1785613820

IET ENERGY ENGINEERING 115
DC Distribution Systems
and Microgrids
Other volumes in this series:
Volume 1 Power Circuit Breaker Theory and Design C.H. Flurscheim (Editor)
Volume 4 Industrial Microwave Heating A.C. Metaxas and R.J. Meredith
Volume 7 Insulators for High Voltages J.S.T. Looms
Volume 8 Variable Frequency AC Motor Drive Systems D. Finney
Volume 10 SF6 Switchgear H.M. Ryan and G.R. Jones
Volume 11 Conduction and Induction Heating E.J. Davies
Volume 13 Statistical Techniques for High Voltage Engineering W. Hauschild and W. Mosch
Volume 14 Uninterruptible Power Supplies J. Platts and J.D. St Aubyn (Editors)
Volume 15 Digital Protection for Power Systems A.T. Johns and S.K. Salman
Volume 16 Electricity Economics and Planning T.W. Berrie
Volume 18 Vacuum Switchgear A. Greenwood
Volume 19 Electrical Safety: A guide to causes and prevention of hazards J. Maxwell Adams
Volume 21 Electricity Distribution Network Design, 2nd Edition E. Lakervi and E.J. Holmes
Volume 22 Artificial Intelligence Techniques in Power Systems K. Warwick, A.O. Ekwue and
R. Aggarwal (Editors)
Volume 24 Power System Commissioning and Maintenance Practice K. Harker
Volume 25 Engineers’ Handbook of Industrial Microwave Heating R.J. Meredith
Volume 26 Small Electric Motors H. Moczala et al.
Volume 27 AC–DC Power System Analysis J. Arrillaga and B.C. Smith
Volume 29 High Voltage Direct Current Transmission, 2nd Edition J. Arrillaga
Volume 30 Flexible AC Transmission Systems (FACTS) Y.-H. Song (Editor)
Volume 31 Embedded Generation N. Jenkins et al.
Volume 32 High Voltage Engineering and Testing, 2nd Edition H.M. Ryan (Editor)
Volume 33 Overvoltage Protection of Low-Voltage Systems, Revised Edition P. Hasse
Volume 36 Voltage Quality in Electrical Power Systems J. Schlabbach et al.
Volume 37 Electrical Steels for Rotating Machines P. Beckley
Volume 38 The Electric Car: Development and future of battery, hybrid and fuel-cell cars
M. Westbrook
Volume 39 Power Systems Electromagnetic Transients Simulation J. Arrillaga and N. Watson
Volume 40 Advances in High Voltage Engineering M. Haddad and D. Warne
Volume 41 Electrical Operation of Electrostatic Precipitators K. Parker
Volume 43 Thermal Power Plant Simulation and Control D. Flynn
Volume 44 Economic Evaluation of Projects in the Electricity Supply Industry H. Khatib
Volume 45 Propulsion Systems for Hybrid Vehicles J. Miller
Volume 46 Distribution Switchgear S. Stewart
Volume 47 Protection of Electricity Distribution Networks, 2nd Edition J. Gers and
E. Holmes
Volume 48 Wood Pole Overhead Lines B. Wareing
Volume 49 Electric Fuses, 3rd Edition A. Wright and G. Newbery
Volume 50 Wind Power Integration: Connection and system operational aspects B. Fox
et al.
Volume 51 Short Circuit Currents J. Schlabbach
Volume 52 Nuclear Power J. Wood
Volume 53 Condition Assessment of High Voltage Insulation in Power System Equipment
R.E. James and Q. Su
Volume 55 Local Energy: Distributed generation of heat and power J. Wood
Volume 56 Condition Monitoring of Rotating Electrical Machines P. Tavner, L. Ran,
J. Penman and H. Sedding
Volume 57 The Control Techniques Drives and Controls Handbook, 2nd Edition B. Drury
Volume 58 Lightning Protection V. Cooray (Editor)
Volume 59 Ultracapacitor Applications J.M. Miller
Volume 62 Lightning Electromagnetics V. Cooray
Volume 63 Energy Storage for Power Systems, 2nd Edition A. Ter-Gazarian
Volume 65 Protection of Electricity Distribution Networks, 3rd Edition J. Gers
Volume 66 High Voltage Engineering Testing, 3rd Edition H. Ryan (Editor)
Volume 67 Multicore Simulation of Power System Transients F.M. Uriate
Volume 68 Distribution System Analysis and Automation J. Gers
Volume 69 The Lightening Flash, 2nd Edition V. Cooray (Editor)
Volume 70 Economic Evaluation of Projects in the Electricity Supply Industry, 3rd Edition
H. Khatib
Volume 72 Control Circuits in Power Electronics: Practical issues in design and
implementation M. Castilla (Editor)
Volume 73 Wide Area Monitoring, Protection and Control Systems: The enabler for
smarter grids A. Vaccaro and A. Zobaa (Editors)
Volume 74 Power Electronic Converters and Systems: Frontiers and applications
A.M. Trzynadlowski (Editor)
Volume 75 Power Distribution Automation B. Das (Editor)
Volume 76 Power System Stability: Modelling, analysis and control B. Om P. Malik
Volume 78 Numerical Analysis of Power System Transients and Dynamics A. Ametani
(Editor)
Volume 79 Vehicle-to-Grid: Linking electric vehicles to the smart grid J. Lu and J. Hossain
(Editors)
Volume 81 Cyber-Physical-Social Systems and Constructs in Electric Power Engineering
S. Suryanarayanan, R. Roche and T.M. Hansen (Editors)
Volume 82 Periodic Control of Power Electronic Converters F. Blaabjerg, K. Zhou, D. Wang
and Y. Yang
Volume 86 Advances in Power System Modelling, Control and Stability Analysis F. Milano
(Editor)
Volume 87 Cogeneration: Technologies, optimisation and implementation
C.A. Frangopoulos (Editor)
Volume 88 Smarter Energy: From smart metering to the smart grid H. Sun, N. Hatziargyriou,
H.V. Poor, L. Carpanini and M.A. Sánchez Fornié (Editors)
Volume 89 Hydrogen Production, Separation and Purification for Energy A. Basile,
F. Dalena, J. Tong and T.N. Veziroğlu (Editors)
Volume 90 Clean Energy Microgrids S. Obara and J. Morel (Editors)
Volume 91 Fuzzy Logic Control in Energy Systems with Design Applications in MATLAB‡/
Simulink‡ İ.H. Altaş
Volume 92 Power Quality in Future Electrical Power Systems A.F. Zobaa and S.H.E.A. Aleem
(Editors)
Volume 93 Cogeneration and District Energy Systems: Modelling, analysis and
optimization M.A. Rosen and S. Koohi-Fayegh
Volume 94 Introduction to the Smart Grid: Concepts, technologies and evolution
S.K. Salman
Volume 95 Communication, Control and Security Challenges for the Smart Grid
S.M. Muyeen and S. Rahman (Editors)
Volume 97 Synchronized Phasor Measurements for Smart Grids M.J.B. Reddy and
D.K. Mohanta (Editors)
Volume 98 Large Scale Grid Integration of Renewable Energy Sources A. Moreno-Munoz
(Editor)
Volume 100 Modeling and Dynamic Behaviour of Hydropower Plants N. Kishor and
J. Fraile-Ardanuy (Editors)
Volume 101 Methane and Hydrogen for Energy Storage R. Carriveau and D.S.-K. Ting
Volume 104 Power Transformer Condition Monitoring and Diagnosis A. Abu-Siada (Editor)
Volume 108 Fault Diagnosis of Induction Motors J. Faiz, V. Ghorbanian and G. Joksimović
Volume 110 High Voltage Power Network Construction K. Harker
Volume 111 Energy Storage at Different Voltage Levels Technology, Integration, and
Market Aspects Zobaa, Ribeiro, Aleem and Afifi (Editors)
Volume 112 Wireless Power Transfer: Theory, technology and application N. Shinohara
Volume 119 Thermal Power Plant Control and Instrumentation: The control of boilers and
HRSGs, 2nd Edition D. Lindsley, J. Grist and D. Parker
Volume 123 Power Systems Electromagnetic Transients Simulation, 2nd Edition N. Watson
and J. Arrillaga
Volume 124 Power Market Transformation B. Murray
Volume 130 Wind and Solar Based Energy Systems for Communities R. Carriveau and
D.S.-K. Ting (Editors)
Volume 131 Metaheuristic Optimization in Power Engineering J. Radosavljević
Volume 905 Power System Protection, 4 volumes
This page intentionally left blank
DC Distribution Systems
and Microgrids
Edited by
Tomislav Dragičević, Pat Wheeler
and Frede Blaabjerg
The Institution of Engineering and Technology

Published by The Institution of Engineering and Technology, London, United Kingdom
The Institution of Engineering and Technology is registered as a Charity in England &
Wales (no. 211014) and Scotland (no. SC038698).
† The Institution of Engineering and Technology 2018
First published 2018
This publication is copyright under the Berne Convention and the Universal Copyright
Convention. All rights reserved. Apart from any fair dealing for the purposes of research
or private study, or criticism or review, as permitted under the Copyright, Designs and
Patents Act 1988, this publication may be reproduced, stored or transmitted, in any
form or by any means, only with the prior permission in writing of the publishers, or in
the case of reprographic reproduction in accordance with the terms of licences issued
by the Copyright Licensing Agency. Enquiries concerning reproduction outside those
terms should be sent to the publisher at the undermentioned address:
The Institution of Engineering and Technology

Michael Faraday House
Six Hills Way, Stevenage
Herts, SG1 2AY, United Kingdom
www.theiet.org
While the authors and publisher believe that the information and guidance given in this
work are correct, all parties must rely upon their own skill and judgement when making
use of them. Neither the authors nor publisher assumes any liability to anyone for any
loss or damage caused by any error or omission in the work, whether such an error or
omission is the result of negligence or any other cause. Any and all such liability is
disclaimed.
The moral rights of the authors to be identified as authors of this work have been
asserted by them in accordance with the Copyright, Designs and Patents Act 1988.
British Library Cataloguing in Publication Data

A catalogue record for this product is available from the British Library
ISBN 978-1-78561-382-1 (hardback)

ISBN 978-1-78561-383-8 (PDF)
Typeset in India by MPS Limited

Printed in the UK by CPI Group (UK) Ltd, Croydon
Contents
Preface xvii
1 DC microgrid control principles – hierarchical control diagram 1

Linglin Chen, Tao Yang, Fei Gao, Serhiy Bozhko, and Patrick Wheeler
1.1 Introduction 1
1.2 The hierarchical control for DC MGs 2
1.3 Primary control 3
1.3.1 Basics of droop control 3
1.3.2 Power sharing errors 6
1.3.3 Droop strategies 8
1.3.4 Dynamic power sharing 10
1.3.5 Interfaces to upper levels 12
1.4 Secondary control 13
1.4.1 Centralized approach 13
1.4.2 Distributed approach 15
1.5 Tertiary control 18
1.6 Summary 20
References 21
2 Distributed and decentralized control of dc microgrids 23

Saeed Peyghami, Hossein Mokhtari, and Frede Blaabjerg
2.1 Introduction 23
2.2 Decentralized approaches 24
2.2.1 Mode-adaptive (autonomous) droop control 24
2.2.2 Nonlinear droop control 28
2.2.3 Frequency droop control 31
2.3 Distributed approaches 34
2.3.1 Fully communicated control 34
2.3.2 Sparse communicated (consensus-based) control 35
2.3.3 Sparse communicated control using current information 36
2.4 Conclusion and future study 40
References 41
viii DC distribution systems and microgrids
3 Stability analysis and stabilization of DC microgrids 43

Alexis Kwasinski
3.1 Dynamic characteristics of DC microgrids 43
3.2 DC microgrids stability analysis 45
3.3 Passive approaches for stabilization of DC microgrids 48
3.4 Control strategies for stable DC microgrids operation 50
3.5 Operation of rectifiers with instantaneous constant power loads 58
3.6 Summary 60
References 60
4 Coordinated protection of DC microgrids 63

Jae-Do Park, Md Habib Ullah, and Bhanu Babaiahgari
4.1 Introduction 63
4.2 Faults in DC power systems 65
4.2.1 Fault types and behavior 65
4.2.2 Fault current analysis 66
4.2.3 Faults in various bus configurations 68
4.3 Coordinated protection techniques 71
4.3.1 AC side protection 72
4.3.2 DC side protection 73
4.3.3 Applications 80
4.4 Summary 85
4.5 Acknowledgment 85
References 85
5 Energy management systems for dc microgrids 91

Amjad Anvari-Moghaddam, Tomislav Dragičević, and Marko Delimar
5.1 Introduction 91
5.2 DC microgrid operation and control fundamentals 92
5.2.1 Power/energy management schemes 92
5.2.2 Control schemes 97
5.3 Interfacing converter control strategies for power/energy
management purposes 105
5.3.1 Voltage control/grid-forming mode 105
5.3.2 Current control/grid-following mode 107
5.4 Illustrative example 108
5.5 Conclusions 110
References 111
6 Control of solid-state transformer-enabled DC microgrids 119

Xu She, Alex Huang, Xunwei Yu, and Yizhe Xu
6.1 Introduction 119
6.2 Solid-state transformer-based microgrid: architecture and benefits 119
Contents ix
6.3 Centralized power management of solid-state transformer-based

DC microgrid 121
6.3.1 Power management strategy 121
6.3.2 Case study 124
6.3.3 Summary 128
6.4 Hierarchical power management of solid-state transformer-enabled
DC microgrid 131
6.4.1 Power management strategy 131
6.4.2 Case study of a small-scale DC microgrid 136
6.4.3 Summary 142
6.5 Control of SST-enabled DC microgrid as a solid-state synchronous
machine (SSSM) 142
6.5.1 Concept of the SSSM 142
6.5.2 Frequency regulation 144
6.5.3 Power up/down reserve support 145
6.5.4 Voltage regulation 147
6.5.5 Case study 147
6.5.6 Summary 150
6.6 Conclusion 151
References 151
7 The load as a controllable energy asset in dc microgrids 153

Robert S. Balog, Morcos Metry, and Mohammad Shadmand
7.1.1 Local area power and energy system 154
7.2 Why control the load? 156
7.2.1 Benefit of load control 156
7.2.2 Is load modulation practical? 159
7.3 Time-scale of energy requirements 161
7.3.1 Short-term transients 161
7.3.2 Long-term transients 163
7.4 Autonomous load control 164
7.4.1 Control 164
7.4.2 Architecture 165
7.4.3 Strategy for controlling the load to be an energy asset 166
7.5 Droop control for stability and information communication 168
7.5.1 Constant power load and its deleterious effect on
dc systems 169
7.5.2 Steady state stabilization 170
7.5.3 Dynamic stabilization 172
7.6 Voltage-based load interruption 175
7.6.1 Power flow analysis 176
7.6.2 Contingency analysis 178
7.6.3 Search algorithm 178
x DC distribution systems and microgrids
7.7 dv/dt-Based dynamic load interruptions 180

7.8 Load prioritization and scheduling 181
7.9 Summary 182
Acknowledgments 183
References 184
8 Electric vehicle charging infrastructure and dc microgrids 189

Srdjan Srdic and Srdjan Lukic
8.1 Overview of EV and EVSE markets and trends 189
8.2 dc Fast charging systems and requirements 190
8.2.1 dc Fast charging systems and standards 190
8.2.2 State-of-the-art EV dc fast chargers 192
8.2.3 dc Fast charger power converter topologies 194
8.3 Microgrid topologies for EV charging 199
8.3.1 State-of-the-art dc fast charger installation 200
8.3.2 Medium-voltage dc fast chargers: a new approach
to high-power EV fast charging 203
8.4 Conclusions and future trends 208
References 209
9 Overview and design of solid-state transformers 215

Levy Costa, Marco Liserre, and Giampaolo Buticchi
9.1 Solid-state transformer: concept 215
9.2 SST in electric distribution grid application 216
9.3 ST architecture classification 218
9.3.1 Power conversion stages 218
9.3.2 Modularity 219
9.3.3 Modularity level 220
9.3.4 dc Link voltage availability 221
9.4 Solid-state transformer and smart transformer architectures
overview 223
9.5 dc–dc Stage: power converter topologies 224
9.5.1 Requirements 224
9.5.2 Basic module topologies 225
9.6 Series resonant converter 226
9.6.1 Current stress 228
9.6.2 Efficiency expectation 229
9.7 Dual active bridge (DAB) converter 229
9.7.1 Modulation strategy 230
9.7.2 Analysis of the DAB using the PSM 231
9.7.3 Current stresses on the DAB 233
9.7.4 Efficiency calculation 235
9.8 Active-bridge converter: control description and tuning 236
Contents xi
9.9 Summary 238

References 239
10 Bipolar-type DC microgrids for high-quality power distribution 245

Sebastian Rivera, Ricardo Lizana, Samir Kouro, and Bin Wu
10.2 Bipolar-type DC distribution systems 247
10.3 Topologies and operational aspects of bipolar LVDC grids 249
10.3.1 Distribution converter topologies 249
10.4 Balancing topologies 253
10.4.1 Bidirectional buck-boost topologies 254
10.4.2 Coupled inductor current redistributor 255
10.4.3 Three-level DC–DC current redistributors 256
10.5 Control schemes 257
10.5.1 Cascade control 257
10.5.2 AC–DC converter control 258
10.6 Summary 262
Acknowledgement 263
References 263
11 Aircraft DC microgrids 267

Fei Gao, Tao Yang, Serhiy Bozhko, and Pat Wheeler
11.2 Aircraft electrical power system 267
11.2.1 Power generation 268
11.2.2 Power distribution 269
11.2.3 Power utilization 273
11.2.4 Energy storage system 274
11.3 Power quality requirements in aircrafts 274
11.4 Aircraft starter/generator control 276
11.5 Control strategies in aircraft DC microgrids 279
11.5.1 Primary control 280
11.5.2 Secondary control 283
11.6 Stability analysis 285
11.7 Chapter summary 288
Acknowledgement 289
References 289
12 Shipboard MVDC microgrids 295

Dong-Choon Lee, Yoon-Cheul Jeung, Dinh Du To, and Duc Dung Le
12.2 Architecture of shipboard MVDC power systems 296
12.2.1 Structure of DC bus 296
xii DC distribution systems and microgrids
12.2.2 Prime movers 297

12.2.3 Components of shipboard MVDC power systems 297
12.3 Control of shipboard MVDC power systems 301
12.3.1 Gen-sets 301
12.3.2 Energy storage systems 302
12.3.3 Power management control 303
12.4 Stability analysis 310
12.5 Faults and protection 312
12.5.1 Architecture of protection schemes 312
12.5.2 Fault detection and location 313
12.6 Summary 317
References 317
13 DC-based EVs and hybrid EVs 321

Ruoyu Hou, Jing Guo, and Ali Emadi
13.1.1 Power electronic system in electrified vehicles 321
13.1.2 DC auxiliary loads in electrified vehicles 323
13.2 Converter topologies in electrified vehicles 325
13.2.1 Conductive HV battery chargers 325
13.2.2 Active power filters in HV battery chargers 328
13.2.3 LV auxiliary power modules 330
13.3 Practical design considerations 331
13.3.1 Selections of switching devices 331
13.3.2 Selections of DC-link capacitors 332
13.3.3 Design considerations of DC bus bars 333
13.4 Topological reconfigurations of DC systems in electrified vehicles 337
13.4.1 Topological reconfigurations in HV charging and
propulsion systems 338
13.4.2 Topological reconfigurations in dual-voltage charging
systems 339
13.5 Conclusions 340
References 340
14 DC data centers 343

Enver Candan and Robert C.N. Pilawa-Podgurski
14.2 Development of DC power distribution in data centers 345
14.3 Efficiency 346
14.4 Reliability 349
14.4.1 Fault tolerance 350
14.4.2 Back-up power 351
14.5 Integration with other DC sources and loads 352
14.5.1 Renewable and distributed energy sources 353
Contents xiii
14.5.2 Cooling 353

14.5.3 Lighting 354
14.6 Installation 354
14.6.1 Isolation 354
14.6.2 Grounding 355
14.6.3 Wiring 355
14.6.4 Connectors 356
14.6.5 Total cost of ownership 356
14.7 Protection 357
14.8 Power quality 358
14.9 Stability 358
14.10 Existing high voltage DC data centers 358
14.11 Key obstacles to widespread adoption of DC data centers 360
14.11.1 Overly optimistic claims 360
14.11.2 Emergence of rack-level UPS 360
14.11.3 Protection and DC circuit breakers 360
14.11.4 Incumbent cost and familiarity advantages 361
14.12 Conclusion 361
References 361
15 DC microgrid in residential buildings 367

Rajeev Kumar Chauhan, Francisco Gonzalez-Longatt,
Bharat Singh Rajpurohit, and Sri Nivas Singh
15.2 Conceptualisation: DC microgrids in buildings 368
15.3 Classification of microgrids 368
15.3.1 AC microgrid system 368
15.3.2 Hybrid AC–DC microgrid system 370
15.3.3 DC microgrid system 371
15.4 Topologies for DC microgrid for residential applications 372
15.4.1 Unipolar LVDC system 372
15.4.2 Bipolar LVDC system 372
15.5 Mathematical analysis of AC vs DC microgrid system 373
15.5.1 Total daily load 373
15.5.2 Voltage, current and power losses in DC supply 374
15.6 Comparison between AC and DC residential buildings 377
15.7 AC residential buildings 378
15.8 DC residential buildings 379
15.9 Automation architecture for smart DC residential buildings 379
15.9.1 Field level 379
15.9.2 Field network 380
15.9.3 Automation level 380
15.9.4 Primary and secondary network 380
15.9.5 Management level 381
xiv DC distribution systems and microgrids
15.10 Advantages, challenges and barriers of smart DC residential

buildings 381
15.11 Comparison AC and DC residential buildings: an illustrative
example 382
15.12 Conclusions 386
References 387
16 DC microgrids for photovoltaic powered systems 389

Rasool Heydari, Saeed Peyghami, Yongheng Yang,
Tomislav Dragičević, and Frede Blaabjerg
16.2 Architecture of dc microgrids 390
16.2.1 MVDC and LVDC microgrid for dc distribution systems 390
16.2.2 LVDC microgrid for space applications 392
16.2.3 MVDC microgrid for marine applications 393
16.2.4 LVDC microgrid for data centers 394
16.2.5 LVDC microgrid for homes 395
16.2.6 MVDC microgrid for oil-drilling applications 396
16.3 dc Microgrids for photovoltaic power plants 397
16.4 PV system modeling 398
16.4.1 Ideal model of a PV cell 398
16.4.2 Single diode with series resistance model 399
16.4.3 Single-diode with shunt resistance model 400
16.4.4 Double-diode model 400
16.5 Power converter technologies 401
16.5.1 Transformerless topologies 401
16.5.2 dc Converters with high-frequency transformers 403
16.5.3 Next generation PV converters 404
16.6 Control of dc microgrids for PV collection plants 406
16.7 Simulation of a droop-controlled PV microgrid 407
16.8 Summary 410
References 411
17 Demonstration sites of dc microgrids 415

Aditya Shekhar, Laura Ramı́rez-Elizondo, Seyedmahdi Izadkhast, and
Pavol Bauer
17.2 Off-grid dc microgrids 415
17.2.2 Components 417
17.2.3 Control and operational safety 417
17.2.4 Socioeconomic impact 417
17.2.5 Discussion 417
Contents xv
17.3 Transportation electrification 418

17.3.1 Shipboard dc microgrids 418
17.3.2 Railway traction with dc supply 421
17.4 Datacenters 423
17.4.2 Components 424
17.4.3 Protection and grounding system 424
17.5 Residential and commercial dc buildings 425
17.5.1 dc House 425
17.5.2 Building interiors 427
17.5.3 dc Distribution grids 429
References 432
Index 437
Preface
The current revolution in power electronics has enabled direct-current (DC)

distribution architectures. These architectures have a potential for higher efficiency
and reliability, better current carrying capacity, and faster response when compared
to conventional alternating-current (AC) architectures. DC distribution also has
advantages in terms of inexistence of reactive power flows, power quality issues
and frequency regulation, resulting in a simpler control architecture and more
robust stability when compared to the AC coupled systems. For these reasons, DC
is more and more used in the modern society at different voltage and power levels.
The most recent emergence of DC is happening at low voltage levels, where
practically all modern appliances (e.g., laptops, cellphones, LED lights, displays,
etc.), air, land and sea vehicles, residential and commercial facilities, as well as
renewable energy plants increasingly rely on the DC power distribution technology.
All these emerging applications can be categorized under the same label, i.e., the
DC ‘‘Microgrid’’. However, as the utility power system will very likely continue to
be operated using the AC architecture, development and accommodation of DC
microgrids in the overall power system holds many integration challenges. On the
other hand, the adoption of DC microgrids in standalone and transport applications
has already begun, but lack of existing standards and protection methodologies
creates a specific set of challenges in this area as well. The aim of this book is to
overview the technologies that have been developed over the last 10–15 years in the
area of DC microgrids.
The book is divided into three parts – control circuits, power architectures, and
real-world applications. The first part comprises seven chapters, which provides a
comprehensive overview of different control approaches focused both on local
control functions of individual converters as discussed in Chapters 1, 3, and 7 (e.g.,
voltage/current control, DC-bus stabilization, point of load converters support), and
on functions that allow their effective coordinated performance as presented in
Chapters 2, 4, and 5 (e.g., distributed power sharing/voltage regulation and energy
management systems) and safe operation (coordinated protection systems). Except
for Chapter 6 that focuses specifically on the control of solid-state transformers,
this portion of the book is rather generic and is thereby applicable to any type of
the power converter topology and DC microgrid architecture. This part will thus
provide essential reading for anyone involved in research in design and imple-
mentation of control algorithms.
Building upon the set of control oriented chapters, the three chapters in the
second part overview several prominent DC microgrid architectures. Architectures
xviii DC distribution systems and microgrids
that are different and more complex compared to the most conventional single bus
structure were intentionally selected for a detailed overview here, as they can offer
additional degrees of flexibility and redundancy. Moreover, due to the rapid
development of power electronics technology that has resulted in increased relia-
bility of power electronic devices, additional hardware required to establish such
architectures is not considered as a drawback any more. In particular, two archi-
tectures overviewed are relatively generic (i.e., the solid-state transformers in
Chapter 9, and bipolar-type DC microgrids in Chapter 10), while the architecture
detailed in Chapter 8 for electrical vehicle charging infrastructure is more tailored
toward that specific application.
The final part, comprised of seven chapters, presents an analysis of a number of
applications in which the control methods and architectures discussed in the first two
parts can be applied in the specific context of a given application. The selected
applications for this book include aircraft and shipboard systems, electrical vehicles,
data centers, residential buildings, and photovoltaic powered systems, respectively.
The book ends with an overview of some of the DC microgrid demonstration sites
around the world.
The Editors would like to thank all the contributors for their excellent work
and cooperation in the preparation of this book. Furthermore, we would like to
acknowledge the excellent and timely assistance of the editorial and production
staff at the Institution of Engineering and Technology (IET).
The Editors
Tomislav Dragičević, Aalborg University, Denmark
Patrick Wheeler, University of Nottingham, UK and China
Frede Blaabjerg, Aalborg University, Denmark
Chapter 1
DC microgrid control principles – hierarchical
control diagram
Linglin Chen1, Tao Yang1, Fei Gao1, Serhiy Bozhko1, and
Patrick Wheeler1
1.1 Introduction
Different electrical components can be conveniently connected to a DC grid as
there are no issues of reactive power, power factor, different frequencies, etc. as
in AC grids. Control is one of the key disciplines in achieving the objective of
smart grid initiatives. The control of DC microgrids (MGs) covers a wide range of
control objectives and responsive dynamics. Therefore, to better understand the
concept and implement the MG control, division of the MG control into relatively
independent sublevels is essential. The utility AC power system has a history of
over a hundred year of development. Due to its unique physical characteristic, MGs
have more feasibility of controlling individual distributed generation (DG) sub-
systems due to their power electronics interfaces. In order to fully exploit the
potential of power electronic converters in the MG, a hierarchical control for AC
MGs was developed in literature [1].
In the hierarchical structure, MG controls are classified into three groups
according to their different functionalities. Generally, in the primary control level,
controllers are responsible for the active and reactive power sharing among
DGs. In the secondary control level, frequency variation and AC voltage ampli-
tude offset are compensated. In the tertiary control level, power management
algorithm is conducted. The recent development of solid-state circuit breakers for
the DC network has made the DC MG an alternative option to the AC MGs.
Compared with AC networks, the DC network has no reactive power issues and
less power cables, which can potentially reduce weight and cost of the power
grid. A similar hierarchical control structure can also be implemented in DC MGs
as shown in Figure 1.1. The details of each control level will be explained in the
following section.
1
Department of Electrical and Electronics Engineering, The University of Nottingham, United Kingdom
2 DC distribution systems and microgrids
MG central control
Tertiary
Bandwidth
Secondary increases
Primary
Inner loops
Local Local Local

control control control
converter converter converter
Microgrid
Figure 1.1 Hierarchical control of an MG
1.2 The hierarchical control for DC MGs
In the past few years, different concepts for primary, secondary and tertiary control
have been developed for DC MGs applications [1–4]. The explanation of each level
can be given as follows:
● The primary control is implemented strictly local within the DG subsystems. In
a DC MG, installation of more than one power source is common. Therefore,
how to manage the DGs to share the load proportionally to their rated power is
an issue to consider. Conventionally, droop control [5] is a preferred option in
this control level as it provides autonomous control features to the DGs in the
presence of communication failure. It effectively deals with the load sharing
among DGs connected on the same DC bus and provides references to sub-
control loops (voltage or current loops). Alternatively, the droop control can
also be perceived as virtual resistance, connected in series with converter on
the DC bus side.
● The secondary control, predominately, sets in an MG central controller which
is usually, together with tertiary control level, remotely located to DGs, thus
communications are required. It generates voltage shifting signals to primary
control level in aim to eliminate the DC voltage deviation caused by primary
droop controls. Moreover, it is also possible to implement the secondary con-
trol level locally inside DGs.
DC microgrid control principles – hierarchical control diagram 3
● The tertiary control takes responsibility of power management between MG

and upper grids and is also commonly referred to as the energy management
system. In addition, tertiary control level has to prepare dispatch schedules
inside MG to achieve the optimum operation [1].
The hierarchical control essentially groups the MG control co-ordinately from the
bottom level power sharing among electronics converters to top level power man-
agement of DC MGs. The response dynamic ranges from milliseconds to a few
days. The primary control guarantees reliable operation of DC MGs in the presence
of communication failure, and the tertiary level provides flexibilities and intelli-
gence to power governance through the intermedium secondary control.
1.3 Primary control
The functionality of this control level is to deal with the power sharing among
different DG units. Generally, the load sharing can be achieved in both centralized
and decentralized ways. In the centralized way [6], usually the DG with the largest
power capacity is chosen to be the dominant power source, and it is responsible for
the regulation of the DC bus voltage (voltage support mode). Other DGs in the DC
MG work in power mode (current mode). They receive power (current) reference
signals from a central controller. In this case, a reliable operation of power sharing
largely depends on the effective communication. As the control bandwidth of the
primary control is the highest among the three control levels (in terms of kHz),
communication within this level between DGs also increases infrastructure
investment. Thus, for large-scale MGs with remotely located DGs, communication
links are not practical in the primary control level, and only local current and
voltage information is available. Thus, as a fully decentralized approach, the droop
control has shown favourable features in large-scale MGs for load sharing. With the
droop control, all measurements take place locally within DGs. The droop control
in DC MGs can be intuitively appreciated as ‘virtual resistance’ [2]. This allows
paralleled DGs to share DC bus currents proportionally in line with their respective
power rating. Due to its modularity and reliability, the droop control has been
widely used in DC MGs in the primary control level [7–9].
1.3.1 Basics of droop control

The original idea of droop control is from AC generators in legacy grid, where the
power generation is controlled by mechanical outputs of prime movers. A speed
governor measures the variation of a synchronous generator (SG) speed (frequency)
and provides the power regulation according to generator’s droop characteristic
[10]. With such characteristic, SGs on a common AC bus can adjust to load
changes according to the variation of their terminal frequencies and voltages. This
allows power sharing of different SGs without extra communication links.
A similar idea was initially used in paralleled inverters for load sharing [11].
If neglecting the line resistance, the active power (P) delivered from each DG unit
is only dependent on frequency ( f), and the reactive power (Q) is solely related to
voltage (V). Therefore, by adjusting the frequency f and the terminal V, the active
power and reactive power from each DG unit can be regulated. These results in the
famous droop regulation in an AC grid as
f f0 ¼ kp ðP P0 Þ (1.1)
V V0 ¼ kq ðQ Q0 Þ (1.2)
with measured terminal voltage V and frequency f, the required P and Q can be
derived from (1.1) and (1.2). The derived P and Q are then sent to the inverter inner
controllers (control level 0), and the DG unit can operate in desired conditions.
A similar concept can also be used to parallel DC converters for power sharing.
Within DC MGs, there is no reactive power, and the active power is only controlled
by terminal DC voltage. Hence, a linear droop characteristic line between V and I
can be implemented. As shown in Figure 1.2, the DC-link voltage vDC and the DC
current iDC is in a linear relation. With the DC current increasing, the DC-link
voltage will be reduced. The relation between DC-link voltage vDC and the DC
current iDC can be expressed as
vDC ¼ V0 RD iDC (1.3)
where iDC is the power converter output DC current; V0 is the voltage at no load
(iDC ¼ 0 A); vDC is the DG terminal voltage; the slope RD is the droop gain and can
also be viewed as a virtual resistor, which will be explained later.
To explain the power sharing using DC V–I droop control concept, a simple
DC MG system is shown in Figure 1.3. The droop curves for the two DGs are
illustrated as in Figure 1.4. The control structure details are depicted in Figure 1.5.
DC currents and voltages of DG1 and DG2 are measured locally and fed into their
own primary controllers. Based on the measured current iDC1 and iDC2, terminal DC
voltage references can be calculated accordingly with the V–I droop curves.
Eventually at steady state, the DC voltage equals to VL. Because the droop gains
RD1 and RD2 are designed based on the rated power of the DG, this will be dis-
cussed later. As shown in Figure 1.4, DG1 and DG2 spend different ‘efforts’ to
share the load current, with iDC1 and iDC2, respectively.
V0
RD
vDC
iDC i
Figure 1.2 V–I droop characteristic

DG 2 DG 1
idc2 idc1
+
VL
RL
Figure 1.3 A simple MG with two DGs and a resistive load
vdc
V0 RD1
RD2
vL
idc
idc2 idc1
Figure 1.4 Droop curves for DG1 and DG2
vdc
vdc Voltage
*
Current PWM
V0 DC source
– loop loop generator
idc
RD
Primary control
Figure 1.5 Practical primary control implementation with V–I droop method [2]
Usually, the inner voltage and current loops have fast dynamic response. The
output impedance of the converter is predominately dependant on the droop gain.
Therefore, the V–I droop gain is termed also as virtual resistance RD [5]. Figure 1.6
illustrates a simplified DC system diagram, where Vo1 and Vo2 are DC voltages at
no load from V–I droop controllers for DG1 and DG2, respectively; RD1 and RD2
are virtual resistance for DG1 and DG2, respectively. If there is no virtual resis-
tance, a small difference between Vo1 and Vo2 will result in a large circulating
DG1 DG2
idc1 RD1 RD2 idc2
+ + +
Vo1 VL Vo2
– RL –
–
Figure 1.6 The equivalent circuit of two DC power supplies
vdc
εv vo
imax idc
imin
Figure 1.7 An illustration of V–I droop with maximum allowed deviation
current between DG1 and DG2. This is because without virtual resistance RD1 and
RD2, the circulating current between DG1 and DG2 is only subject to cable resis-
tance which is essentially very small (neglected in Figure 1.6).
In general, two aspects should be considered when designing the value of
the virtual impedance. One is the maximum allowed voltage deviation (ev in
Figure 1.7). A larger virtual impedance will make the V–I output characteristic
‘softer’ (as the slope is larger). Another aspect to consider is the system stability, as
the droop gain alters the output impedance. A detailed analysis concerning stability
analysis in DC MGs can be found in [5]. Thus, in this chapter, only constraint of the
maximum allowed DC bus voltage deviation is considered. Assuming that ev is the
maximum allowed voltage deviation, the allowed maximum RD is given as
ev
RD ¼ (1.4)
2imax
where imax is the maximum output current. The relation of ev and imax is illustrated
in Figure 1.7.
1.3.2 Power sharing errors

The droop V–I droop control provides autonomous local control to DGs at the price
of ‘soft’ V–I characteristic (here the term ‘soft’ is used compared with an ideal stiff
voltage source). Therefore, voltage deviation will be expected when different loads
are applied to the DC system [7]. Another issue for using the conventional V–I
droop is that some power sharing error would occur during operations. Two main
reasons for this error are discussed below:
1. Error in DC voltage control. In practical situations, due to the measurement
error from voltage and current sensors, the DC bus voltage at each DG unit
might be different even if the references are set the same. If we consider no
load conditions, this control error gives different V0 for each DG unit (Vo1 and
Vo2), as shown in Figure 1.8. The voltage error within DGs will potentially
cause large power sharing discrepancies in MG. As can be seen from
Figure 1.8, if two DGs are with the same power rating and virtual impedances
(i.e. RD), the current error between two DGs is smaller (I1I2) with a larger
droop gain. On the other hand, although a smaller droop gain produces
larger current sharing errors (I1oI2o), it means a stiffer voltage source and a
smaller voltage deviation due to load changes. This results in a dilemma in
designing the droop gain for primary control.
2. Cable resistance: In large-scale DC MGs, DG terminal voltages might be a
variant with load power in presence of cable resistance. A simple DC MG
which takes accounts of cable resistance is shown in Figure 1.9. The cable
resistance between the converter DC terminal and the common DC bus is
considered as Rline1 and Rline2.
Vdc Large Small

V01 droop droop
V02
VL
idc
I1 I2 I1o I2o
Figure 1.8 Unequal load sharing due to load distribution in DC MG
DG1 DG2
RD1 Rline1 Rline2 RD2 idc2
idc1
+ +
+ + +
vdc1 vL vdc2 vo2
vo1 RL
– – –
– –
Figure 1.9 A simplified DC MG with cable resistance

The DG1 node voltage vDC1 can be derived as in the following equation:
vDC1 ¼ V0 RD1 iDC1 (1.5)
where vDC1 is the voltage reference produced by V–I droop control in MG1; iDC1 is
the output current from DG1; RD1 is the virtual output resistance (droop gain) for
DG1; and V0 is the output voltage at no load. If the line impedance of Rline1 and
Rline2 is considered, (1.5) can be written as
vL ¼ V01 RD1 iDC1 Rline1 iDC1 (1.6)
vL ¼ V02 RD2 iDC2 Rline2 iDC2 (1.7)
Combining (1.6) and (1.7), and assuming Vo1 equals to Vo2, yields
1 1
iDC1 : iDC2 ¼ : (1.8)
RD1 þ Rline1 RD2 þ Rline2
From (1.8), we can conclude that the current sharing between DGs is not only
dependent on the droop gain but also the line impedance. In large-scale DC systems
where the line impedance is not negligible, the current sharing error cannot be
avoided with conventional droop control.
1.3.3 Droop strategies

Different droop control techniques for power electronics converters in DC MGs
have been developed [5,12]. These MGs implement the basic concept of droop
control in different ways and can be categorized into two types, i.e. voltage-mode
droop and current-mode droop. In the voltage-mode droop control, the output cur-
rent is measured and fed to the V–I droop. The voltage reference is calculated and
fed into inner voltage controller, as shown in Figure 1.10(a). In the voltage-mode
droop control, the V–I droop can also be replaced by the V–P droop as the DC
current is essential reflecting the power, as shown in Figure 1.10(b). In the current-
mode droop control, the output voltage is measured and fed to the I–V droop. The
current reference is calculated and fed into inner current controller, as shown in
Figure 1.10(c). Similarly, the I–V droop can be replaced by the P–V droop as shown
in Figure 1.10(d) [13].
Depending on the DC control strategy, the converter can be operated either in
the current mode or voltage mode [14]. The current-mode droop control scheme is
shown in Figure 1.11(b), with the current reference derived from the I–V droop
characteristic in Figure 1.11(a), based on the DC voltage measurement.
Although the voltage-mode and current-mode droop controls seem to be dif-
ferent, the DC current sharing in steady states for both methods is the same in
presence of cable resistance [13]. The difference for them is the implication on the
system stability [5]. When using current-mode droop control, the current output of
the converter is dependent on the droop gain. An increased DC current control loop
bandwidth is helpful for stability enhancement of the system. However, the upper
boundary of DC current control bandwidth is limited by the right half plane (RHP)
V–P droop
V–I droop idc
idc characteristic
characteristic +
vdc
vdc + P *
vdc
* vdc
idc vdc P
idc vdc
–
–
{
(a) (b)
I–V droop P–V droop

characteristic characteristic
idc + +
idc
* P*
Vdc Vdc P*
vdc vdc
vdc
vdc – –
{ {
(c) (d)
Figure 1.10 Droop characteristic employed in VSCs: (a) V–I droop, (b) V–P
droop, (c) I–V droop and (d) P–V droop
I–V droop idc

characteristic
Idc controller
idc idc ref
Vdc Idc Iqref VSC vector
vdc PI vdc
control
idc
vdc
{
(a) (b)
Figure 1.11 I–V droop and its corresponding current-mode control scheme:
(a) I–V droop characteristic and (b) current-mode control scheme
zero in the transfer function. For the voltage-mode droop-controlled system, the DC
voltage dynamics are affected by the DC voltage control bandwidth and the droop
gain. Increasing the droop gain will reduce the voltage loop bandwidth. In contrast
to the current-mode approach, the voltage-mode droop control regulates the term-
inal voltage based on current measurements. The RHP zero causes high gain
instability within the vDC controller, i.e. the system will easily go unstable when the
vDC controller has a high gain value. Furthermore, when utilizing voltage-mode
droop control, the voltage loop bandwidth is mainly determined by the voltage
controller rather than the droop gain. Greater details on stability analysis can be
found in literature [5].
The discussed, droop control techniques so far have assumed linear relations
between the DC voltage and DC current. As opposed to linear droop regulations
mentioned above, a nonlinear droop curve is proposed in [12]. The droop gain can
V
P*
Figure 1.12 Nonlinear droop control scheme considering available headroom for
each DGs
be adapted according to the available headroom available for each individual DGs.
The droop gain is defined as

P jP j l
K ¼ K0 (1.9)
P
where K0 is the rated droop gain, K is the adapted coefficient value, P* is the rated
power and P is the real-time output power. Essentially, the droop gain K gets larger
in heavy load conditions, as shown in Figure 1.12. Therefore, the power sharing
error is reduced in presence of DC voltage control discrepancy.
1.3.4 Dynamic power sharing

Under sudden change of load, a desired power sharing may not be guaranteed
solely with conventional droop controls. This is mainly due to the fact that an MG
may incorporate different types of DG units with different dynamic bandwidth and
performance. For example, fuel cells usually have much slower dynamics than PV
panels due to the limitation of chemical reactions. Although the rated power of fuel
cell could be larger than other renewables and storage sources, in presence of a
sudden load change, extra power required by a load with faster dynamics could
only be provided by faster DGs (PVs, batteries, super-capacitors, etc.). In this
section, power sharing in the transient condition is addressed. Therefore, to ensure a
reasonable power sharing in the transition process, dynamic power sharing is
inevitable and essential for MGs.
The basic idea of the dynamic power sharing is to compensate the required
power from different DGs according to the load and DG dynamic spectrums.
Dynamic power sharing can be achieved by configuring a dynamic compensator
in series with the droop control loop. The approach is referred as frequency-
coordinating virtual impedance [15]. It separates the dynamic spectrum into dif-
ferent frequency ranges and implements the compensation by inserting a filter in
series with the I–V droop admittance as shown in Figure 1.13. The error between
the reference voltage Vo at no load and measured output voltage vDC is sent to the
P–V droop block. In series with the droop block is a high-pass filter. The output of
the filter is the power reference P*, and the power reference P* is then converted
Frequency coordinating virtual impedance
idc
Z(s)
Vo P* i* Current PWM DC
+ 1/Rd ÷
– loop generator source
ω
vdc
Figure 1.13 Frequency coordinating virtual impedance with I–V droop [15]
Hybrid ESS
Grid
Lead-acid
AC
DC
DC
DC
Load SC
AC DC
DC DC
FC
Gas
DC H2
AC DC O2
DC
PV
DC
DC
Figure 1.14 Composite of a DC MG with hybrid energy storages
into a current reference i*. Consequently, the high-pass filter alters the virtual
impedance, and the equivalent output impedance becomes
ZðsÞ
Zdroop ðsÞ ¼ (1.10)
Rd
As an example, an MG with both lead acid and super-capacitor energy storages is
shown in Figure 1.14. The lead-acid battery has slower dynamics than the super-
capacitor. Thus, the slower load dynamic spectrum is loaded to battery converter,
while the faster dynamic load is assigned to super-capacitor converter as depicted
in Figure 1.15. A low-pass filter is inserted in series with the droop admittance in
battery control loop to shape the equivalent virtual impedance as illustrated in
Figure 1.16. This allows the output impedance of the battery converter increases
with frequency. On the contrary, a high-pass filter is implemented in the control
loop of super capacitor, and the output impedance of the super capacitor converter
shows low impedance in the medium frequency range. Dynamic range greater than
control bandwidth is mainly managed by passive DC bus filtering capacitors.
idc
Zbat(s)=ωc/(s+ωc)
+ 1/Rd ÷
ω
vdc
(a)
idc
Zsc(s)=s/(s+ωc)
+ 1/Rd ÷
ω
vdc
(b)
Figure 1.15 Frequency coordination for hybrid ESS: (a) control loop for battery
and (b) control loop for super-capacitor [15]
Z (dB)
80
Zbat
Zsup
60 Zsup // Zbat
40
20
0
0.1 1 100 1k f (Hz)
Battery Super-cap DCcaps
Figure 1.16 Frequency-domain impedance comparison of converters [15]
1.3.5 Interfaces to upper levels

Primary control level needs interfaces to interact with upper levels. According to
above discussions on the primary control, interfaces are no load voltage reference
value (V0), droop gain (RD) and saturation limits (P and Pþ). These variables can
be manipulated by secondary and tertiary control levels to ensure power quality and
accommodate flexible power management. They are illustrated in Figure 1.17.
Implications of changing interface variables on droop curve are shown in
Figure 1.18. Typical saturation limits P and Pþ are normally controlled according
Z(s) idc +
Vo – P* i*
+ 1/Rd ÷ T(s) Y vdc
ω –
P– P+
Figure 1.17 Interfaces of primary level to upper levels
P+
P–
Vo Rd
Figure 1.18 Implications of changing interface variables to the droop curve
to the requirement of energy management [16]. The voltage reference value V0 is

adjusted to restore the DC voltage offset caused by the primary droop control. The
droop gain RD is modified to reschedule the power sharing between DGs.
1.4 Secondary control
The secondary control aims to restore the DC bus voltage deviation caused by
primary droop control. When MG is connected to an upper grid, the secondary
controller receives command from upper tertiary controllers. This allows voltages
at common coupling points to track the upper grid voltage. The secondary control
can be implemented locally inside DGs. It can also be integrated within the tertiary
control in an MG central controller, as shown in Figure 1.1. When the secondary
control is implemented locally inside DGs, this level of control will be decen-
tralized and is referred to as distributed control. On the other hand, if the secondary
control is integrated into the upper level controller (controller in the tertiary level),
it will be centralized (thus referred as centralized approach).
1.4.1 Centralized approach

As mentioned before, the MG secondary controller is used to recover the DC bus
voltage deviation caused by primary control. In order to achieve this functionality,
the DC bus voltage is measured and sent to a central controller. A low bandwidth
communication channel can be used for this application [2]. The central controller
will collect the data from local sensors, process the data and send out commands to
local primary controllers. As the signals are always transmitted in one direction, the
communication is normally very reliable.
As shown in Figure 1.19, the DC bus voltage is measured remotely and sent to
a central controller. The error between the measured DC bus voltage and its
reference value VMG-ref is proceed in the controller in the secondary control level.
Communication link Remote measurement
δVo1
vdc1
*
vdc PWM
vdc Voltage Current
Vo DC source
VMG* –
GV
RD1
DC bus
Secondary Control idc1
δVo2
vdc2
vdc* Voltage Current PWM
Vo DC source
RD2
idc2
Primary control
Figure 1.19 A conceptual diagram of centralized primary and secondary controls

of a DC MG with power and information flow
Vdc
δVo
V *MG
VMG_p
idc
Figure 1.20 Droop curve lifted up by the secondary controller
A droop shifting value dvo is generated from the secondary controller and sent to
each DGs through communication links. To adjust power sharing dynamically,
different shifting values can be sent to different DGs. When embedding the droop
shifting into the DGs, the V–I droop (1.1) becomes
vDC ¼ v0 þ dv0 RD iDC (1.11)
Another figure showing the curve shifting is shown in Figure 1.20. As can be
seen, a droop shifting value dV0 from the secondary controller is used to adjust the
no-load voltage V0 in the primary control level. After this compensation, the droop

curve is lifted up, and DC voltage is increased from VMG_P to VMG .
As can be seen from Figures 1.19 and 1.20, a reliable operation of secondary
control depends on effective communication between the two levels and a reliable
MG central controller. Any fault that occurs in the communication channel or the
central controller may jeopardize the functionality of the secondary control.

Under these faulty conditions, the deviation of the DC bus voltage is totally sub-
jected to load change.
1.4.2 Distributed approach

To improve system reliabilities, the distributed secondary control have been pro-
posed [7,17–19]. In this approach, the DC bus voltage regulation is entirely realized
locally within DGs. This is the most significant difference compared with the
centralized approach. The communication between DGs can be achieved either
through dedicated communication channels or through the power line.
1.4.2.1 Communication through dedicated communication channels

A general diagram is shown in Figure 1.21. In this diagram, the average current
sharing (ACS) [7] technique is implemented in the secondary control level. This
technique is incorporated within each DG. The communication links can be either
analogue or digital.
The operation principle of an analogue communication bus can be explained in
Figure 1.22. The voltage VNn can be calculated as
Pn
Vi =Ri
VNn ¼ Pi¼1
n (1.12)
i¼1 1=Ri
Communication
Secondary
ACS ACS ACS control
Droop Droop Droop Primary

inner loops inner loops inner loops control
Converter-1 Converter-2 Converter-3
DC microgrid
Load-1 Load-2
Physical link Communication link

(power flow) (information flow)
Figure 1.21 Decentralized primary and secondary controls of a DC MG with

power and information flow [7]
VNn N
R1 R2 Rn
+ + +
V1 V2 ... Vn
– – –
Figure 1.22 Equivalent circuit for the average current signal bus
R1
Rd idc
*
– vdc Voltage Current PWM
Vo DC source
loop loop generator
Average current signal bus
dVo
∑i vdc
K
n
DC bus
R2
Rd idc
*
Vo DC source
loop loop generator
dVo
∑i vdc
K
n
Figure 1.23 ACS control for parallel DC–DC converters using analogue
communication
If resistance R1, R2, . . . , Rn are equal, (1.12) can be written as

Pn
Vi
VNn ¼ i¼1 (1.13)
n
From (1.13), we can conclude that if the resistance R1, R2, . . . , Rn are equal, the
voltage VNn is the average of V1, V2, . . . , Vn.
This concept has been adopted in ACS control for paralleled DC–DC con-
verters as shown in Figure 1.23 [7]. The currents of DC/DC converters are mea-
sured with current sensors. The outputs of these sensors are essentially some
voltages which are proportionate to the value of measured currents. When con-
necting the current sensor output to a resistor (R1, R2), an equivalent circuit in
pu
idcj
1/irated
j
Rd
idcj
*
Vo DC source
loop loop generator
DACS dVo
Communication link
n vdc
∑i
j =1
pu
j
irated K
j
DC bus
n
pu
idcm
1/imrated
Rd idcm
*
Vo DC source
DACS dV loop
o
loop generator
n
∑i
j =1
pu
j
i mrated K
vdc
n
Figure 1.24 ACS-based distributed control using CAN bus as the communication
channel
Figure 1.22 will be created. The voltage on the average current signal bus is, hence,
reflecting the average value of DC currents within different DC/DC converters.
As analysed in the primary control section, with conventional droop control, the
voltage variation is subjected to load changes (current changes). If the load
increases, according to the droop curve, the DC voltage will decrease. Thus, the
average current measured from average current signal bus will also increase.
Therefore, the droop shifting value is increased, and the droop curve is shifted up,
as shown in Figure 1.20. The voltage deviation caused by primary droop control is
compensated.
Although ACS method is an effective approach to restore DC bus voltage, this
method is only effective when DGs are configured close to each other. When DGs
are widely spread with long distance in between, since the average current signal
bus has to be deployed along with power lines, this makes average current bus
susceptible to noise. This will degrade the system performance when using ACS
method for the secondary control.
To overcome the above-mentioned shortcomings of ACS, ACS-based dis-
tributed method using controller area network (CAN bus) has been proposed, as
shown Figure 1.24 [7]. The nature of CAN is to provide reliable digital commu-
nications between controllers and devices without a host. This makes CAN bus
rather suitable for the secondary control application. Since all the current infor-
mation shared on the CAN bus are normalized values, this makes the current
sharing algorithm simpler and MG easier to expand. To distinct from analogue-
based ACS, the CAN bus-based implementation is termed as digital ACS (DACS).
Normalized averaged current is calculated in DACS and transformed into a droop
shifting signal by multiplying its current normal base irated
j and shifting gain kj.
idcj Receiving
Secondary BPF
control Broadcasting
–
iref_b
dVo Gr(s)
+ iref_PLS idcj
DC bus
*
vdc PWM DC source
Vo Gv(s) Gi(s)
generator
– vdc –
Rd
DGj
DGm
Figure 1.25 Block diagram of the PLS-based decentralized secondary control

scheme [19]
Afterwards, the droop shifting signal Dv0j (the same purpose as dvo in Figure 1.20)
is sent to the primary controller to restore the bus voltage.
1.4.2.2 Communication through power lines

The utilization of power distribution line as a communication channel to carry and
transmit both power and information is nothing new. The investigation on Power
Line Signalling (PLS) has been started not long after electrical power supply
became widespread. The idea of PLS has been introduced to DC MG in [19] as
shown in Figure 1.25. Sinusoidal signal is injected into the power line by using a
resonant controller in DGj. The sinusoidal reference is generated from the local
secondary controller. A band-pass filter is employed to extract the frequency
component signal embedded in the DC bus voltage. By using such communication,
the current information can be shared among DGs, and DC voltage compensation
can be achieved in the same way as discussed in above sub-section.
1.5 Tertiary control

Whenever there is a requirement to connect MG to an upper stiff grid, a DC bus
voltage synchronization process has to be initiated. Distinctively from AC MGs,
only DC bus voltage needs to be synchronized in the DC MGs as shown in Fig-
ure 1.26. A bypass switch is required to remain off to avoid inrush current before
the secondary control tracks down VDC. Once the deviation between VMG and VDC
is controlled under a predefined threshold, a mode transfer signal will be sent from
tertiary controller to turn on the bypass switch. The tertiary controller at this point
T
will also alter the secondary reference (VMG ) from VDC to VMG . By doing this, the
whole MG will be seen by the stiff DC grid as a current source, since there is a
+
VDC VMG
DC microgrid
–
Stiff DC grid sampling
Secondary dVo
Primary control level

* control
VMG
level
–
iG* + GT (s)
T
VMG
Operation control
Tertiary control level
Figure 1.26 Tertiary control to manage power exchange between MG and upper
grid with synchronization loop of a DC MG [2]
Vdc Vdc Vdc
VH Mode 1
VM Mode 2
VL Mode 3
MPPT
P P P
Pg–
– +
Pbat Pbat Pg+
Renewables Battery Utility grid
Figure 1.27 Adaptive autonomous mode structure
current (or power) control loop in tertiary control level and the current reference is
set by the command iG .
Once the MG is connected to the upper grid, it communicates with the dis-
tribution system operator (DSO) or transmission system operator (TSO) [1] and the
secondary control level. System power management can be implemented based on
DC Bus Signalling (DBS). The concept of DBS technique has been introduced in
literature [15,16]. In this method, the DC bus voltage itself is used as a commu-
nication signal between DGs. Agreements have been made beforehand among
DGs. DGs will take different operation actions under different DC bus voltage
ranges. Sources in the DC MGs can be classified into three main categories:
renewables, storages and utility grid as shown in Figure 1.27. When the bus voltage
is higher than VH (Mode 1), DC bus voltage is regulated by renewables and battery/
utility grid receives power generated from renewables. When bus voltage falls
below VH but stays above VM (Mode 2), DC bus voltage is supported by battery.
In Mode 2, renewables are working in Maximum Power Point Tracking and power
Vdc Vdc Vdc
Mode 1
Mode 2
Mode 3
MPPT
P P P
–
Pbat +
Pbat Pg– Pg+
Renewables Depleted battery Utility grid
Figure 1.28 Adaptive autonomous droops with depleted battery
is feeding into the utility grid. When the bus voltage drops below VM (Mode 3), grid
converters adjust the DC bus voltage. Both renewables and battery work as current
sources. Due to the existence of the secondary control level, the final steady state
bus voltage is always restored to the reference. All the curves shown in Figure 1.27
can be shifted up or down as an entity by the secondary controller.
Each droop curve in Figure 1.27 can be modified in real time online by tertiary
control according to system requirements. For instance, when battery has low State
of Charge (SOC), Pþ bat is forced to zero and the droop curve for utility grid is lifted
up accordingly, as shown in Figure 1.28.
Based on the above mode structure, DSO/TSO gives command to tertiary
control level to determine the power flow between MG and the upper grid by
changing Pþ
g and Pg . Based on information gathered from both DSO/TSO and the
MG, the tertiary controller prepares the source and storage dispatch schedule to
optimize the operation which is communicated to the secondary and primary levels.
From this point, tertiary control level shares similarity for both AC and DC MGs.
Therefore, the existing solutions on energy management for AC MGs can be
adapted for DC MG applications. Greater details concerning energy management
will be discussed in other chapters.
1.6 Summary
In this chapter, the hierarchical control of DC MG is introduced. The definitions for

each control levels have been discussed. Primary control is responsible for DGs
load sharing and is predominately implemented using the droop control. The droop
control can be perceived as a virtual resistance, and its value can affect system
stability and maximum DC bus voltage deviation. Two inherent issues with con-
ventional droop control are discussed. Both terminal DC voltage control accuracy
and cable resistance have impacts on the power sharing among DGs. Droop control
can be mainly divided into two groups: current-based droop and voltage-based
droop. The difference is the implication on system stability. In addition, another
nonlinear droop curve is also mentioned with adaptive droop gain. It can reduce
power sharing error in heavy load. Furthermore, dynamic power sharing during
load transition is also discussed in this section. Different load dynamic spectrums
are assigned to different DGs according to their dynamic response. Finally, to
receive commands from upper control levels, interfaces are pointed out without
changing the basic primary control loop structure.
The purpose of the secondary control is to restore the DC bus voltage deviation
caused by conventional droop control from primary control level. It can be
implemented remotely in an MG central controller (Centralized approach) or
locally inside each DG (distributed approach). Both dedicated analogue and digital
communication links can be used in a decentralized secondary control to transmit
current signals. To enhance the reliability of the control, a communication link
based on power lines is also possible.
The tertiary control level is responsible for the connecting process of MG to
the upper grid. A basic mode structure based on DBS is introduced to accommodate
energy-management algorithms.
References
[1] T. L. Vandoorn, J. C. Vasquez, J. De Kooning, J. M. Guerrero, and
L. Vandevelde, ‘‘Microgrids: Hierarchical Control and an Overview of the
Control and Reserve Management Strategies,’’ IEEE Ind. Electron. Mag.,
vol. 7, no. 4, pp. 42–55, Dec. 2013.
[2] J. M. Guerrero, J. C. Vasquez, and R. Teodorescu, ‘‘Hierarchical control of
droop-controlled DC and AC microgrids—a general approach towards
standardization,’’ in 2009 35th Annual Conference of IEEE Industrial
Electronics, 2009, pp. 4305–4310.
[3] C. N. Papadimitriou, E. I. Zountouridou, and N. D. Hatziargyriou, ‘‘Review
of Hierarchical Control in DC Microgrids,’’ Electr. Power Syst. Res.,
vol. 122, pp. 159–167, May 2015.
[4] C. Jin, P. Wang, J. Xiao, Y. Tang, and F. H. Choo, ‘‘Implementation of
Hierarchical Control in DC Microgrids,’’ IEEE Trans. Ind. Electron.,
vol. 61, no. 8, pp. 4032–4042, Aug. 2014.
[5] F. Gao, S. Bozhko, A. Costabeber, et al., ‘‘Comparative Stability Analysis of
Droop Control Approaches in Voltage-Source-Converter-Based DC Micro-
grids,’’ IEEE Trans. Power Electron., vol. 32, no. 3, pp. 2395–2415, Mar. 2017.
[6] D. Xu, H. Li, Y. Zhu, K. Shi, and C. Hu, ‘‘High-surety Microgrid: Super
Uninterruptable Power Supply with Multiple Renewable Energy Sources,’’
Electr. Power Compon. Syst., vol. 43, no. 8–10, pp. 839–853, Jun. 2015.
[7] S. Anand, B. G. Fernandes, and J. Guerrero, ‘‘Distributed Control to Ensure
Proportional Load Sharing and Improve Voltage Regulation in Low-Voltage
DC Microgrids,’’ IEEE Trans. Power Electron., vol. 28, no. 4, pp.
1900–1913, Apr. 2013.
[8] Y. A.-R. I. Mohamed, ‘‘Robust Droop and DC-Bus Voltage Control for
Effective Stabilization and Power Sharing in VSC Multiterminal DC Grids,’’
IEEE Trans. Power Electron., vol. 33, no. 5, pp. 4373–4395, May 2018.
[9] P. Karlsson and J. Svensson, ‘‘DC Bus Voltage Control for a Distributed
Power System,’’ IEEE Trans. Power Electron., vol. 18, no. 6, pp. 1405–1412,
Nov. 2003.
[10] J. Jiang, ‘‘Design of an Optimal Robust Governor for Hydraulic
Turbine Generating Units,’’ IEEE Trans. Energy Convers., vol. 10, no. 1,
pp. 188–194, Mar. 1995.
[11] K. De Brabandere, B. Bolsens, J. Van den Keybus, A. Woyte, J. Driesen,
and R. Belmans, ‘‘A Voltage and Frequency Droop Control Method
for Parallel Inverters,’’ IEEE Trans. Power Electron., vol. 22, no. 4,
pp. 1107–1115, Jul. 2007.
[12] N. R. Chaudhuri and B. Chaudhuri, ‘‘Adaptive Droop Control for Effective
Power Sharing in Multi-Terminal DC (MTDC) Grids,’’ IEEE Trans. Power
Syst., vol. 28, no. 1, pp. 21–29, Feb. 2013.
[13] J. Beerten and R. Belmans, ‘‘Analysis of Power Sharing and Voltage
Deviations in Droop-Controlled DC Grids,’’ IEEE Trans. Power Syst.,
vol. 28, no. 4, pp. 4588–4597, Nov. 2013.
[14] F. Gao, Y. Gu, S. Bozhko, G. Asher, and P. Wheeler, ‘‘Analysis of droop
control methods in DC microgrid,’’ in 2014 16th European Conference on
Power Electronics and Applications, 2014, pp. 1–9.
[15] Y. Gu, W. Li, and X. He, ‘‘Frequency-Coordinating Virtual Impedance for
Autonomous Power Management of DC Microgrid,’’ IEEE Trans. Power
Electron., vol. 30, no. 4, pp. 2328–2337, Apr. 2015.
[16] Y. Gu, X. Xiang, W. Li, and X. He, ‘‘Mode-Adaptive Decentralized Control
for Renewable DC Microgrid with Enhanced Reliability and Flexibility,’’
IEEE Trans. Power Electron., vol. 29, no. 9, pp. 5072–5080, Sep. 2014.
[17] K. Sun, L. Zhang, Y. Xing, and J. M. Guerrero, ‘‘A Distributed Control
Strategy Based on DC Bus Signaling for Modular Photovoltaic Generation
Systems With Battery Energy Storage,’’ IEEE Trans. Power Electron., vol.
26, no. 10, pp. 3032–3045, Oct. 2011.
[18] L. Zhang, T. Wu, Y. Xing, K. Sun, and J. M. Gurrero, ‘‘Power control of DC
microgrid using DC bus signaling,’’ in 2011 Twenty-Sixth Annual IEEE
Applied Power Electronics Conference and Exposition (APEC), 2011,
pp. 1926–1932.
[19] T. Dragicevic, J. M. Guerrero, and J. C. Vasquez, ‘‘A Distributed Control
Strategy for Coordination of an Autonomous LVDC Microgrid Based
on Power-Line Signaling,’’ IEEE Trans. Ind. Electron., vol. 61, no. 7,
pp. 3313–3326, Jul. 2014.
Chapter 2
Distributed and decentralized control of
dc microgrids
Saeed Peyghami1, Hossein Mokhtari2, and
Frede Blaabjerg1
2.1 Introduction
Power sharing control of dc power sources in dc microgrids is an important issue in
order to obtain a stable and reliable operation [1–7]. Power sharing approaches in
three levels of the hierarchal control system are described in the last chapter. Two
major objectives of this method of control are proportional power sharing and
appropriate voltage regulation in dc microgrids which can be carried out using
centralized and decentralized/distributed approaches [2,3,6,7]. Centralized
approaches deal with lower reliability and resiliency. In order to enhance the resi-
liency of the power sharing methods, distributed and decentralized controls have
been introduced. Distributed approaches employ sparse communication systems
instead of point-to-point communication control among different converters (and
maybe a central control unit). On the contrary, decentralized approaches use no
communication (or even physical communication link) in the corresponding control
system.
Decentralized approaches can be categorized as
1. Mode adaptive (autonomous) droop control
2. Nonlinear droop control
3. Frequency droop control
The voltage droop method is the simplest decentralized approach employed in dc
microgrids. This method can perfectly control the load sharing among power
sources in the case of short distances where the line resistances can be neglected. In
this case, mode adaptive or autonomous droop approach can be used to control the
output power of different (dispatchable and nondispatchable) sources in the dc
microgrid. Considering the line resistances of long distances among converters, the
conventional droop method cannot properly carry out the power sharing objectives,
1
Department of Energy Technology, Aalborg University, Denmark
2
Department of Electrical Engineering, Sharif University of Technology, Iran
and hence, nonlinear droop approach and frequency droop methods have been
introduced in order to reach the power management objectives.
Furthermore, some distributed approaches have been presented which can be
classified as
1. Fully communicated control
2. Sparse communicated (consensus-based) control
3. Sparse communicated control using current information.
In the distributed approaches, each converter is equipped with a secondary con-
troller cascaded by the primary controller. A communication system is employed to
share the voltage and current information between all of the converters. In a simple
distributed secondary control, it is required that all of the converters have access to
the information each other. This approach may be complex and has a lower resi-
liency; however, consensus strategy with a sparse communication network has been
presented and can use the neighboring converter information to reach an
acceptable operating condition. Furthermore, another sparse communicated control
approach can reach the power management objectives by only employing the cur-
rent information of the converters. In this chapter, decentralized and distributed
control concepts are discussed.
2.2 Decentralized approaches

In this section, the decentralized power sharing approaches are conceptually
explained. Decentralized methods do not require any communication link between
the converters and hence introduce higher reliability and stability. The voltage
droop control is the simplest decentralized method employed in dc microgrids. This
method can perfectly control the load sharing among the power sources in the case
of short distances where the line resistances can be neglected. In this case, mode
adaptive or autonomous droop approach can be used to control the output power of
different (dispatchable and nondispatchable) sources in the dc microgrid. Con-
sidering the line resistances of long distances among converters, the conventional
droop method cannot properly control the power sharing, and hence, nonlinear
droop approach and frequency droop methods have been presented in order to
appropriately manage the load power among converters. The decentralized
approaches are discussed in the following section.
2.2.1 Mode-adaptive (autonomous) droop control

dc Microgrids, in practice, contain dispatchable, nondispatchable and storage units.
A voltage droop-based approach can be employed to control the power and energy
flow in the microgrid by defining suitable droop characteristics in order for dif-
ferent kinds of energy units to operate in an autonomous mode. A simplified dc
microgrid is shown in Figure 2.1. In the following section, possible droop schemes
for different kinds of energy units are explained.
Distributed and decentralized control of dc microgrids 25
R1
Interlinking
Dispatchable unit converter
AC grid
R4
R2
Non dispatchable unit
R3
Distributed storage Distributed loads
Figure 2.1 Typical dc microgrid with dispatchable and nondispatchable units,

distributed storage, interlinking converter and distributed loads—DG.
DG, distributed generation
Power demand Voltage SoC range Voltage Voltage Voltage
SoC 0% SoC 100%

Rated power
Rated power
Rated power
Rated power
Rated power
Rated power
MPPT range
SoC range
Pinjected 0 Pabsorbed Pcharge 0 Pdischarge 0 Prated 0 Prated
(a) (b) (c) (d)
Figure 2.2 Droop characteristics for different energy units in dc microgrid with
constant droop slopes; (a) IC, (b) storage converter, (c) dispatchable
units and (d) nondispatchable units
Typical droop characteristics for an interlinking converter (IC), a distributed

storage unit, and dispatchable and nondispatchable units are shown in Figure 2.2
[8]. For an IC, a bidirectional droop can be considered, such as the one shown in
Figure 2.2(a). The maximum injected power into the grid must be determined by
the transmission/distribution system operator (TSO/DSO). However, the required
power can be supported by the grid which can be determined by the droop
characteristic shown in Figure 2.2(a). For the storage droop control shown in
Figure 2.2(b), such as a battery, a regenerative fuel cell, or a flywheel, the max-
imum and minimum limits can be determined by the energy level or state of charge
(SoC) level of the battery. Dispatchable energy units such as microturbines and fuel
cells can be controlled with a droop curve as shown in Figure 2.2(c). Finally, as
shown in Figure 2.2(d), nondispatchable units such as photovoltaic arrays and

wind turbines can operate in a droop mode from zero to an Maximum Power Point
Tracking (MPPT) power. Therefore, MPPT-based units can support the microgrid
under MPPT power. With this approach, only the maximum and minimum limits of
the droop characteristics are updated, based on the SoC or MPPT limits. Further-
more, in the droop mode operation, converters work as grid supporting converters,
where at a constant power mode, they work as grid feeding converters.
Adjusting power limits may cause voltage instability because of the lack of a
voltage forming or supporting converter at some loading and sourcing conditions.
For example, if the power demand of an IC is zero, the storage is fully charged, and
the nondispatchable unit operates at an MPPT mode; hence, the droop character-
istics must be modified as shown in Figure 2.3(a), (b) and (d), respectively. If the
load is equal to the sum of the rated powers of the dispatchable unit and the MPPT
power of a nondispatchable unit, the voltage can have any value within the gray
region in Figure 2.3. In this area, there is no voltage forming converter, and the
voltage cannot converge to a steady-state value.
To avoid all converters operating at constant power mode at the same time,
which may occur at some loading and sourcing conditions, in addition to adapting
the power limits, the slope of the droop controls should be adjusted, and this
approach is illustrated in Figure 2.4. However, a control system needs to be
Voltage Voltage Voltage Voltage
Pinjected 0 Pabsorbed Pcharge 0 Pdischarge 0 Prated 0 PMPPT Prated

(a) (b) (c) (d)
Figure 2.3 An unstable region in droop characteristics of different energy units in a

droop controlled dc microgrid with constant droop slopes; (a) IC, (b)
storage converter, (c) dispatchable units and (d) nondispatchable units
Power demand Voltage SoC range Voltage Voltage Voltage
SoC 100%
Rated power
Rated power
Rated power
Rated power
Rated power
Rated power
SoC 0%
MPPT range
SoC range
Pinjected 0 Pabsorbed Pcharge 0 Pdischarge 0 Prated 0 Prated
(a) (b) (c) (d)
Figure 2.4 Droop characteristics for different energy units with adjustable
droop slopes; (a) IC, (b) storage converter, (c) dispatchable units
and (d) nondispatchable units
designed to guarantee the system’s stability at different loading and sourcing

conditions, since the droop gains affect the system’s stability.
The droop characteristics shown in Figures 2.2 and 2.4 guarantee parallel
operation of the ICs in the dc microgrid. However, in the case of islanded hybrid
ac–dc microgrids, as well as the absence of the tertiary control DSO/TSO, it would
be better to support the microgrid demand by its internal sources. Hence, the excess
power or required power can be supported by the ICs. This approach is capable of
controlling the energy flow between ac–dc microgrids [9]. The droop character-
istics of two parallel ICs are shown in Figure 2.5(b) and (c). If the voltage lies
between VH and VL, the power of the ICs is zero. Otherwise, the output power of
the ICs can be determined by the droop characteristics as shown in Figure 2.5(b).
Furthermore, the power sharing between ICs can be properly managed by this
droop control approach [9].
Different types of droop characteristics may be defined for various energy
sources in the dc microgrid. To operate the microgrid in an autonomous mode, it is
important to determine a voltage range for the droop gain of each unit in order to
Interlinking
R1 converter 1
R4
Dispatchable unit
Interlinking AC grid
R2 converter 2
R5
Nondispatchable unit
R3
Distributed storage Distributed loads

(a)
IC 1 Voltage IC 2 Voltage
VH Operating point VH
VL VL
(b) 0 Power (c) 0 Power
Figure 2.5 Typical ac–dc microgrid with two ICs; (a) the single line diagram
of the power system and (b), (c) the droop characteristics of the
interlinking converters (ICs)
Voltage SoC range Voltage Voltage

Power demand
SoC 100% Mode I
VH
Rated power
Mode II
Rated power
VL
Rated power
Rated power
Rated power
Mode III SoC 0%
MPPT range
SoC range
Pinjected 0 Pabsorbed Pcharge 0 Pdischarge 0 Prated

(a) (b) (c)
Figure 2.6 An autonomous droop-based power sharing in dc microgrid; (a) IC,

(b) battery converter, (c) photovoltaic array
Converter 1 X1 R1 PCC R2 X2
Converter 2
Io1 Io2
Cdc Cdc
+ V Vo1 Line 1 Line 2 Vo2 Vi2 +
i1
– –
PWM PWM
Vo1 Vo2
Voltage Inner Inner Voltage,
current controller controller current
Io1 Io2
Vref,1 Vref,2
V* Droop Droop
+– Rd,1 Rd,2 +– V*
controller controller
Figure 2.7 Simplified dc microgrid with primary droop control
have at least one voltage forming converter in the microgrid. For example, an
autonomous control system based on the droop is shown in Figure 2.6 [10]. As can
be seen in Figure 2.6, in Mode I, the voltage is controlled by the photovoltaic array.
In Mode II, a battery is responsible for forming the voltage, and in Mode III, the IC
controls the dc-link voltage.
2.2.2 Nonlinear droop control

In the conventional droop control, considering the higher virtual resistor Rd,i
improves the current sharing accuracy but deteriorates the voltage regulation. The
relationship between the current sharing error and voltage regulation by increasing
the virtual resistors Rd,1 ¼ Rd,2 for the simplified dc microgrid is shown in Fig-
ure 2.7 and is illustrated in Figure 2.8. In heavy-load conditions, unlike light load
conditions, the current sharing accuracy is very important, since overloading a
converter may affect its corresponding lifecycle. Therefore, large virtual resistors
need to be employed in heavy-load condition. However, larger virtual resistors
create a large voltage drop in the microgrid; hence, an adaptive technique is
required to compensate for the voltage drop. This is a secondary controller.
80 Current sharing error

Voltage regulation
60
40
%
20
0
0 0.5 1 1.5 2 2.5 3
Droop gain (Rd,1 = Rd,2)
Figure 2.8 Performance of the conventional droop control for the simplified dc
microgrid, current sharing error and voltage regulation
Secondary control approaches are explained in the first chapter. The secondary
controller requires communication infrastructures to share the voltage and current
information of converters to regulate the dc-link voltage. In order to avoid utilizing
a communication system, an adaptive nonlinear approach has been used, presented
in [4]. In this approach, the reference voltage Vref,i of the ith converter can be
defined as (2.1), where mi and a are the droop curve constant and coefficients,
respectively [4]. mi can be determined by (2.2), where Vmin is the minimum
allowable voltage and Imax are the maximum current of ith converter.
Vref ;i ¼ V mi Iia (2.1)

V Vmin
mi ¼ a (2.2)
Imax;i
The nonlinear droop curve is graphically shown in Figure 2.9. As can be seen, the
slope of the droop curve slope is larger in the higher currents, which reflects
accurate current sharing in heavy-loading condition. Furthermore, the dc-link
voltage remains in the acceptable interval. Considering the effective droop curves
(the tangent line of the nonlinear curve) shown in Figure 2.9, the droop parameters
including the slope and voltage references can automatically be adapted by varying
the load current. Actually, with the conventional droop approach, the droop curve
needs to be shifted by a secondary controller to regulate the dc voltage at an
acceptable interval. However, by utilizing the nonlinear droop method, the effec-
tive voltage reference can be determined by the droop curve [4], where the effective
voltage shifting can be calculated as follows:
ða 1ÞðV Vmin Þ a
DVi ¼ a Ii (2.3)
Imax;i
The current sharing error and voltage regulation performance of the conventional
and nonlinear droop methods are illustrated in Figure 2.10. As is shown, employing
V Nonlinear droop curve

Effective droop gain
Effective voltage shifting
∆V3 Rd2 Rd3
∆V2
V* ∆V1
Rd1
Vmin
I
I1 I2 I3
Figure 2.9 Characteristics of the nonlinear droop control method
40 Droop gain = 0.2

Droop gain = 2
Voltage regulation (%)
30 Nonlinear approach
20
10
0
2 4 6 8 10 12 14 16 18
(a) Load current (A)
40
Current sharing error (%)
Droop gain = 0.2

Droop gain = 2
30 Nonlinear approach
20
10
0
2 4 6 8 10 12 14 16 18
(b) Load current (A)
Figure 2.10 Performance comparison of the nonlinear method with the

conventional droop control; (a) voltage regulation and (b) current
sharing error
small droop gains causes acceptable voltage regulation but leads to high current
sharing error. Although selecting higher droop gains causes appropriate current
sharing, it leads to a large voltage drop in the higher load currents. However,
employing the nonlinear curve introduces an acceptable current sharing in a heavy-
load condition and appropriates voltage regulation in different load currents as can
be seen from Figure 2.10.
2.2.3 Frequency droop control

A typical dc microgrid with distributed loads is shown in Figure 2.11(a). The dc
source can be a dispatchable unit such as a fuel cell module or a hybrid battery/
nondispatchable unit such as a photovoltaic array, which can control the dc-link
voltage as a voltage source converter. The proposed control approach for the kth
unit is shown in Figure 2.11(b). Each converter modulates a small ac voltage
R2+jX2 Converter N
Converter 1 ViN
Vi1 IoN V
lo1 oN
SN ILN +
+ IL1 S1 P2 –
– Vo1
R1+jX1
R3+jX3
P1
P3
R4+jX4 R6+jX6
P4
Vi2 Converter 2 Converter k Vik
Io2 R5+jX5 P5 lok Vok
+ IL2 S2 Sk ILK +
– Vo2 RN+jXN –
PN
(a)
Power sharing controller for kth unit Inner voltage and
dk current regulators Sk
iok fk
dfk Eq. 1 2π ∫ Sine A
~
Vk PWM
Vo*
Vok Power Qk
dp G(s) –+ +
+– Gv(s) +– Gi(s)
iok calc.
Vok ILk
(b) (c)
Figure 2.11 Block diagram of (a) a typical dc MG with distributed loads,

(b) proposed control structure for the kth converter, (c) inner voltage
and current control loops—Gv(s) and Gi(s) are PI-based inner
voltage and current controllers. PI, proportional–integrator
superimposed onto the dc voltage, where the frequency of the ac voltage is pro-
portional to the output dc current of the converter. The rated frequency should be
selected to be smaller than the bandwidth of the inner voltage controller, and hence,
to be regulated by a proportional–integrator (PI)-based voltage regulator. There-
fore, the inner voltage (Gv(s)) and current (Gi(s)) controllers in Figure 2.11(c) can
modulate the reference voltage including dc voltage and superimposed ac voltage.
From the ac voltage point of view, the converters are working like a synchronous
generator; hence, they can be coordinated together with the common frequency
[2,11,12]. From the power system dynamics and control theory, for analyzing the
dynamic behavior of a Synchronous Generator (SG) in an ac power system, it can
be modeled as two SGs; one being the specified SG and the other modeling the
entire power system. Moreover, the two SGs can be simplified as a single-machine-
infinite-bus, where the infinite bus is considered as a stiff ac source [13]. Therefore,
since the proposed approach is based on the SG principles, without losing the
generality, a simplified dc MG, with two converters connected to a load at a point
of common coupling (PCC), is considered, and the block diagram of the system
with the corresponding signals is shown in Figure 2.11.
According to Figure 2.12, if the output dc voltage of the converters (Vo1, Vo2)
is settled at a reference value (Vo* ), their output dc current (Io1, Io2) will be inversely
proportional to the corresponding line resistances (i.e., Io1/Io2 ¼ R2/R1), where R1
and R2 denote the line resistance of the first and second converter, respectively.
Adjusting the output dc voltage of the converters is the only option for controlling
the corresponding output currents at a desired value, for example, proportional to
their rated current, which requires the coordination of the converters. For the
coordination between converters, a small ac voltage, i.e., ~v k ¼ A sin ð2pfk t), is
superimposed onto the dc voltage reference and modulated by each converter.
~
Converter 1 io1 = lo1+io1 X1 R1 Load
+ io1 Line 1
Vi1 S1 vo1 VPCC
–
~
vo1 = Vo1 + vo1 A A
~
vo2 = Vo2 + vo2 d Vo2
Vo1
Converter 2 io2 = lo2 + io2

~
X2 R2
+ S2 io2 Line 2
Vi2 vo2
–
Figure 2.12 Conceptual illustration of the proposed strategy showing the injected
ac voltages and corresponding currents in a simplified dc MG based
on two dc–dc converters
The amplitude of the superimposed voltage (denoted as A) is considered as a small

constant value with a small ripple factor; however, it should be detectable by
measurement. Furthermore, the corresponding frequency should be proportional to
the output current of the converter. It can be defined as [2,11,12]
fk ¼ f dfk iok (2.4)
where f * (50 Hz) is the nominal frequency, iok being the output current, and dfk
is the frequency droop gain of the kth converter determined by the following
equation:
fmax fmin
dfk ¼ ; k ¼ 1; 2 (2.5)
In;k
with fmax/fmin being the maximum/minimum frequency for tuning the droop gain
and In,k is the nominal current of kth converter. At steady-state condition, the fre-
quency of the units has the same value; hence, the output current of the units has to
be shared in proportion to their rated current as shown in (2.6), where hi denotes
the average value and x is the ratio of the rated current of the converters.
hio1 i Io1 In;1 df 2

¼ ¼ ¼ ¼x (2.6)
hio2 i Io2 In;2 df 1
Frequency droop can be used to coordinate the converters by a common injected

frequency; therefore, the dc currents need to be regulated by the frequency to
control the power sharing. However, the dc currents are determined by the dc
voltages as
Vo1 VPCC Vo2 VPCC
Io1 ¼ ; Io2 ¼ (2.7)
R1 R2
with R1 and R2 being the line resistances. Therefore, the dc voltages should be
adjusted by an ac injected variable related to the frequency. According to Fig-
ure 2.11 and (2.6), the phase angle of each unit (d1, d2) can be found as
2p 2p
d1 ¼ f df 1 Io1 ; d2 ¼ f df 2 Io2 (2.8)
S S
where S is the Laplace operator. The relative phase (d) between the converters,
thus, is equal to
2p
d ¼ d1 d2 ¼ df 2 Io2 df 1 Io1 (2.9)
S
If the output currents are not proportional to the rated currents, the phase angle will
not be zero; hence, this phase difference causes a small ac power flow. As the load
impedance is higher than the line impedances, the small ac power will only flow
between the converters. According to the ac power flow theory, ac power is pro-
portional to the ac currents (~i 1 ,~i 2 ). Furthermore, the ac currents are proportional to
the line impedances. Thereby, the ac power contains the information of the line
impedances. On the other hand, in LV systems with low X/R ratio, the reactive
power can accurately be controlled by the frequency [14]. Therefore, employing the
injected reactive power (Q) of the converters to adjust the dc voltage reference (Vo* )
causes proper current sharing. Applying the proposed control algorithm, the output
dc voltage of the converters can be written as
Vo1 ¼ Vo dp Q1 GðsÞ; Vo2 ¼ Vo dp Q2 GðsÞ (2.10)
where dp is the coupling gain between dc voltage and reactive power, and G(s) ¼
wc/(sþwc) is a low pass filter to eliminate the high-frequency component of the
calculated reactive power. Therefore, the frequency droop can be used to coordinate
the converters, and the small ac power can be employed to adjust the dc voltage and,
consequently, the dc currents. Each converter can be controlled by the local mea-
sured values, and hence, like SGs, there is no need for any communication network
[2,11,12]. Furthermore, the injected ac voltage by the converters must be synchro-
nized with the ac component of the grid voltage at the startup time. The phase of the
connection bus voltage can be extracted using a phase locked loop (PLL) block.
On the other hand, in ac systems, synchronization methods are employed to make
the converter voltage close to the grid voltage in order to limit the inrush current at
the start time, which may damage the converter switches for the high currents.
However, the injected ac voltage and consequently the ac currents are very small in
the proposed approach, and hence, the converters can be connected without utilizing
a PLL. Hence, they can be synchronized based on the droop control functionality as
the grid supporting voltage source converters in ac grids [15].
2.3 Distributed approaches

In the distributed approaches, each converter is equipped with a secondary con-
troller cascaded by the primary controller. A communication system is employed to
share the voltage and current information of converters between all of them. In a
simple distributed secondary control, it is a requirement that all of the converters
have access to the information of all the others. This approach may be complex and
have a lower resiliency. However, consensus strategy with a sparse communication
network has been presented which can use the neighboring converter information to
reach an acceptable operating condition, i.e., suitable load sharing and appropriate
voltage regulation. Furthermore, another sparsely communicated control approach
can fulfill the power management objectives by only employing the current infor-
mation of the converters. The distributed approaches are explained in the following
section.
2.3.1 Fully communicated control

In this approach, secondary controllers can locally regulate the voltage of the
microgrid. A general distributed secondary controller is shown in Figure 2.13.
Primary
Secondary controller controller
V* VL
V* Voltage regulator Converter 1 RN+1
controllers
V = [V1,...,VN, Vavg V1 1
dV + V1
Inner
VL1,...,VLM]
T Averaging –+ PI +
+ +– R1
I = [I1,...,IN]T Averaging +– PI
Rd1 I1
Current regulator
VL
RN+M
V* M
V* Converter N
controllers
V = [V1,...,VN,
Vavg dV VN
Inner
+ VN
–+ +–+ RN
T
VL1,...,VLM] Averaging PI +
T
I = [I1,...,IN] Averaging +– PI IN
RdN
Current regulator
Figure 2.13 Distributed secondary controller
Vi
ith converter estimator
i
Vavg
+
j + aij 1/s +
–
Vavg
Neighbor average
estimator
Figure 2.14 Consensus global average voltage estimator [19]
A communication network collects the information of the voltage of the desired

buses, V ¼ [V1, V2, . . . , VM]T, and the output current of the converters, I ¼ [I1,
I2, . . . , IN]T. The secondary controller of each converter calculates the average
voltage (Vavg) and weighted average current of the converters and, then, regulates
its output voltage and current accordingly. Implementing the secondary controller
in the distributed policy improves the reliability of the system, since the central
controller is replaced by distributed regulators.
2.3.2 Sparse communicated (consensus-based) control

However, regulating the average of the voltages and currents requires more data
transmission through the microgrid. To overcome this problem, consensus algo-
rithms are employed to regulate the global variables in distributed systems [3,16].
In consensus algorithms, each converter needs to communicate with the neigh-
boring converters. An updating protocol as shown in Figure 2.14, also called
dynamic consensus protocol, carries out the estimated global average voltage of the
neighboring converter and the local voltage to estimate the global average voltage
[3,17–20]. The average voltage of the jth converter, V javg, can be used to calculate
the average voltage of the ith converter, V iavg, with the dynamic consensus
Primary
Secondary controller controller
V1 V*
N 1 V* Voltage regulator Converter 1
controllers
Vavg Dynamic Vavg + dV V1
V1
Inner
+ R1
From Nth N consensus – PI + ++
–
Conv. Ipu
+– PI I1
Rd1
To 2nd Current regulator
Conv.
VN V*
N–1 N Converter N
controllers
Vavg Dynamic Vavg dV VN
Inner
+ VN
From (N–1)th N–1 consensus –+ PI + +–+ RN
Conv. Ipu
+– PI IN
RdN
To 1st Current regulator
Conv.
Figure 2.15 Distributed secondary controller
estimator shown in Figure 2.14, where the coefficient aij is the weight of infor-
mation exchanged between converters i and j. After some iterations, the estimated
average voltage of the converters converges to a value which is equal to the average
voltage of the converters (Vavg). As shown in Figure 2.15, this average voltage is
regulated by the secondary controller and settles at the reference value.
On the other hand, the per-unit current of the neighboring converter can be
used to compensate the mismatch of the currents of all converters. This approach is
analogous to the circular-chain-control (CCCs) in parallel inverters in ac microgrids,
which is used to improve the current mismatches among the inverters [14,21–24].
In this approach, each converter shares the average voltage and per-unit current
with its neighboring converter. Therefore, sparse communication among the con-
verters with low volume of transmitted information improves the reliability and
stability of the system [3]. However, in consensus algorithms, the average voltage
of the grid supporting buses is only regulated. In practice, the loads may not be
connected to the grid supporting converters and might be distributed over the
microgrid. Therefore, regulating the voltage of the grid supporting buses may not
guarantee the voltage regulation at load buses.
2.3.3 Sparse communicated control using current information

Distributed secondary control using current information is shown in Figure 2.16
including two controllers for load sharing and voltage restoring in the microgrid
[25]. A current regulator is used to increase the load sharing accuracy of the con-
ventional droop controller, and a voltage regulator regulates the average voltage of
the microgrid to the nominal values, which are explained in the following section.
2.3.3.1 Current regulator

Current sharing among converters is conventionally performed by a droop gain Rdk
which can be defined as follows for the kth converter:
Vmax Vmin
Rdk ¼ (2.11)
Ink
Control unit of kth converter

V*
Voltage regulator Vk
V ref
dVi E = Vavg + dVV
controllers
++ Gk
++ – Gv(S) – +–
Inner
N Iavg dVi Rdk
I
1
N Σ a
j
+– Gi(S)
Ik
Current data
j=1 j
1/ak
Current regulator
Secondary control Primary control
l1, l2,...,lN Ik
Figure 2.16 Proposed control approach; voltage regulator Gv, current regulator
Gi, and low-bandwidth communication link with delay function of
Gd ðsÞ ¼ ð1=ð1 þ tsÞÞ [25]
where Vmax and Vmin are the maximum and minimum allowable voltage range,
respectively, and Ink is the rated current of the kth converter. Due to the differences
in line resistances, the output current of converters cannot be proportionally dis-
patched among the converters. The current regulator calculates the weighted
average currents of converters and regulates the corresponding output current
proportional to the rated current of the converter. The average current (Iavg) can be
calculated as [25]
1X N
Ij
Iavg ¼ j ¼ 1 : N; (2.12)
N j¼1 aj
where N is the number of converters, Ij is the measured current and aj is the sharing
coefficient of jth converter, respectively. Furthermore, as with the CCCs approach,
the converters can only employ the neighboring converters information to reach
proportional load sharing.
A simplified single line model of a dc microgrid with two converters is shown
in Figure 2.17. In steady state, droop gain acts as a series resistor (Rd1 and Rd2). The
secondary current regulator behaves as a small positive/negative resistor (rd1 and
rd2) such that the total resistance of each line becomes proportional to the corre-
sponding rated current. Hence, the relationship between rated current (Inj) and
sharing coefficient (aj) and total line resistance between jth converter and PCC can
be written as
Ini aj Rdj þ rdj þ rj
¼ ¼ ; i; j ¼ 1 : N ; i 6¼ j (2.13)
Inj ai Rdi þ rdi þ ri
where rj is the resistance of the line connected to jth converter.
Converter 1 δvi,1
Rd1 I1 r1 Load
rd1 Line 1
Vr E1 V1 VPCC R
Converter 2 δvi,2
Rd2 I2 r2
rd2 Line 2
Vr E2 V2
Figure 2.17 Simplified model of two converter-based dc microgrid
VPCC + r1I1
Vdc* +δvi
Rd I VPCC + r2I2
–δvi
Rd
VPCC
I
I1= I2
Figure 2.18 Proposed droop characteristic adjustment
The effect of a current regulator in a power sharing system is schematically

described in Figure 2.18. The dashed graph shows the effect of conventional droop
gain. The secondary current regulator alters the slope of this droop characteristics
to reach the same current between the two converters. Sharing coefficients are
assumed to be one for both converters. Therefore, the accurate current sharing is
obtained with a droop controller and a current regulator. However, the dc voltage of
the microgrid is dropped, and a distributed voltage regulator is required to restore
the average voltage of the microgrid.
2.3.3.2 Voltage regulator

This regulator compensates for the voltage drop though the droop gains of the
converters. From Figure 2.17, the output voltage of each converter (V1 and V2) can
be calculated as

V1 ¼ Vr Rd1 I1 dvi1
(2.14)
V2 ¼ Vr Rd2 I2 dvi2
where Vr is the reference value of the voltage loop. If sharing coefficients are
assumed to be one, then the droop gains must be equal (Rd1 ¼ Rd2 ¼ Rd) and I1 ¼ I2
at steady state. The output values of the current regulators can be obtained as
8
>
> I1 þ I2 I2 I1
< dvi1 ¼
>
2
I1 Gd Gi ðsÞ ¼
2
Gd Gi ðsÞ
(2.15)
>
> dvi2 ¼ I1 þ I2 I2 Gd Gi ðsÞ ¼ I1 I2 Gd Gi ðsÞ
>
:
2 2
where Gi(s) is the PI controller and Gd(s) is the delay of communication link. At
steady state, I1 ¼ I2, and hence dvi1 þ dvi2 ¼ 0. Therefore, the average voltage of the
microgrid is
1
V avg ¼ ðV1 þ V2 Þ ¼ Vr Rd I (2.16)
2
Applying the primary droop controller and secondary current regulator causes an
average voltage drop equal to RdI. From the single line model of the microgrid
shown in Figure 2.17, internal voltages (i.e., E1 and E2) are equal to the average
voltage calculated by (2.16). Therefore, the distributed voltage regulator can esti-
mate the internal voltage and regulate it at the reference value. In fact, the cor-
rection term (dvv) shifts up the droop characteristics in Figure 2.18 to restore the
average voltage of the microgrid which can be calculated as (2.17), where V* is the
rated voltage of the microgrid.

dvv1 ¼ ðV ðV1 þ dvi1 ÞÞGv ðsÞ
(2.17)
dvv2 ¼ ðV ðV2 þ dvi2 ÞÞGv ðsÞ
The relationship between converter voltage and PCC voltage can be shown as

V1 VPCC ¼ r1 I1
(2.18)
V2 VPCC ¼ r2 I2
where
VPCC ¼ RL ðI1 þ I2 Þ (2.19)
The set point value for the primary controller is

V1 ¼ Vref 1 Rd I1
(2.20)
V2 ¼ Vref 2 Rd I2
and the set point value for the inner voltage loop can be determined by the primary
controller as

V1 ¼ Vref 1 Rd I1
(2.21)
V2 ¼ Vref 2 Rd I2
Substituting (2.17), (2.18) and (2.20) in (2.21) results in

8
>
> Rd I1
> V1 ¼ V dvi1
< 1 þ Gv ðsÞ
(2.22)
>
> Rd I2
>
: V2 ¼ V dvi2
1 þ Gv ðsÞ
This equation shows that the term of RdI which is related to the primary controller
can be eliminated in low frequencies, i.e., in the secondary controller frequency
bandwidth. Therefore, the primary controller shares the load current between
converters based on droop gain, and the secondary controller reduces the mismatch
in current sharing as well as decreasing the voltage drop of the droop gain. This
approach uses the only current information to appropriately control the load sharing
and voltage regulation in the dc microgrid.
2.4 Conclusion and future study
In this chapter, distributed and decentralized control approaches for dc microgrids

are discussed. Distributed approaches employ a communication system among
different converters in order to regulate the dc voltage and improve the load sharing
accuracy. Some of the distributed methods utilize point-to-point communication
links among converters; however, some of them use a sparse communication sys-
tem based on consensus protocol. Sparse communication-based control approaches
are more reliable and resilient than the fully communicated methods.
On the contrary, the decentralized methods use no communication (physical
link) between converters to reach the power sharing objectives. In these approa-
ches, the control system of each converter uses local voltage and current informa-
tion to control the output power (current) of the corresponding converter. Since
these converters do not need to communicate with other converters, the overall
stability and reliability can be enhanced. The centralized control approaches can be
categorized as mode-adaptive (autonomous) droop control, nonlinear droop control
and frequency droop control, and these methods are conceptually discussed in this
chapter.
For future studies, it is important to establish new power sharing controls with
higher resiliency and reliability other than droop-based methods. These methods
should be applicable for both dispatchable and nondispatchable resources. Fur-
thermore, transient conditions such as connecting/disconnecting a converter, or
fault conditions should be taken into account in the control system to reach
stable and reliable operation. Moreover, stability issues such as constant power load
problems should be analyzed with the newly suggested control algorithms to ensure
the viability of control systems in real-operation conditions.
References
[1] Peyghami, S., Mokhtari, H., Davari, P., Loh, P.C., Blaabjerg, F.: ‘On Sec-
ondary Control Approaches for Voltage Regulation in DC Microgrids’ IEEE
Trans. Ind. Appl., 2017, 53, (5), pp. 4855–4862.
[2] Peyghami, S., Davari, P., Mokhtari, H., Loh, P.C., Blaabjerg, F.: ‘Syn-
chronverter-Enabled DC Power Sharing Approach for LVDC Microgrids’
IEEE Trans. Power Electron., 2017, 32, (10), pp. 8089–8099.
[3] Nasirian, V., Davoudi, A., Lewis, F.L., Guerrero, J.M.: ‘Distributed Adap-
tive Droop Control for dc Distribution Systems’ IEEE Trans. Energy Con-
vers., 2014, 29, (4), pp. 944–956.
[4] Khorsandi, A., Ashourloo, M., Mokhtari, H., Iravani, R.: ‘Automatic Droop
Control for a Low Voltage DC Microgrid’ IET Gener. Transm. Distrib.,
2016, 10, (1), pp. 41–47.
[5] Khorsandi, A., Ashourloo, M., Mokhtari, H.: ‘A Decentralized Control
Method for a Low-Voltage DC Microgrid’ IEEE Trans. Energy Convers.,
2014, 29, (4), pp. 793–801.
[6] Guerrero, J.M., Vasquez, J.C., Matas, J., De Vicuña, L.G., Castilla, M.:
‘Hierarchical Control of Droop-Controlled AC and DC Microgrids—
A General Approach Toward Standardization’ IEEE Trans. Ind. Electron.,
2011, 58, (1), pp. 158–172.
[7] Peyghami, S., Mokhtari, H., Blaabjerg, F.: ‘Hierarchical Power Sharing
Control in DC Microgrids’, in Magdi S Mahmoud (Ed.): ‘Microgrid:
advanced control methods and renewable energy system integration’
(Oxford: Elsevier Science & Technology, 2017, first), pp. 63–100.
[8] Boroyevich, D., Cvetković, I., Dong, D., Burgos, R., Wang, F., Lee, F.:
‘Future electronic power distribution systems—A contemplative view’ Proc.
Int. Conf. Optim. Electr. Electron. Equipment, OPTIM, 2010, pp. 1369–1380.
[9] Loh, P.C., Li, D., Chai, Y.K., Blaabjerg, F.: ‘Autonomous Operation of ac–
dc Microgrids with Minimised Interlinking Energy Flow’ IET Power Elec-
tron., 2013, 6, (8), pp. 1650–1657.
[10] Gu, Y., Xiang, X., Li, W., He, X.: ‘Mode-Adaptive Decentralized Control
for Renewable DC Microgrid with Enhanced Reliability and Flexibility’
IEEE Trans. Power Electron., 2014, 29, (9), pp. 5072–5080.
[11] Peyghami, S., Mokhtari, H., Loh, P.C., Davari, P., Blaabjerg, F.: ‘Distributed
Primary and Secondary Power Sharing in a Droop-Controlled LVDC
Microgrid with Merged AC and DC Characteristics’ IEEE Trans. Smart
Grid, 2018, 9, (3), pp. 2284–2294.
[12] Peyghami, S., Mokhtari, H., Blaabjerg, F.: ‘Decentralized Load Sharing in a
Low-Voltage Direct Current Microgrid with an Adaptive Droop Approach
Based on a Superimposed Frequency’ IEEE J. Emerg. Sel. Top. Power
Electron., 2017, 5, (3), pp. 1205–1215.
[13] Kundur, P., Balu, N., Lauby, M.: ‘Power system stability and control’
(New York: McGraw-Hill, 1994).
[14] Guerrero, J.M., Hang, L., Uceda, J.: ‘Control of Distributed Uninterruptible
Power Supply Systems’ IEEE Trans. Ind. Electron., 2008, 55, (8), pp. 2845–
2859.
[15] Rocabert, J., Luna, A., Blaabjerg, F., Rodriguez, P.: ‘Control of Power
Converters in AC Microgrids’ IEEE Trans. Power Electron., 2012, 27, (11),
pp. 4734–4749.
[16] Lu, X., Guerrero, J.M., Sun, K., Vasquez, J.C.: ‘An Improved Droop Control
Method for DC Microgrids Based on Low Bandwidth Communication with
DC Bus Voltage Restoration and Enhanced Current Sharing Accuracy’ IEEE
Trans. Power Electron., 2014, 29, (4), pp. 1800–1812.
[17] Meng, L., Dragicevic, T., Vasquez, J.C., Guerrero, J.M.: ‘Tertiary and Sec-
ondary Control Levels for Efficiency Optimization and System Damping
in Droop Controlled DC–DC Converters’ IEEE Trans. Smart Grid, 2015,
6, (6), pp. 2615–2626.
[18] Meng, L., Dragicevic, T., Guerrero, J.M., Vasquez, J.C.: ‘Dynamic con-
sensus algorithm based distributed global efficiency optimization of a droop
controlled DC microgrid’ ENERGYCON 2014—IEEE Int. Energy Conf.,
2014, pp. 1276–1283.
[19] Shafiee, Q., Dragicevic, T., Vasquez, J.C., Guerrero, J.M.: ‘Hierarchical
Control for Multiple DC-Microgrids Clusters’ IEEE Trans. Energy Convers.,
2014, 29, (4), pp. 922–933.
[20] Moayedi, S., Davoudi, A.: ‘Distributed Tertiary Control of DC Microgrid
Clusters’ IEEE Trans. Power Electron., 2015, 31, (2), pp. 1717–1733.
[21] Wu, T.-F., Chen, Y.-K., Huang, Y.-H.: ‘3C Strategy for Inverters in Parallel
Operation Achieving an Equal Current Distribution’ IEEE Trans. Ind.
Electron., 2000, 47, (2), pp. 273–281.
[22] Ding, G., Gao, F., Zhang, S., Loh, P.C., Blaabjerg, F.: ‘Control of Hybrid
AC/DC Microgrid under Islanding Operational Conditions’ J. Mod. Power
Syst. Clean Energy, 2014, 2, (3), pp. 223–232.
[23] Hatziargyriou, N., Asano, H., Iravani, R., Marnay, C.: ‘Microgrids’ IEEE
Power Energy Mag., 2007, 5, (4), pp. 78–94.
[24] Lu, X., Guerrero, J.M., Sun, K., Vasquez, J.C., Teodorescu, R., Huang, L.:
‘Hierarchical Control of Parallel AC–DC Converter Interfaces for Hybrid
Microgrids’ IEEE Trans. Smart Grid, 2014, 5, (2), pp. 683–692.
[25] Peyghami-Akhuleh, S., Mokhtari, H., Loh, P.C., Blaabjerg, F.: ‘Distributed
secondary control in DC microgrids with low-bandwidth communication
link’ 2016 7th Power Electronics and Drive Systems Technologies Con-
ference (PEDSTC), IEEE, 2016, pp. 641–645.
Chapter 3
Stability analysis and stabilization
of DC microgrids
Alexis Kwasinski1
3.1 Dynamic characteristics of DC microgrids

Stability issues in DC microgrids can be characterized by dynamics associated to
time constants ranging various time scales. At longer time scales, sometimes ran-
ging from seconds to hours, stability concerns are related to the need of matching
generated and consumed powers. That is, statically, generated power needs to
match consumed power in loads and sources, whereas, dynamically, power gen-
eration sources response ramps should be able to follow load changes. Still, these
stability concerns associated to generation and demand matching, both statically
and dynamically, are fundamentally similar to same stability concerns observed in
AC power systems and microgrids. These stability issues due to electric power
generation and consumption mismatches tend to be more noticeable in microgrids
than in power due to the lower combined inertia of microsources and because
microsources maximum power ratings tend to be closer to individual loads power
consumption. The common approach to address these stability concerns is to add
energy storage [1] that can act as temporary sources by discharging when power
demand is higher than supply or temporarily as loads by charging when power
demand is lower than supply. Additionally, several studies have proposed control
approaches that simulate the presence of inertia in microgrids [2,3]. In the case of
DC microgrids, [4] discusses analogies between stored energy in electrical gen-
erators’ rotors of conventional AC grids and the role of stored energy in capacitors
of DC microgrids.
As indicated, stability issues associated to power generation and consumption
mismatches are similar in AC and DC microgrids. However, stability issues at
shorter time scales are more commonly observed in DC microgrids than in AC
systems because these issues originate in the presence of loads interfaced by power
electronic converters acting as instantaneous constant-power loads (CPLs), which
tend to be more prevalent in DC microgrids than on AC ones because in DC
microgrids almost all type of loads tend to be powered through a point-of-load
1
Department of Electrical and Computer Engineering, University of Pittsburgh, USA
(POL) converter. Hence, within the context of this book, it is more relevant to focus
the discussion of DC microgrids stability on the effect of CPLs and
suitable approaches for mitigating the stability issues that these loads introduce.
As Figure 3.1(a) exemplifies, instantaneous CPLs are commonly realized by
very efficient POL converters with fast controllers that regulate their output voltage
tightly. Since the output voltage of the POL converter remains sufficiently constant,
its output power with a resistive load will remain approximately constant, too.
Hence, its input power will also remain constant despite input voltage changes
because the POL converter is assumed to have a sufficiently high efficiency so
that the input power equals to the output power. Figure 3.1(b) shows a common
practical instantaneous CPL: a data center server. In the particular case of this
figure, the servers are powered in part by photovoltaic arrays through a 380 V DC
power distribution network.
Mathematically, ideal instantaneous CPLs are modeled by
PL
i ðt Þ ¼ (3.1)
v ðt Þ
where v(t) is the voltage at the CPL, i(t) is the current through the CPL and PL is the
power of the CPL. In practical applications, CPLs do not maintain indefinitely a
same value for PL, but although PL may change over time, changes in PL occur at
time scales much longer than the time constants associated to the dynamics of the
system under study. Additionally, real CPLs only show the behavior modeled by
(3.1) above a given threshold voltage. Below such voltage, protections that are part
of the CPL controller act in order to disconnect such load to prevent the current to
reach excessively high values. Nevertheless, in order to provide a systematic ana-
lysis of DC microgrids stability characteristics and control that is not dependent on
Point-of-load DC–DC
Constant DC
converter
output
voltage
DC Point-of-load converter +
input power stage VL R
voltage –
Fast output voltage regulator
Constant power
(a) (b)
Figure 3.1 DC load systems. (a) A point-of-load converter acting as a CPLs.

(b) An actual CPL: servers in a data center (Texas Advanced
Computing Center). The detail on the upper-right corner show the
servers power supplies acting as CPLs
Stability analysis and stabilization of DC microgrids 45
specific design choices, such as selection of CPLs low voltage threshold, for the
rest of this chapter, a voltage threshold is omitted and, thus, it is assumed that CPLs
behave as described by (3.1) irrespective of the value of v(t).
The next section of this chapter discusses the effect of CPLs on DC microgrids
stability characteristics followed by two sections describing approaches for miti-
gating the destabilizing effect of CPLs. These mitigating approaches are described
first considering passive methods followed by actively controlled strategies.
3.2 DC microgrids stability analysis
Stability challenges appear with CPLs when connecting at least two converters in a
cascade configuration, as shown in Figure 3.2. In a DC microgrid, one of these
converters would be a POL converter, whereas the other would be a converter
interfacing an electric power source with the power distribution grid powering the
POL converter. In Figure 3.2, the converter interfacing the power source is repre-
sented by a buck converter. However, the same general stability characteristics are
observed when using other type of converters to interface a power source with the
microgrid power distribution grid [5,6]. In order to initiate the discussion of DC
microgrids stability characteristics with CPLs, consider that the buck converter
interfacing a power source is controlled with a fixed duty cycle D and that it
operates in continuous conduction mode. Hence, this buck converter can be
mathematically represented by its average model given by
8
>
> di L
>L
< ¼ DE v C
dt
>
> dv C PL (3.2)
>
:C ¼ iL
dt v C
v C > e; i L 0

Power source converter Point-of-load (POL)

interface (PSCI) converter
L i Main
L bus
+ Point-of-load converter +
C +
Vo vC power stage VL R
E –
– –
Power
source
Controller Fast output voltage regulator
Constant power (PL)
Figure 3.2 A POL converter acting as a CPL to a PSCI located upstream

in a cascade configuration
where E is the power source voltage; v C is the power source converter interface
(PSCI) average capacitor voltage, which equals the DC bus average voltage; iL is
the PSCI average inductor current; PL is the CPL power; L is the PSCI inductance;
and C is the PSCI output capacitor capacitance. The equilibrium point for this
converter is then
0 1 0 1
DE VO
VC
¼ @ PL A ¼ @ P L A (3.3)
IL
DE VO
It is possible to observe that this equilibrium point is not stable. In order to prove
this characteristic, it is relatively simple to find that the characteristic equation for
the small signal model of the buck PSCI with a CPL is given by
PL 1
l2 lþ ¼0 (3.4)
CVC2 LC
where l are the unknown eigenvalues. Since the coefficient for the first-order term
is negative, the characteristic polynomial associated to the buck PSCI with a CPL
does not satisfy the Routh–Hurwitz criterion and, thus, the equilibrium point for a
system represented by (3.1) is not stable.
Although small signal models allows to determine the stability characteristics
of the equilibrium point in a relatively simple way, linearization analytic methods
omit important behavior characteristics of converters with CPLs. A nonlinear
analysis approach, such as the one in [5–7], shows that depending on the PSCI
capacitor voltage and inductor current initial conditions, it is possible to observe
that the PSCI shows two behaviors: for sufficiently large initial capacitor voltages,
the voltage and current waveforms eventually settle into an oscillatory behavior,
which in a state space plot represents a limit cycle behavior. Otherwise, the capa-
citor voltage drops and the inductor current increases to very high values. As [5–7]
also show, it is possible to find that the curve—a separatrix—in the state space that
acts as a boundary to both behaviors can be approximated to
PL Cv2C
iL ¼ ðE v C Þ (3.5)
vC LPL
A simulation based on the model represented by (3.1) can be used to exemplify the
two possible behaviors of a PSCI with a CPL. For example, consider a buck PSCI
with the following parameters: E ¼ 400 V, C ¼ 3 mF, L ¼ 0.15 mH, D ¼ 0.5 and
PL ¼ 50 kW. As Figure 3.3 shows, different choices for initial conditions result in
either the oscillatory behavior or the voltage collapse-high current conditions.
Evidently, both of these behaviors are undesirable in a DC microgrid in which the
PSCI output voltage is intended to be constant. Although the assumption for the
analysis is that the converter operates in continuous conduction mode, in [7], it was
shown that in most practical conditions, a more exact model, than (3.1) that con-
siders operation in discontinuous conduction mode, does not deviate significantly
from the already discussed observations. Additionally, as it was indicated above,
–
iL (A)
1,200
Separatrix Oscillatory limit
cycle behavior
1,000
Limit cycle
800
600
400 Equilibrium
point
Voltage
200 collapse –
behavior vC (V)
0
0 100 200 300 400 500
–200
Figure 3.3 Various trajectories from different conditions in state space

showing the two possible behaviors of a PSCI with a CPL
shown in Figure 3.2
–
iL [A]
1,200 1,200
1,000 1,000
800 – 800
iL [A]
600 600
400 400
v–C [V]
200 200
t (s) v–C [V]
0 0
0 0.02 0.04 0.06 0.08 0.1 0 100 200 300 400 500
–200 –200
(a) (b)
Figure 3.4 Time domain (a) and state-space response (b) of a buck PSCI
attempting to regulate its output voltage that is supplied to a CPL
although the analysis uses a buck converter as an example for the discussion of the
observations, as [5,6] shows, the same general behavior described for the buck
converter is also observed in other DC–DC converter topologies.
One common initial thought to address the stability issues introduced by CPLs
and to achieve a constant DC voltage on the microgrid main bus is to attempt to
regulate the PSCI output voltage with a PI controller. However, as Figure 3.4 shows
with the same PSCI used in Figure 3.3 but now with a PI controller with iL(t ¼ 0) ¼
1 A, vC (t ¼ 0) ¼ 150 V, ki ¼ 10, kp ¼ 0.1 that attempts to regulate the output vol-
tage at 200 V, the oscillatory behavior still persists exemplifying the fact that a
common PI controller, that could achieve a stable equilibrium point with a PSCI
powering a resistive load, may not achieve a stable equilibrium point when the load
is an equivalent CPL. Hence, the next sections will discuss various strategies to
mitigate the stabilizing effect of CPLs and to achieve a constant regulated DC
voltage in a microgrid power distribution grid.
3.3 Passive approaches for stabilization of DC microgrids
Passive approaches for stabilization of DC microgrids with CPLs can be identified

by a necessary condition for the stability derived from (3.5). As it is explained in
[7], a necessary but not sufficient condition for stability is that the trajectory of any
PSCI stays on the right side of the separatrix given by (3.5). That is,
PL Cv2C
iL > ðE v C Þ (3.6)
vC LPL
Hence, the stability characteristics of a DC microgrid with CPLs improves when
C increases, L decreases or PL decreases. Still, (3.6) only provides an initial
approximation of suitable passive approaches for stabilizing DC microgrids.
In order to further determine passive stabilization approaches for DC micro-
grids, consider the expanded model of a buck PSCI
8
>
> diL
<L ¼ qðtÞðE RSW iL Þ ð1 qðtÞÞðVD þ iL RD Þ iL RL vC
dt
>
> C dvC ¼ iL PL vC (3.7)
:
dt v C RO
with iL 0; vC > e
in which q(t) is the main switch switching function, RSW is the conduction resis-
tance of the main switch, RD is the conduction resistance of the diode, VD is the
diode’s forward voltage drop, RL is the inductor’s series resistance and RO is an
output resistance acting as an additional load in parallel with the CPL. The new
characteristic equation of the linearized average model is

Ri PL 1 Ri 1 PL 1
l2 þ þ l þ þ ¼0 (3.8)
L CVO2 RO C L RO C CVO2 LC
where Ri is the sum of RSWD, RL and RD(1D). Based on the Routh–Hurwitz
criterion, the two necessary and sufficient conditions for a stable equilibrium
point are

2 C 1
P L < VO Ri þ (3.9)
L RO
and

1 1
PL < VO2 þ (3.10)
Ri R O
Although both (3.9) and (3.10) need to be satisfied in order to ensure a

stable equilibrium point, from a practical perspective in most applications (3.10)
is satisfied if (3.9) is satisfied because Ri is usually less than 1 W and C is within the
same order or an order of magnitude higher than L.
From (3.9), it is possible to identify several passive approaches for improving
the stability characteristics of DC microgrids. One of such approaches is to reduce
L. However, this is not a practical approach because the inductance is usually
determined by other design needs and constraints. Another approach is to add an
output resistance in parallel to the CPL. If it is assumed in (3.9) that Ri ¼ 0, then the
maximum output resistance that needs to be added in parallel with the CPL equals
VO2 =PL . As Figure 3.5 exemplifies, adding an output resistance of 0.5 W to the PSCI
simulated in Figure 3.3 achieves a stable equilibrium point, but if such resistor is
added for the only purpose to achieve a stable equilibrium point, then the overall
power efficiency would be 50% at best (power efficiency in the example in
Figure 3.5 equals 0.38 because only 38% of the PSCI output power is dissipated
in the CPL).
Another approach to achieve a stable equilibrium point suggested by (3.9) is to
increase the PSCI’s output capacitance. However, since Ri tends to be in the order
of milli-Ohms or tens of milli-Ohms, the minimum capacitance needed in order to
achieve stable equilibrium points are relatively high, leading to large and bulky
designs. Such result is exemplified in Figure 3.6 in which the PSCI simulated in
Figure 3.3 was augmented with an RSW of 50 mW and with a new output capaci-
tance of 50 mF. One practical alternative in DC microgrids to increase the PSCI’s
output capacitance is to add batteries directly connected to the microgrid power
distribution network. However, such solution may lead to some control limitations,
particularly if droop control is sought.
As it has been discussed, although passive approaches allow achieving a
stable operating point, they tend to present important practical limitations. Thus,
the next section discusses alternative strategies to achieve the same goal of a
stable equilibrium point though controller action.
–
iL [A]
1,400 1,400
1,200 1,200
1,000 1,000
800 – 800
iL [A]
600 600
400 400
v–C [V]
200 200
t [s] v–C [V]
0 0
0 0.02 0.04 0.06 0.08 0.1 0 50 100 150 200 250 300 350
(a) (b)
Figure 3.5 Time domain (a) and state-space behavior (b) of a buck PSCI
powering a CPL and a resistive load
–
iL [A]
1,000 1,200
800 1,000
800
600
600
400 –
iL [A] 400
200
v–C [V] 200
v–C [V]
t (s)
0 0
0 0.02 0.04 0.06 0.08 0.1 0 100 200 300 400 500
–200 –200
(a) (b)
Figure 3.6 Time domain (a) and state-space behavior (b) of a buck PSCI with
increased output capacitance and powering a CPL
3.4 Control strategies for stable DC microgrids operation

Control-based approaches for achieving stable operation points in DC microgrids
with CPLs can be divided into time-based approaches and geometric control stra-
tegies. Arguably, the most commonly discussed time-based approach involves the
insertion of virtual resistances to PSCIs through controller action. Since these
inserted resistances are not real, their insertion do not affect the converter power
efficiency. As it is shown in [5–7], such approach may lead in the case of buck
PSCIs like that represented by (3.2) to a linear proportional-differential (PD) con-
troller of the form
Ri1 C de Ri1
d¼ þ eþD (3.11)
E dt Ri2 E
where e is the error signal resulting from the difference between the measured
output voltage and the desired output voltage, Ri1 is a virtual resistance in series
with the PSCI’s inductor and Ri2 is a virtual resistance in parallel to the PSCI’s
output capacitor. As indicated in [5], asymptotic stability is observed only if
Ri1 PL
> 2 (3.12)
L VO C
Still, global asymptotic stability is not observed in practical applications due to the
fact that the duty cycle is constrained to values between 0 and 1. An example of the
result of using a PD controller to achieve a stable equilibrium point is shown in
Figure 3.7. In this figure, a buck PSCI was simulated with the same parameters
of the converter used in Figure 3.3 with initial conditions iL(t ¼ 0) ¼ 1 A and
vC (t ¼ 0) ¼ 150 V, and a proportional gain of 0.5 and differential gain of 1 103,
implying a value of Ri1 of 133.33 W and Ri2 of 0.66 W.
The work in [5,6,8] shows that PD controllers also achieve a stable equilibrium
point for other types of PSCIs, such as boost and buck-boost converters. However,
although a properly designed PD controller may achieve a stable operation point,
–
iL [A]
500 1,200
400 1,000
– 800
300 iL [A]
600
200
v–C [V] 400
100 200
t [s] v–C [V]
0 0
0 0.02 0.04 0.06 0.08 0.1 0 100 200 300 400 500
–100 –200
(a) (b)
Figure 3.7 Time domain (a) and state-space behavior (b) of a buck PSCI
powering a CPL and controlled with a PD controller
VC [V]
VO,NL
VC = f(IL1)
VC = f(IL2)
IL [A]
IL2,max IL1,max
Figure 3.8 Droop lines for two parallel connected PSCIs
it does not provide output voltage regulation. Nevertheless, as shown in [5], a

stable equilibrium point with output voltage regulation can be achieved by adding
an integral term to the PD controller in (3.11) provided that the integral gain is low
enough so that the dynamics of the integral term are slower than those associated to
the PD terms.
Dynamics introduced by droop controllers create a control action somewhat
analogous to that associated to the aforementioned PD controllers. The main
concept of conventional droop controllers, at what it is commonly referred as
the primary controller, is to insert a virtual resistance at the PSCI output via con-
troller action with the goal of achieving autonomous control goals, such as load
sharing among PSCIs connected in parallel. As a result, the output voltage vs.
current relationship follows a line such as the one shown in Figure 3.8 for two
buck PSCIs connected in parallel. Still, droop relationships show static operational
relationships.
Buck PSCI #1 Point-of-load converter

L1 i Main
L1 bus
+ +
C1 VO Point-of-load converter
d1 power stage VL R
E1 vc
– –
Controller Fast output voltage regulator
Buck PSCI #2 PL
L2 i
L2
C2
d2
E2
Controller
Figure 3.9 Two parallel-connected buck PSCIs sharing a CPL with a droop
controller in each PSCI
In order to evaluate the dynamic effect of adding virtual droop resistances to

two PSCIs with a CPL, consider the following dynamic average model of the
resulting system [9]
8
>
> di L1
>
> L1 ¼ d1 E1 v C
>
> dt
>
<
di L2
> L2 ¼ d2 E2 v C (3.13)
>
> dt
>
>
>
> dv P
: ðC1 þ C2 Þ C ¼ i L1 þ i L2 L
dt vC
in which the corresponding variables and parameters are depicted in Figure 3.9.
As also shown in [9], the duty cycles d1 and d2 are controlled so that if VO,NL is the
no-load output voltage, then
V0;NL Rd;1 i L1
d1 ¼ (3.14)
E1
and
V0;NL Rd;2 i L2
d2 ¼ (3.15)
E2
imply the insertion of virtual resistances Rd,1 and Rd,2 in series with each of the
inductors L1 and L2, respectively. Since the average current across these compo-
nents equal the output current of the PSCIs, and the steady state average output
voltage of a buck converter equals the product of the input voltage and the duty
cycle, (3.14) and (3.15) imply a static linear droop relationship between the output
voltage and output current. Dynamically, [9] shows that the addition of a droop
resistance allows limit cycle oscillations to be damped provided that the following
conditions obtained by studying the characteristic equation of the linearized system
are met:
Rd;1 Rd;2 PL
þ > (3.16)
L1 L2 ðC1 þ C2 ÞVO2
L1 þ L2 þ Rd;1 Rd;2 ðC1 þ C2 Þ PL
> 2 (3.17)
Rd;2 L1 þ Rd;1 L2 VO
V2
Rd;1 jjRd;2 < O (3.18)
PL
" #
P2L Rd;1 Rd;2 Rd;1 Rd;2 1 Rd;1 Rd;2
þ þ þ þ 2
VO4 ðC1 þ C2 Þ2 L1 L2 L1 L2 C1 þ C2 L21 L2
" 2 # " #
PL Rd;1 Rd;2 PL 1 1
2 þ þ >0
VO ðC1 þ C2 Þ L1 L2 VO2 ðC1 þ C2 Þ2 L1 L2
(3.19)
where
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
VO;NL þ 2
VO;NL 4PL Rd;1 jjRd;2
VO ¼
2
The effectiveness of the droop resistances insertion in order to damp the effect of
the CPL was verified experimentally with a reduced scale microgrid in [9] by using
a CPL of 100 W and two parallel buck PSCIs with E1 ¼ 35 V, E2 ¼ 30 V, C1 ¼
C2 ¼ 1,000 mF, L1 ¼ 640 mH, L2 ¼ 630 mH and VO,NL ¼ 25.7 V. Initially, the PSCIs
were controlled in open loop with a constant duty cycle equal to 0.734 for the buck
converter #1 and equal to 0.857. As Figure 3.10 shows, the result of such operation
is limit cycle oscillations for all state variables. Then, a primary droop controller as
the one represented by (3.15) and (3.14) with Rd1 ¼ 2.075 W and Rd2 ¼ 1.709 W is
enabled causing, as Figure 3.10 shows, damping of the oscillations and verifying
reaching a stable equilibrium point.
Although the insertion of virtual droop resistance through controller action
damps the limit cycle oscillations, they also naturally cause voltage deviations,
which in some cases may be undesirable. The conventional solution to this issue is
to add a so-called secondary droop control that regulates the main bus voltage by
shifting vertically the static droop lines in Figure 3.8. Thus, as indicated in [9],
iL1
iL2
vC
Open loop Primary
control
Figure 3.10 Experimental results showing the transition from open loop
operation to primary droop control for two buck PSCIs with a CPL.
Image courtesy of Mahesh Srinivasan
a secondary control regulating action can be achieved by adding an integral term

with a gain ki in (3.15) and (3.14). The resulting control action is then given by
ð

V0;NL Rd;1 i L1 þ ki VO;NL v C dt
d1 ¼ (3.20)
E1
and
ð

V0;NL Rd;2 i L2 þ ki VO;NL v C dt
d2 ¼ (3.21)
E2
with this controller, the new conditions for a stable equilibrium point become [9]
Rd;1 Rd;2 PL
þ > (3.22)
L1 L2 ðC1 þ C2 ÞVO;NL
2
!
1 ki Rd;1 ki Rd;2 Rd;1 Rd;2 PL
þ þ þ >0 (3.23)
C1 þ C2 L2 L1 L2 L1 L1 L2 L1 L2 VO;NL 2

ki Rd;1 þ Rd;2
>0 (3.24)
L1 L2 ðC1 þ C2 Þ
" #
P2L Rd;1 Rd;2 Rd;1 Rd;2 1 Rd;1 Rd;2
þ þ þ þ
4
VO;NL ðC 1 þ C 2 Þ2 L1 L2 L1 L2 C1 þ C2 L21 L22
" 2 #
PL Rd;1 Rd;2 1 1 1
2 þ þ >0 (3.25)
VO;NL ðC1 þ C2 Þ L1 L2 ðC1 þ C2 Þ L1 L2
iL1
iL2
vC
Added
Primary secondary
control control
Figure 3.11 Experimental results showing the addition of regulation through a

secondary droop controller for two buck PSCIs with a CPL. Image
courtesy of Mahesh Srinivasan
iL1
iL2
vC
P = 100 W P = 120 W
Figure 3.12 Experimental results showing droop control with load regulation for
two buck PSCIs with a CPL. Image courtesy of Mahesh Srinivasan
Output voltage regulation while still achieving a stable operating point was also
verified experimentally in [9]. As Figure 3.11 shows, the addition of integral reg-
ulators with ki ¼ 4.5 to the system used to produce Figure 3.10 compensate for the
primary droop controller voltage deviations and restores the main bus voltage to its
nominal value of 25.7 V. As it can be verified by comparing the current traces
before and after the secondary controller is enabled, the currents share the load
current maintaining a constant ratio of 1.22. As Figures 3.12 and 3.13 exemplify,
the addition of the integral stage of the controller allows to regulate the main bus
iL1
iL2
vC
E1 = 35 V E1 = 30 V
Figure 3.13 Experimental results showing droop control with line regulation for
two buck PSCIs with a CPL. Image courtesy of Mahesh Srinivasan
voltage at a constant reference value even when the load’s power or one or more
input voltages change.
Although the discussion in this section still focused on buck PSCIs, virtual
droop resistances applied to other DC–DC converters will damp the limit cycle
oscillations caused by CPLs. In addition to the discussion in [9], the validity of this
conclusion and the possibility to achieve voltage regulation with an integral sec-
ondary droop controller was also described in [10,11].
Although the addition of a virtual resistance through controller action makes
the operating point stable, such controllers tend to have relatively slow dynamic
response. Moreover, in the case of the PD controller, the differential term intro-
duces noise susceptibility to the PSCIs. Geometric controllers provide another
approach to mitigate the destabilizing effect of CPLs and to achieve a
stable operating point but with a relatively fast dynamic response and without
introducing noise susceptibility. In geometric controllers, the state of the main
switch changes when the state space trajectory defined by the PSCI’s state variables
crosses a boundary. In practice, the boundary (also called switching surface) is
replaced by a hysteresis band containing such boundary in order to avoid chattering
when the trajectory tends to be confined along the switching surface. In general, it
is possible to distinguish three behaviors when a trajectory intersects a switching
surface [12]:
● Refractive behavior when the trajectories on one side of the boundary are
incident, but they are exiting on the other side of the boundary.
● Rejective behavior when trajectories on either side of the boundary are exiting.
● Reflective behavior when trajectories on both sides of the boundary are inci-
dent. Once a trajectory reaches a switching surface in a reflective behavior
region, it remains confined to the boundary. Hence, a reflective behavior could
iL [A]
Linear
boundary
(k < 0)
Refractive area
Reflective
stable area
Reflective
unstable area
PL = constant
vC [V]
Equilibrium point
Figure 3.14 Boundary control behavior areas for a buck PSCI with a CPL
be associated to a stable equilibrium point when the confined trajectory tends

along the switching surface to such an equilibrium point. However, a reflective
behavior region could be associated to an unstable operating point when the
confined trajectory moves away from such point.
As it was identified in [13] and it is shown with Figure 3.14, both refractive and
reflective (stable and unstable) behaviors are observed when a linear boundary of
the form
PL
i L ¼ k ðv C V O Þ þ (3.26)
VO
is used for a buck converter powering a CPL. In (3.26), k is a design parameter that
mathematically represents the slope of the line acting as the switching boundary. In
order to achieve a stable operating point by having the trajectory intersecting the
switching boundary at a stable reflective region instead of an unstable reflective
region k needs to be negative [13]. Such need for a negative value of k in order to
achieve a stable operating point is also observed for other converters powering
CPLs, such as boost and buck-boost converters [6,14]. Also, [6,13,14] show that
although for some of these converters, a linear boundary naturally achieves line or
load regulation, it is also possible to simply enhance the controller so it can auto-
matically displace the switching boundary to achieve line or load regulation. Fig-
ure 3.15 shows an example of the trajectories observed both in time domain and in
the state space when a buck PSCI with L ¼ 500 mH, C ¼ 1 mF, E ¼ 22.2 V, VO ¼
18 V and a CPL of 108 W transitions from open loop operation to a linear boundary
controller with k ¼ 1.
iL [A] Linear boundary

vC [V ]
(k = –1) Closed-loop
40 18 trajectory
16
20 Open-loop
14
trajectory
12
20 40 60 80 t (ms) 10
iL [A] Closed loop 8
Open loop
20 6
PL = constant
4
10 2 vC [V ]
0
0 6 12 18 24 30
t (ms) –2
(a) 20 40 60 80 (b) Equilibrium point
Figure 3.15 Transition from open-loop to a closed-loop boundary control for a

buck PSCI with a CPL. (a) Time domain and (b) state-space
representation
L iL
+
+ C vC PL
|vS|
–
–
Figure 3.16 Simplified model of a rectifier with a CPL in which |vS| represents a
rectified unfiltered sinusoidal voltage signal
3.5 Operation of rectifiers with instantaneous constant

power loads
Although this chapter has focused primarily on the operation of DC–DC converters
with CPLs, in DC microgrids, it is possible to have rectifiers directly interfacing a
source and the microgrid main bus. Still, it is expected that in most applications,
there would be a DC–DC converter at the output of the rectifier in order to improve
regulation and filter design. However, the case of rectifiers directly interfacing an
electric power source to a CPL was explored in [14]. In [14], it was shown that
without an adequate mitigating approach, CPLs may cause either an oscillatory,
limit cycle, behavior or a voltage collapse and high current condition similar to
those discussed for a buck PSCI with a CPL. Since [14] explains that a rectifier,
such as the one with a simplified model in Figure 3.16, with a second-order low-
pass output filter behaves analogously to a buck converter with a CPL, then a
necessary condition for the operating point to be an unstable focus that may show
an oscillatory limit cycle behavior if
rffiffiffiffi
PL C
2
<2 (3.27)
VS;O L
vC(t) [V]
12
4
iL(t) [A]
t (ms)
0
0 20 40 60 80
Figure 3.17 Example of limit cycle period-1 behavior of a rectifier with a CPL
where C and L are the rectifier output filter capacitor and inductor, respectively,
and VS,O is an equivalent constant input voltage that equals |vs| when |vs| ¼ vC [14].
Condition (3.27) is necessary but not sufficient to observe a limit cycle behavior
considering the rectifier output capacitor voltage and inductor current as state
variables. In order to observe a limit cycle behavior, in addition to satisfy (3.27),
the initial capacitor voltage needs to be high enough. An approximate value for
such minimal initial capacitor voltage is given by [14]
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
" #ffi
u
u 1 P2 L1=4 2 1=4
P L P
v C ð0 Þ > t
L
L
2fVin L þ (3.28)
2fVin 4C 4C C
where Vin is the peak AC input voltage to the rectifier.

Both conditions (3.27) and (3.28) are necessary conditions for observing limit
cycle oscillations at the rectifier output which are typically presented as the
waveforms shown in Figure 3.17. Such behavior is called period-1 in [14]. How-
ever, since (3.27) and (3.28) are not sufficient conditions, it is still possible to
observe the alternative rectifier behavior: voltage collapse with increasing inductor
currents. As indicated in [14], closer to the limit between period-1 to voltage col-
lapse behavior it is possible to observe another oscillatory behavior called period-2
in [14], and it is shown in Figure 3.18.
Although observing the limit cycle oscillatory behavior tends to be better than
the voltage collapse behavior, the oscillations still cause output voltage variations
that can be approximated to
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
PL
DvC ¼ Vin Vin2 (3.29)
Cf
vC (t) [V]
12
iL(t) [A]
4
t (ms)
0
0 20 40 60 80
Figure 3.18 Example of limit cycle period-2 behavior of a rectifier with a CPL
This voltage variation is typically higher than conventional voltage ripple in rec-
tifiers with an equivalent resistive load [14].
3.6 Summary
This chapter has discussed the effects of CPLs in DC-microgrids. Such loads are
commonly observed in DC microgrids because of the extensive use of power
electronic circuits to interface loads. As it is explained, due to their nonlinear nat-
ure, CPL creates a complex dynamic system that present two possible behaviors
with an unstable equilibrium point unless PSCIs are properly controlled. In this
chapter, both time domain and geometric control approaches that achieve
stable equilibrium points in the presence of CPLs are explained. Passive approa-
ches to compensate the destabilizing effect of CPLs are also discussed, but these
approaches often lead to costly and bulky designs. Complex dynamics are also
observed in rectifiers with CPLs, which often present excessively large output
voltage variations.
References
[1] A. Toliyat and A. Kwasinski, ‘‘Energy Storage Sizing for Effective Primary
and Secondary Control of Low-Inertia Microgrids,’’ in Proc. 2015 IEEE
Power Electronics for Distributed Generation Conference, pp. 1–7, Aachen,
Germany, June 2015.
[2] N. Sony, S. Doolla and M. C. Chandorkar, ‘‘Improvement of Transient
Response in Microgrids Using Virtual Inertia,’’ IEEE Transactions on Power
Delivery, vol. 28, no. 3, pp. 1830–1838, July 2013.
[3] X. Zhu, J. Cai, Q. Yan, J. Chen and X. Wang, ‘‘Virtual Inertia Control of
Wind-Battery-Based Islanded DC Micro-Grid,’’ in Proc. International
Conference on Renewable Power Generation (RPG 2015), pp. 1–7, October
2015.
[4] T. L. Vandoorn, B. Meersman, J. D. M. De Kooning and L. Vandevelde,
‘‘Analogy Between Conventional Grid Control and Islanded Microgrid
Control Based on a Global DC-Link Voltage Droop,’’ IEEE Transactions on
Power Delivery, vol. 27, no. 3, pp. 1405–1414, July 2012.
[5] A. Kwasinski, W. Weaver, and R. Balog, ‘‘Micro-Grids in Local Area Power
and Energy Systems,’’ Cambridge University Press, Cambridge, U.K., 2016.
[6] A. Kwasinski and C. N. Onwuchekwa, ‘‘Dynamic Behavior and Stabilization
of DC Microgrids with Instantaneous Constant-Power Loads,’’ IEEE
Transactions on Power Electronics, vol. 26, no. 3, pp. 822–834, March 2011.
[7] A. Kwasinski and P. T. Krein, ‘‘Passivity-Based Control of Buck Converters
with Constant-Power Loads,’’ in Proc. Power Electronics Specialist Con-
ference 2007 (PESC), pp. 259–265, Orlando, FL, USA, June 17–21, 2007.
[8] A. Kwasinski and P. T. Krein, ‘‘Stabilization of Constant Power Loads in
DC–DC Converters Using Passivity-Based Control,’’ in Proc. 2007 Inter-
national Telecommunications Energy Conference (INTELEC), pp. 867–874,
Rome, Italy, September 30–October 4, 2007.
[9] M. Srinivasan, ‘‘Hierarchical Control of DC Microgrids with Constant
Power Loads,’’ Ph.D. Dissertation, The University of Texas at Austin,
Austin, Texas, USA, 2017.
[10] M. Srinivasan and A. Kwasinski, ‘‘Decentralized Control of a Vehicular
Microgrid with Constant Power Loads,’’ in Proc. 2014 IEEE International
Electric Vehicle Conference, pp. 1–8, Florence, Italy, December 2014.
[11] M. Srinivasan and A. Kwasinski, ‘‘A Linear Damping Scheme for Higher
Order dc–dc Converters Supplying Constant-Power Loads in a dc Micro-
grid,’’ in Proc. 19th European Conference on Power Electronics and Appli-
cations (EPE) – Energy Conversion Congress and Exposition (ECCE)
Europe, pp. 1–10, Warsaw, Poland, September 2017.
[12] P. T. Krein, ‘‘Elements of Power Electronics,’’ Oxford University Press,
New York, NY, 1998.
[13] C. N. Onwuchekwa and A. Kwasinski, ‘‘Analysis of Boundary Control for
Buck Converters with Instantaneous Constant-Power Loads,’’ IEEE Trans-
actions on Power Electronics, vol. 25, no. 8, pp. 2018–2032, August 2010.
[14] C. N. Onwuchekwa and A. Kwasinski, ‘‘Dynamic Behavior of Single-Phase
Full-Wave Uncontrolled Rectifiers with Instantaneous Constant-Power
Loads,’’ in Proc. 2011 IEEE ECCE, pp. 3472–3479, Phoenix, AZ, USA,
September 17–22, 2011.
Chapter 4
Coordinated protection of DC microgrids
Jae-Do Park1, Md Habib Ullah1,
and Bhanu Babaiahgari1
4.1 Introduction
In the 2000s, the development of renewable energy resources and the advance of
power electronics technology, as well as the issues of fossil-fuel depletion and their
impact on environment, have driven extensive research on distributed power sys-
tems. The energy policy of many governments in the world competitively increases
the requirement of the penetration of renewable energy resources and distributed
generation. DC power systems have received renewed attention as a solution for the
distributed generation to replace or augment AC systems in various applications in
transmission and distribution systems. The amount of equipment that operates on
DC power is increasing, including photovoltaic (PV) panels, batteries, fuel cells,
and even AC motors driven by adjustable speed drives. Improved power electronics
converters allow DC power to be converted easily and efficiently. Unlike AC
systems, multiple sources can readily operate attached to a single DC bus because
they do not need to be synchronized. The system efficiency and reliability can be
improved with DC power delivery, because the DC-to-AC power conversion layer
can be eliminated from the back-to-back conversion scheme. Moreover, DC sys-
tems have an advantage in cable size because they do not suffer from skin effect,
and there is no need for reactive power. Multiterminal DC (MTDC) systems have
been investigated for high-voltage DC (HVDC) transmission systems and low-
voltage DC (LVDC) distribution systems as well. DC power systems have been
investigated for various microgrid systems, such as naval ships, traction systems,
data centers, buildings, aircrafts, and microgrids.
However, compared to the AC power systems that have been in operation as the
standard power system for the last two centuries after the ‘‘war of current’’ in the
early 1800s, the DC power systems have much shorter operation history. Standards
and accumulated experience are substantially insufficient. Moreover, it has been
applied to the systems in confined areas in many cases. Lack of knowledge base is
true as well when it comes to the system protection, especially for large systems.
1
Department of Electrical Engineering, University of Colorado Denver, USA
Although the AC system protection technology is very mature, most of their devices
cannot be directly applied in DC systems. Typically, DC switchgear is more
expensive and may not always be available for certain systems.
The protection of electrical power systems is fundamentally essential for safe
and stable operation; it prevents personnel injury and minimizes the service inter-
ruption and system damage from equipment failure, human error, and disastrous
natural events. And this is one of the biggest challenges of DC power systems,
because of the nonalternating nature and fast dynamics of DC power. Furthermore,
although there are various devices that monitor the power system, detect abnorm-
alities, and perform automatic actions to protect the system and sustain the opera-
tion, development of a well-coordinated protection scheme for DC power system is
not a trivial task. As DC power system design is not quite a standardized procedure
yet, its protection system is developed on the case-by-case basis and designer’s
empirical techniques and methods would play an important role in design.
Typically, a DC microgrid has multiple terminals connected to the bus, to which
individual systems such as load, source, and bidirectional devices can interface
through power converters, such as voltage-source converters (VSCs) and line-
commutated converters (LCCs). Thyristors in LCCs have high-voltage and power
capability, but lower control performance compared to IGBTs in VSCs. LCCs have
been used for simpler PtP configurations. Recently, multimodular converter systems
have been widely investigated for increased voltage and power capacity of DC bus
systems. One of the challenges associated with the protection of VSC-based DC
microgrid systems is that the fault current must be detected and extinguished quickly
as the fault current withstand rating of typical VSCs is only twice the converter full-
load current. Unlike AC systems, the typical sources in DC microgrid interfaced
through VSCs do not have enough inertia to supply fault current while maintaining
bus voltage. And the VSCs will trip and disconnect themselves from the bus when
the bus voltage drops below a certain threshold. These make it difficult to locate the
fault because the transient that contains fault circuit information might not be
available to protection devices. Recently, semiconductor-based and hybrid circuit
breakers (CBs) and fault current limiters have been investigated, and they have been
used for fast DC current interruption and overcurrent limitation. Typically, CBs were
proposed for converter and battery protection, while fuses and molded-case CBs
(MCCBs) were used for feeder protection [1].
As well as the detection and interruption where traditional differential and
overcurrent relaying techniques and main/back-up scheme can be applied, location
and isolation of the fault in the microgrid are important tasks of a protection system
for fast recovery and impact minimization, which requires cooperation between
protective devices in the system. Although the line impedance method and travel-
ing wave method have been adopted as industry standard for AC systems [2], it is
difficult to directly apply to DC systems due to the inherent absence of frequency
and phase parameters and much faster current dynamics.
In this chapter, nature of the faults in DC microgrid due to lack of inductance
in the lines, typical DC power system configurations, coordinated protection
schemes, and their applications for DC power systems will be reviewed.
Coordinated protection of DC microgrids 65
4.2 Faults in DC power systems

4.2.1 Fault types and behavior
A line-to-ground fault is said to have occurred in a line segment when a path
between either positive or negative pole and ground is created, and it is the most
common type of fault. A line-to-line fault is defined as a path between positive and
negative pole created in a line segment. A typical line-to-ground and line-to-line
fault in a bipolar DC bus are shown in Figure 4.1.
If a short-circuit fault occurs in a DC bus, a large current flows in the line
accompanied with terminal voltage drop. Relays will send trip signals to CBs and
VSCs. Since the DC link voltage has dropped to a low level even to zero, the
terminal capacitors will discharge through the fault and the fault current increases
from AC main as well. This overcurrent condition might damage the reverse diodes
and possibly the entire system, if the fault is not isolated quickly. The severity of a
fault depends on the fault impedance and system configuration.
DC system grounding plays a critical role from the protection perspective. For
a system where both AC and DC side are ungrounded, a line-to-ground fault does
not create a current circulating in the ground, so the system still can operate;
however, the fault has to be cleared before another line-to-ground fault occurs to
generate a short circuit between lines. If the grounding points in a DC network
include the neutral-ground link of the AC side or the DC-link midpoint, a line-to-
ground fault in a transmission system shown in Figure 4.2 will form a ground loop
with the grounding points and the AC main. In this case, the VSC cannot block the
current flowing in from the AC side even if all the IGBTs are open, because the
current bypasses IGBTs in VSC and flows through the antiparallel diodes as it does
in a bridge rectifier. The fault current will flow from the bus capacitors that dis-
charge through the fault path, making the fault current high-peak and fast-rising. In
a system with positive or negative pole solidly grounded, a line-to-ground fault on
(a) (b)
Figure 4.1 Types of faults in typical bipolar DC bus: (a) line-to-ground fault and
(b) line-to-line fault
CDC
CDC
Figure 4.2 Transmission line-to-ground fault fed by AC main and bus

capacitance. The AC source feeds the fault current through
antiparallel diodes regardless of the VSC switch status
If 1
R L if
If 2
Es + C
–
t
(a) (b)
Figure 4.3 (a) Line-to-line fault equivalent circuit model. The fault current
consists of two current components from the source and bus
capacitors, respectively. (b) Typical fault current waveform.
The transient can be seen at the peak of the fault current
the grounded pole will not have an impact on the system, but a fault on the other
pole will cause a line-to-line short circuit through ground.
4.2.2 Fault current analysis

To explain further about the fault currents mathematically, a simple equivalent cir-
cuit model is considered and a line-to-line fault is assumed as shown in Figure 4.3(a).
The current feeding the fault ifault consists of two current components if 1 and if 2 from
the source and bus capacitors, respectively. A mathematical expression for the fault
current when the equivalent resistance in the fault circuit is small enough to cause an
pffiffiffiffiffiffiffiffiffi
oscillation (R < 2 L=C ) can be given as
qffiffiffiffiffiffiffiffiffiffiffiffiffi
Es Es
Ifault ðtÞ ¼ 1 eðR=LÞt þ pffiffiffiffiffiffiffiffiffiffiffiffiffi eat sin wo 1 z2 t (4.1)
R wo 1 z2 L
where Es is the line voltage, R, L, and C is the equivalent resistance, inductance, and
capacitance in the fault path, respectively. The equivalent impedance of the fault path,
pffiffiffiffiffiffiffi
including fault and line impedance, determines the natural frequency wo ¼ 1= LC
and damping factor of the fault current z ¼ a=wo , where a ¼ R=ð2LÞ [3]. A typical
waveform of fault current with this model is shown in Figure 4.3(b), and the transient
can be seen at the peak of the fault current. However, in case the fault resistance or
bus capacitance is large enough, such as the line-to-ground fault where the ground
resistance is considerably high, the fault current can be modeled as a first-order
damping thus having no oscillation.
It is very important to study short-circuit currents in a DC power system and
how it can be calculated. It has to be noted here that the load current is not taken
into consideration while calculating short-circuit current. A standard approximation
function of short-circuit fault current proposed by International Electrotechnical
Commission (IEC) [4] is shown in Figure 4.4, and the equations governing the
function are given by
1 et=t1
i1 ðtÞ ¼ ip ; 0 t tp (4.2)
1 etp =t1
where ip is peak short-circuit current, tp is the peak time, Ik is the quasi-steady-state
short-circuit current, t1 is the rise time constant, t2 is the decay time constant, and
Tk is the short-circuit duration.
i2 ðtÞ ¼ Ik þ ðip Ik Þeððttp Þ=t2 Þ (4.3)

The two currents i1 ðtÞ and i2 ðtÞ determine the maximum short-circuit current that
governs the rating of electrical equipment and the minimum short-circuit current
that can be taken as the basis for fuse and protection device ratings, respectively.
It is assumed that the quasi-steady-state current Ik is the value of the short-circuit
current 1 s after the beginning of short circuit.
i
τ1
ip
i2(t)
Ik
τ2
i1(t)
0 tP Tk t
Figure 4.4 Standard approximation of a short-circuit fault current as a function

of time [4], where ip is peak short-circuit current, tp is peak time, Ik is
quasi-steady-state short-circuit current, t1 is rise time constant, t2 is
decay time constant, and Tk is short-circuit duration
Zc Zc
Γ Rf
Figure 4.5 Reflection of voltage surge at the fault location. As the surge arrives at
the fault location from VSC terminal, some parts of it are reflected
according to the reflection coefficient G as a function of fault
resistance Rf
For a HVDC system that has a long transmission line, a line-to-ground or line-
to-line fault can cause a voltage reflection, especially when the fault resistance is
small [5,6]. When a short-circuit fault happens, the voltage at the fault location
decreases and the fault current rapidly increases. Immediately negative voltage sur-
ges start to flow from the fault location into the VSC terminals, where a part of the
negative voltage surge is reflected back as a positive surge due to the termination of
cable capacitance. This traveling wave that is reflected between the fault location
and VSC terminal could increase to a very high magnitude. Figure 4.5 shows the
fault resistance Rf , reflection of surges, reflection coefficient G, and characteristic
impedance Zc , which can be given as follows with the line impedance parameters.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 R þ jwL
G¼ ; Zc ¼ (4.4)
1 þ 2 Rf =Zc G þ jwC
4.2.3 Faults in various bus configurations

4.2.3.1 Point-to-point
The point-to-point (PtP) configuration has conventionally been employed in HVDC
power transmission systems with a converter station at each end, which can be
classified into monopolar and bipolar schemes.
In the monopolar configuration (Figure 4.6(a) and (b)), there is one converter in
each terminal connected through a transmission line. There are two variations:
asymmetrical and symmetrical structure. In asymmetrical monopolar system, the
converters are connected between an earthed pole and a positive pole. Earthed pole
provides earthing through a low-voltage conductor which gives a return path with
ground. A metallic return path can also be used. On the contrary, in symmetrical
monopolar scheme, the converters are connected between high-voltage independent
poles of opposite polarity: positive and negative. Symmetrical monopolar system can
be grounded with high-resistance grounding or ungrounded; high-resistance
grounding prevents larger fault current but causes higher ground potential.
Bipolar configuration (Figure 4.6(c)) is widely employed in HVDC systems [7].
It is the combination of two monopolar systems where two converters are connected
+ Pole + Pole + Pole + Pole
Converter 2 Converter 2
Converter 1 Converter 1
Earthed pole – Pole – Pole
(a) (b)
+ Pole HVDC line + Pole
Converter 3
Converter 1
Earthed pole
Converter 4
Converter 2
– Pole – Pole
(c)
Figure 4.6 Point-to-point configurations: (a) asymmetrical monopole,

(b) symmetrical monopole, and (c) bipole
in series between two poles at each terminal. The polarities of the poles are positive
and negative similar to the symmetric monopolar system, but there is an earthing or a
neutral point between the poles. The neutral points in the system can either be
connected to or disconnected from each other. No current flows through these points
in balanced operation. One of the poles can maintain the normal system operation
while there is disturbance in the other pole, and the AC side protection can be used to
ensure the converter protection, as well as the DC side protection schemes.
The possible faults in PtP configurations are line-to-ground fault, line-to-line
fault, and AC side faults. The monopolar system is highly vulnerable to a line-to-
ground fault as the power transmission gets fully interrupted unless a parallel path is
available, but normal system operation can be maintained with other poles in bipolar
configurations and it turns into an asymmetrical configuration; however, 50%
reduced power capacity can cause an overload to the sending side converters.
Although less likely to happen, the line-to-line fault causes very high current in the
system. As well as the direct contact between lines, two line-to-ground faults in both
poles can cause line-to-line fault if the system is ungrounded in symmetrical
monopolar system. The unit protection such as distance, directional, and differential
relaying can be effective solutions against faults in PtP configurations [8–10].
4.2.3.2 Multiterminal
In multiterminal configurations, a number of DC terminals are combined to form a
power system that has improved efficiency, reliability, and flexibility, especially
during fault conditions as power can be transmitted by using multiple paths. The
system can withstand a single point of failure and effectively operate without affecting
the whole system. Various network configurations are possible for MTDC systems.
In the ring configuration shown in Figure 4.7(a), a DC microgrid system uses a
closed-loop bus. It is suitable for high-voltage operation due to less voltage fluctuation
[11,12]. This type of system can be operated either in closed or open loop depending
on operational conditions. Typically, converters are connected at each terminal
through CBs to control the power flow between the bus and connected branches. The
CBs can isolate faulted devices without shutting down the whole system. Because
the load can be supported from the either side of the connection, the system can
continue to operate even if some faulted bus segments are separated. The bus can
typically be configured with multiple zones or sections with static switches or CBs;
hence, the unit protection or directional protection techniques can be readily applic-
able for fault detection, which can be implemented by a number of intelligent CBs
placed in the bus to protect the different zones during fault conditions.
Generally, radial configuration is utilized in low-voltage and low-power sys-
tems [13]. In this structure, separate distribution feeders from a single DC grid are
fed by distributed generators. The feeders and distributed generators are connected
in the grid with interface converters and protected by CBs or other protective
devices. Figure 4.7(b) shows a radial DC power system. It is widely used in power
distribution systems due to its major advantages such as low initial cost, simplicity,
1 2
Ring bus DC bus DC bus DC bus

1 1
Generation 1
side
2 Distribution 1 2 n
side n
n
n
3 n 1 2 n
(a) (b) (c)
1
n 2
DC bus DC bus
1 3
5 3
2 n
4
DC bus DC bus
(d) (e)
Figure 4.7 Multiterminal configurations: (a) ring, (b) radial, (c) ladder, (d) star,
and (e) mesh. Rectangular boxes denote circuit breakers
and effective coordination of protective devices [14]. However, it is not as reliable

since any fault on a feeder close to the main grid can hamper the downstream
consumers. Also, disconnecting one transmission feeder would cause power inter-
ruption to the whole system.
The ladder configuration is a dual DC bus architecture that consists of dis-
tributed arrays of power generation, converters, and loads which are connected to
the two buses as rungs in a ladder. Figure 4.7(c) shows a typical ladder configuration
of DC power system. The rungs can be operated as an individual or as a cascaded
power system. During the cascaded operation mode, each rung can share power
with others using multiple paths through the DC buses. Since there is a redundancy
in the arrays of primary power sources, this configuration can prevent the system
from single point of failure, and it can withstand multiple failures without affecting
the critical loads and power output of the system.
The star configuration is simple and secured topology that has been widely
used in large-scale offshore HVDC wind farms [15,16]. In a star configuration, all
the DC terminals are connected in a common star node through HVDC CBs as
illustrated in Figure 4.7(d). The main drawback of this configuration is that any
fault in the common node may cause voltage collapse and shut off the whole system
[17]. Sophisticated protection schemes are required in order to protect the central
node and enhance the system reliability.
The mesh configuration (Figure 4.7(e)) would be the most reliable configuration
in electrical distribution system although the network can be complex. This config-
uration consists of multiple paths between two points so that power can reroute in fault
conditions. In a meshed network, power can flow in any direction regardless of
upstream or downstream concept. The redundancy of path further improves the degree
of flexibility in power system planning. However, interface converters need to be
placed at the key points of the network to control the power flow in the system [18].
The converters and transmission lines are protected by CBs from both ends. This
structure can also withstand single or multiple points of failure. The drawback of this
configuration is power losses in the redundant paths and the associated cable cost.
4.3 Coordinated protection techniques

Coordination between protective devices is important to improve the protection
selectivity and operation continuity, as well as its capability to properly shut down
or separate the system from the possible faults or abnormal conditions. Conven-
tional concepts of AC system’s protective coordination, such as main/back-up
protection, time-interval/current settings, instantaneous trip (IT) devices, over-
current/distance relay settings, zone overlapping, and automatic reclosing, can
mostly be applied to the protection of DC power systems; however, the char-
acteristics of DC power needs to be taken into consideration.
Typically, DC microgrids have generation sources and bidirectional devices,
such as energy storage, as well as the loads. In case they are connected to the AC
grid through PWM converters, the power can flow bidirectionally through the point
of AC/DC coupling. Furthermore, the presence of energy storage or distributed

generators should be carefully considered for the fault detection/location and
restoration after the fault. Continuously flowing DC fault current is difficult to
interrupt due to the arc, and its fast-rising dynamics can impact the instantaneous
selectivity and time-interval/current settings. The location of the fault is challen-
ging as well because the fault transient that contains information about the fault
location happens at the peak of the fault current, which has to be interrupted as soon
as possible due to the lack of capability to sustain the fault current in typical VSC-
based DC microgrids. These make the design of protection system for DC micro-
grids complex compared to the typical radial AC power systems.
4.3.1 AC side protection

4.3.1.1 Fault detection
In case the DC microgrid has a connection point to an AC grid, AC CBs disconnect
the DC microgrid from the AC systems when some abnormalities occur in the AC
side. Possible AC side faults include phase-to-ground fault, phase-to-phase fault,
phase unbalance, converter transformer differential, transformer short-circuit, AC
overvoltage, AC overcurrent, AC filter overload/unbalance, and AC > DC current
differential fault. Caution needs to be taken for the DC microgrid protection system
not to detect AC side fault as a DC system fault as they would generate a dis-
turbance on DC side voltage and current, which typically appears as oscillations in
DC bus voltage and current. A method to use the rate of change of DC bus current
(diDC =dt) to differentiate AC side faults from DC faults, because diDC =dt is lower
for an AC side fault than that of a DC side fault due to the filtering effect of DC bus
capacitors [19]. Once the AC CB connecting a source VSC to AC grid trips, the DC
microgrid can experience a substantial power loss that requires coordination
between bus voltage controller, energy storage controller, and load side controllers
to mitigate the impact and continue the operation [20,21]. The internal faults of the
source VSC, e.g., failures on IGBTs or control power supply, that typically trip the
associated AC CB will also generate a similar effect.
4.3.1.2 Protection by AC circuit breakers
DC system protection can be achieved by using AC system’s CBs as well. The dif-
ferential protection scheme can be applied in the AC side of two-terminal DC systems
as shown in Figure 4.8. This approach can typically be seen in the applications like
PtP-type HVDC systems where the DC bus works as a conduit between AC systems,
and it needs cooperation between the AC CBs, power converters, and relays mon-
itoring DC side voltage and current. For the DC side fault conditions such as
DC under- or overvoltage, DC pole voltage differential, DC overcurrent, DC > AC
current differential fault, the converters are blocked and AC CBs are tripped. For a
LLC-based HVDC system, in conjunction with the trip of AC CBs, the thyristor rec-
tifier valves can be controlled in the forced retard mode to deenergize the bus, while
the inverter operates in bypass mode as shown in Figure 4.9 [22]. The rectifier polarity
will be reversed so that the bus current can decrease (retard mode) and the inverter
bypasses the DC current through bypass pairs (bypass mode). The protection system
87 87
CT CT
CT CT
CT CT
Figure 4.8 Differential protection scheme used in the AC sides of a two-terminal

DC system. The ANSI standard device number 87 is for differential
protection relay
+ + – +
– – + –
(a) (b)
Figure 4.9 LCC operation modes: (a) normal operation, (b) forced retard
operation. The rectifier polarity will be reversed so that the bus
current can decrease, while the inverter bypasses the DC current
through bypass pairs
has to be able to coordinate to properly discharge surge arresters and capacitors in the
bus. The system can enter the reclosing cycle, and the voltage level can be coordinated
in the restart mode (e.g., full or reduced voltage).
For MTDC systems, a ‘‘hand-shaking’’ method using fast-acting DC switches
as well as AC CBs has been proposed [23] (Figure 4.10). Although switches can
only open when there is no current, they can provide a lower cost solution com-
pared to those with DC CBs. Downside is that the DC bus has to be deenergized
because all of the AC CBs trip when a fault occurs. Using AC CBs would be one of
the economical ways to protect DC systems, but the slow interruption speed may be
an issue as the fastest AC CBs can interrupt the current in two cycles [24]. Fur-
thermore, it would completely shut down the whole system, which could be inef-
ficient especially for multiterminal systems. However, AC CBs can be readily used
as a back-up protection measure.
4.3.2 DC side protection

4.3.2.1 Fault detection
When a short-circuit fault happens in a DC bus, e.g., between line to ground or line
to line, the source devices connected to the bus, such as generators, bus capacitors,
and energy storage, will experience excessive current, and loads, as well as the
sources, will detect bus voltage drop. Typically DC bus fault can be detected by
AC#1 AC#2 AC#1 AC#2

VSC#1 VSC#2 VSC#1 VSC#2
VSC#3 VSC#3
AC#3 AC#3
(a) (b)
AC#1 AC#2
VSC#1 VSC#2
VSC#3
(c) AC#3
Figure 4.10 Handshaking method [23]: (a) fault current flows to the fault point. (b)
All AC CBs open followed by appropriate DC switches after the fault
current is extinguished. Each VSC opens one of the associated DC
switches that had outgoing current at the time of fault. If all of them
are outgoing, the one with highest magnitude is opened. (c) Reclose
the switch if it sees the bus voltage after the AC CBs are reclosed
monitoring bus undervoltage and overcurrent, and various detection schemes have
been proposed, including large current change, rise time, and oscillation pattern
methods [25], dv=dt [5,26], di=dt and d 2 i=dt2 measurement [27], traveling wave
[28], wavelet-based [29,30], and model-based [31] detection. Using the first- and
second-order differentiation of voltage or current measurement can reduce detec-
tion time so that a smaller capacity CB can be used, but it is susceptible to signal
noise and filtering could decrease the detection speed.
Short-circuit detection of line-to-ground fault, which is most common, is
relatively straightforward unless the bus is ungrounded or high-resistance grounded
as can be seen in traction and shipboard power systems, or the fault resistance is
large. In the large fault resistance cases, the fault current is difficult to distinguish
from a normal load current or detect it with ground fault relay. For an ungrounded
bus that is adopted to avoid problems like corrosion by ground leakage current,
there is no fault current when just one line is short-circuited to ground. This is an
advantage because the system can keep operating with one line short-circuited to
ground, but the impact is catastrophic if the first fault is not cleared before the
second line-to-ground fault happens in another pole. Although a ground fault can be
detected by measuring the voltage between the pole and ground, it is challenging to
extract more information for protection because of the low fault current level.
Unit protection techniques that compare the magnitudes or directions of
quantities at each end of a defined zone or unit, e.g., current differential protection,
can provide faster detection and better selectivity and have been widely used
typically for the primary protection method for AC systems. The current differ-
ential methods are readily applicable to DC systems and actually suitable to meet
the much faster protection requirements. Although it will require communication
between relays which increases the cost, the development of modern power systems
with increased number of sensors and communication infrastructure can help the
expansion of the usage of unit protection schemes. Differential detection schemes
have been suggested for various types and applications of DC power systems, and
many of them require high-speed data communication between protective devices.
Although it provides higher selectivity and flexibility, the communication-based
approach needs some independent back-up plan because communication error or
glitch needs to be taken into consideration. Distance protection techniques are also
widely used for AC systems to detect the fault and coordinate CBs, but small
impedance of DC cables would cause significant problems using distance relaying
methods for coordination (Figure 4.11) [5].
300
No fault detection
1st Zone 2nd Zone
250 U<,1
Voltage [kV]
dU
200
dt <,1
150
100
F1 (10 Ω)
50 F2 (0 Ω)
–6,000 –5,000 –4,000 –3,000 –2,000 –1,000 0

Voltage derivative [kV/ms]
Figure 4.11 A dv=dt-based detection scheme [5]: protection zones and thresholds
in the voltage/voltage derivative plane is shown. Locus of the voltage
U and voltage derivative dU =dt is plotted for two line-to-ground
faults F1 (near) and F2 (far) from the measuring point. The system is
assumed to be asymmetric monopolar with pole voltage 320 kV. It
can be seen that the fault can be detected using the undervoltage
threshold, e.g., U<;1 and derivative threshold ðdU =dtÞ<;1
Short-circuited
power pole
Figure 4.12 Fault current flows through short-circuited power pole via
antiparallel diode of a DC CB
Most of the devices participating in a DC microgrid are interfaced with the bus
through active power converters, such as LCCs and VSCs, which will disconnect
from the bus to protect themselves by blocking all the switches to avoid overload or
malfunctions from DC link undervoltage when a bus fault occurs. Diode rectifiers
are passive, but they typically have contactors and fuses for separation. Therefore,
it would be necessary for the protection system to interrupt the fault current and
isolate the faulted section as soon as possible to minimize the impact of the fault
and downtime of the system. However, if the DC bus is interfaced to AC grid
through a VSC, it should be noted that the VSC cannot block the fault current from
AC side. This happens because the bus voltage rapidly decreases due to the dis-
charge of DC bus capacitors and sources cannot supply enough current to sustain
the bus voltage. Hence, the protection system in DC side has to manage fault
conditions during this quite short critical time period or should coordinate with the
AC CBs to achieve a back-up protection from AC side. Similarly, if a fault happens
in VSC’s internal power poles, a single solid-state DC CB in the line cannot
block the fault current coming from the DC bus side through antiparallel diode
(Figure 4.12). Bidirectional solid-state CBs can be used, but high conduction loss
can be an issue.
4.3.2.2 Postfault processes

The fault detection is typically followed by fault current interruption, fault location,
and isolation processes. Fault current interruption in DC side can be implemented
using DC CBs of mechanical, solid-state, and hybrid types and fast-acting DC
switches. Typically, DC power systems are configured as shown in Figure 4.7, and
the CBs can be placed at the terminals of each VSC or at the end of each line.
Physical separation of the line can be implemented with additional DC switches.
Although having DC CBs in all of the lines would be most efficient for protection,
it may not be economical considering the cost of the DC CBs especially for the
medium-voltage level. DC/DC converters and choppers have been investigated for
DC fault current interruption as well. Conventional protective device coordination
techniques in AC systems, such as short-circuit calculations, main/back-up pro-
tection, and current-time coordinated settings, can be applied to the solid-state DC
CBs, fast-acting fuses, and no-fuse mechanical CBs in the DC system as well.
However, the conventional time-current coordination between CBs could be chal-
lenging in DC systems due to the fast dynamics, bidirectional nature of power flow,
and very short time interval that is allowed between CBs.
To effectively interrupt the bus from the AC and DC feeder, power converters
such as thyristor rectifiers and DC/DC converters have been suggested to limit and
block the fault current [32,33]. Fault current interruption can be coordinated with
current limiter to reduce the interruption time [34]. Modern protection systems
have adopted communication-based schemes to increase selectivity and flexibility.
It is relatively easy to isolate the faulted segment when the differential or direc-
tional detection is used (Figure 4.13) [35].
Fault location is an important task for protection systems to reduce the repair
time and system downtime. In AC systems, the fault location has been typically
performed by microcontroller-based relays that calculate the location of the fault
using short-circuit study data, traveling wave measurements, and impedance-based
methods. For the impedance-based algorithms, the related voltage and current
phase information is required, and this can be provided by phasor measurement
unit. The impedance-based distance calculation concept can readily be applied to
20
Probe voltage [V]
10
0
–10
–20
Power line 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Probe unit Time [s]
100
Probe current [A]
controller iP
Measurements 50
Control signals
Bypass switch 0
CP
for DC probing –50
–100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time [s]
+ 2.5
VP
–
Magnitude
2
Current sensor 1.5
1
0.5
Switch-mode power supply 0
0 50 100 150 200 250 300 350
for DC/AC probe power Frequency [Hz]
(a) (b)
Figure 4.13 Sinusoidal DC fault probe unit [35]: (a) probe unit. (b) Probe
voltage, current, and FFT result (from top). The probe unit applies
voltage with sweeping frequency and the current response shows the
characteristic of fault circuit
the fault location in DC microgrids. Techniques that extract information from the
instantaneous measurement have been suggested [36–38]; however, due to the
quite lower reactance and shorter length of the cables compared to those of AC
systems, it is not trivial to measure from the transient current especially when the
fault resistance is low. Furthermore, the transient happens at the top of the fault
current; however, the protection system must interrupt the fault current as
quickly as possible before the transient can be measured because the converters
typically cannot sustain the large fault current. Some off-line measurement
approaches using signal analysis have been proposed, e.g., traveling-wave-based
methods [39,40], wavelet [30,41], and probe signal injection [3,42]. Although
these can concurrently achieve fast fault current interruption and location, they
require additional hardware equipment and can be time-consuming or inaccurate
for complex configurations.
Isolation of the fault will minimize the affected area and maximize operation
continuity once the fault current is interrupted. The isolation of the faulted device,
e.g., a converter, from the bus is relatively straightforward because it typically trips
itself out, but it can generate some disturbance on the bus. For example, when a
source converter is lost from a DC microgrid system, the bus voltage will be main-
tained by other source converters if the load is sustainable with remaining sources;
however, the voltage will overshoot if that converter is back on line. Chopper circuits
[43,44], snubber circuits [45,46], and controller-based approaches [47,48] have
been suggested to suppress the overvoltage transient (Figure 4.14). Compared to the
typical control scheme that a single controller supplies the control actions for normal
and fault conditions, these multiple controller approaches have separate controllers
DC chopper
REC SEC
(a)
Ds Ll Ds Ll
Cs Cs
Rs
Rg Rs
Rg Rs
Rs
Cs Ds Cs
Ll Ll
Ds
(b) (c)
Figure 4.14 DC bus overvoltage suppression: (a) DC chopper, (b) discharge-

suppressing RCD snubber, (c) charge–discharge RCD snubber
Load
AC AC
microgrid 1 microgrid 2
VSC VSC
A F A
Bus 2 Bus 3 Bus 4
20 kV DC B B B B B B F B
Bus 1
B
VSC A
B
F1 Load
AC
network
B Bus 5
B B
A F
A Bus 6
VSC
VSC
PV Load
Wind system
turbine
Figure 4.15 Exemplary medium voltage DC (MVDC) network, where boxed A, B,

and F denote circuit breaker, static switch, and fuse, respectively
[49]. A local protection unit is installed at each connection point of
breaker to detect the faults comparing the line current and
predetermined threshold value. Once fault F1 is detected, the circuit
breakers will interrupt the fault current, and the line end static
switches will isolate the faulty section afterwards
that regulate steady-state variables such as power and bus voltage, and transient
disturbances, e.g., loss of a converter, respectively. In a fault condition, the fault
controller takes control to maintain the best possible performance and continuous
operation by redistributing setpoints or shutting down load converters. The idea of
multiple hierarchical controllers is similar to the coordination of protective devices
in AC systems.
In case the fault happens in the bus, the faulted segment has to be isolated so
that the other intact parts of the microgrid continue to operate. This task is espe-
cially important for the mission-critical power systems. For example, the zones of
DC shipboard power systems are implemented with DC CBs, static switches and
redundant buses, so that the load can be continuously supported after the fault is
isolated. The system can be reconfigured to operate without the faulted section and
coordination between protection relays and source converters may be required in
this reconfiguration or self-healing process. In the system shown in Figure 4.15
[49], a fault can be isolated by stopping the source converters, opening the CBs,
locate the fault, and open associated static switches. Then, the system can resume
operation by restarting the source converters; however, load shedding or load
redistribution would be necessary because of the reduced capacity.
4.3.3 Applications
4.3.3.1 High-voltage DC transmission
An example can be found in wind farms collector bus system where long distance
power transmission is required from off-shore wind farms to on-shore grid. Gen-
erally, wind turbines are connected to a collector bus via step-up transformer and
high-capacity converters through collector feeders. The collector bus is equipped
with protective devices to respond during faults in both collection and transmission
side. For the faults in HVDC link, unit protection schemes such as current differ-
ential, directional, and distance relaying are mostly applicable. Since it is long-
distance transmission, nonunit protection such as traveling-wave and wavelet
transformation schemes can also be potential solutions. All of these protection
schemes should have primary and back-up protection capabilities.
However, any fault in collection side also causes problem in operation of DC
collector bus. Generally, the wind turbine generators are protected by unit protec-
tion schemes with individual generator protection relay and CBs, and the trans-
former has either a fuse or separate protection relay based on the capacity. If any
fault happens near the wind turbine, the generator relay coordinates with the
transformer or converter protection devices to isolate the faulty section from the
collector feeder which is connected to the collector bus. If this primary protection
fails, then the collector bus relay is activated as a back-up protection measure.
However, to enhance the reliability of the protection, effective coordination and
communication of the protective devices need to be ensured.
4.3.3.2 Shipboard power system

Conventional AC shipboard power system is moving toward the next generation
medium-voltage DC (MVDC) architecture as DC is advantageous over AC power
system, e.g., higher efficiency and reliability, no issues with frequency and reactive
power regulation, and simpler interface and control. For the MVDC and LVDC
shipboard power systems, fault tolerant DC zonal electrical distribution system
(DCZEDS) concept has been widely investigated. The DCZEDS allows a number
of electrical zones connected to the main bus to improve the survivability of the
system during abnormal events. For the protection of DCZEDS, unit-based and
breaker-based approaches have been extensively investigated [50].
In the unit-based approach, the DCZEDS allows two separate longitudinal
buses along the length of the ship that provides redundant paths to deliver power to
the critical loads from both the buses through power converters and no-load elec-
tromechanical switches [51]. If fault happens in a certain zone, this approach
enables the fault-feeding converters of inherent current limiting and voltage fold
back capabilities to decouple the fault effect from upstream generators and coor-
dinate with the downstream no-load contactors to deenergize the faulty zone. If
fault happens in one longitudinal bus, the DCZEDS utilizes the other bus to supply
the critical loads during system reconfiguration (Figure 4.16).
The breaker-based protection approach relies on the parallel combination of
protection devices such as DC CBs and no-load contactors. The DCZEDS form a
Port longitudinal bus
S-322 Zone 3 S-321 S-322 Zone 2 S-221 S-122 Zone 1 S-121
S-323 S-223 S-123
DC DC DC
S-324 load S-224 load S-124 load
HV HV HV
input input input
PCR PCR PCR
AC AC AC
S-314 load S-214 load S-114 load
S-313 S-213 S-113
S-312 S-311 S-212 S-211 S-112 S-111

Starboard longitudinal bus
Figure 4.16 An unit-based protection scheme for a shipboard DC zonal electrical

distribution system [51]. The adjacent zones consist of switches and
phase controlled rectifier (PCR) with communication links that
detect the fault from predetermined current threshold value, locate
the fault comparing the vector direction of current in each zone,
and isolate the faulted section afterwards
ring bus configuration by connecting the longitudinal buses with solid-state CBs
which allows to isolate the faulty section without deenergizing the entire bus. In a
fault condition, DC CBs should react very fast to extinguish the fault current to zero
before the upstream converters shut down by themselves. However, to coordinate
the fault isolation and system reconfiguration, effective communication is required
between the converters, CBs, and switches, and at the same time, fault location
schemes are also required. Traveling-wave-based techniques have been suggested
for locating fault as it offers sufficient accuracy [52]. Artificial neural network and
wavelets have been suggested as well [53].
4.3.3.3 Traction power systems

Mass transit systems such as light-rail and subway systems are growing rapidly
throughout the world due to the urbanization and massive traffic growth. Single-phase
AC as well as DC power system has been employed for traction system. A typical
substation in DC traction power systems consists of a transformer and rectifiers from
high-voltage AC supply which feeds bus bars through high-speed DC CBs and
bypasses switches. The bus voltage used in DC traction power systems is relatively
low (600/750/1,500 V), and the ground impedance is high [54]; as a consequence, it is
challenging to differentiate between the line-to-ground fault current and start-up
current drawn from substation which are close in magnitude. It becomes worse if a
short-circuit fault occurs at a distant place from substations (Figure 4.17).
di=dt Measurement-based protection methods have been proposed [55]; if cur-
rent outstrips the setting limit at any instant of time with higher di=dt than the
threshold and lasts longer than the setting time interval, then a short-circuit fault is
detected, otherwise it is a start-up current. Auto reclosers are also installed at the
i ∆t
Setting duration Trip
∆t1 Trip
(2)
(4)
(3)
(1)
Setting ∆i
∆tre
Setting di/dt Protection reset t
Figure 4.17 A di=dt protection scheme for traction power system [55]. If di=dt
rises and exceeds the set point, tripping decision will be taken only
when both Di and Dt are higher than the setting limits [curves (2) and
(3)]. However, if there is any momentary reduction in rising di=dt but
Dt is higher than the protection reset interval (Dtre ), then there will
be no tripping [curve (4)]
track feeders to see if the fault is temporary. However, typically, the traction power
system is ungrounded or grounded with high resistance to mitigate ground leakage
current and corrosion, which makes it difficult to detect line-to-ground fault. The
system experiences a line-to-line fault if ground fault happens at the same time on
another pole. Hence, it is required to detect and locate the line-to-ground fault
quickly to avoid unwanted discontinuity in the service due to a line-to-line fault.
Techniques such as off-line impedance probe unit [35], traveling wave-based
methods [56], superconducting fault current limiter [57], and high-speed digital relay-
based protection [58] have been suggested for traction power system fault protection.
4.3.3.4 Photovoltaic park

A number of parallel strings consist of series connected PV panels form a PV
generation system, which is tied to the main bus through power electronics con-
verters and protective devices. Possible faults in a PV power system include arc,
open-circuit, line-to-ground, and line-to-line fault. The line-to-ground fault hap-
pens most often, and line-to-line fault occurs typically due to water ingress, animal
chewing, mechanical damage, or DC junction box corrosion [59].
Considering the short length of bus, an unit protection with DC switches can
be utilized for PV parks. Overcurrent protection devices (OCPDs, e.g., fuse) and
ground-fault protection devices (GFPDs) are used as well in typical PV systems
(Figure 4.18) to detect line-to-line and line-to-ground faults, respectively. A line-to-
line fault in PV strings is difficult to detect and protection becomes challenging if
solar irradiance is low or a fault occurs between two points of close electrical
potential [60]. The OCPDs respond to the fault with typically 2.1 times of short-
circuit current at standard test condition (1,000 W m–2, 25 C, 1.5 air mass) [60], but
IPV
Blocking
I1 diode I2 In
OCPD
GFPD Converter
PV PV PV
F1: L-G
F2: L-L
PV PV PV
PV PV PV DC bus
Figure 4.18 A typical PV array protection system. During line-to-ground and

line-to-line fault, fault current is interrupted by GFPD and OCPD,
respectively
at low solar irradiation specially at cloudy day or evening, fault current level is
considerably low. If the fault remains undetected, it could potentially damage the
system, and furthermore, it consumes power from the healthy section as a load.
Although a blocking diode can be placed on each string to cut off the back-feeding
current, it challenges the OCPDs in fault interruption [59]. Furthermore, GFPDs in a
traditional PV system may trip the whole system for a line-to-ground fault. Recent
studies have been mainly focused on fault location and detection strategies. A vol-
tage transducers-based scheme has been proposed in [61] to protect against line-to-
ground as well as line-to-line fault, in which the top and bottom end string voltage is
compared with predefined threshold to classify and locate the fault. Differential
protection [62] and machine learning-based protection [63] strategies have been
suggested as well for PV system fault location and detection.
4.3.3.5 Data centers

In the past decade, there has been a significant advancement in data center from a
small server room to large data centers. Besides AC grids, conventional diesel gen-
erators and uninterruptible power supply (UPS) are also employed to supply power
during grid outage. Data center loads are mostly DC, comprised of IT equipment
such as servers, routers, switches, and energy storage devices. From AC grid to the
loads, several AC–DC converter stations are installed with associated protective
devices typically in radial configuration. However, data center requires highly reli-
able and accurate fault protection schemes to maintain the service availability.
The unit protection technique with microcontroller-based relay can be applic-
able for the primary detection and location. If the relays and DC CBs fail to respond
to the relay’s trip signal, the converters will shut off with self-protection capability as
back-up protection. However, it should be noted that the power distribution system in
DC data centers is highly capacitive in nature because of UPSs and multiple power
supply units in parallel, which causes problem during short-circuit failure [64].
During a fault, the capacitor discharges so that the voltage falls down rapidly and the
amplitude of the fault current rises to a very high level within a very short time.
Moreover, this current can be detected by protective devices other than those located
on the fault path and unnecessarily trips the intact sections in the system. A relay
should be installed on the bus with coordinated fast reacting protective devices, so
that the faulty segment of the bus can be isolated without interrupting the healthy
sections.
4.3.3.6 Aircraft power systems

An aircraft power system is mostly small-scale AC/DC hybrid system with 230 V AC/
360–800 Hz and 270 V DC, consists of small generators, transmission–distribution
system, electrical components, and onboard loads, as well as some energy storage
devices as an emergency power supply for vital loads [66,67]. The common faults
in aircraft are arc fault, overloads, and electrical short-circuit fault, and differential
IDG 1
External 2 IDG
AC
APU 2 Receptacle
GCU 1 GAPCU
GCU 2
APU line cont EXT power cont
Gen line cont 1 Gen line cont 2
Bus tie cont 1

Transfer busbar Bus tie cont 2
CB CB
Main AC bus 1 Main AC bus 2
CSM/
G
CSM/G control Vital AC bus Semi-vital AC bus
FCLCB unit FCLCB
TR1 Vital TR Vital DC bus Semi-vital AC bus TR2
DC bus 1 DC battery bus 1 DC battery bus 1 DC bus 1

DC norm bus 2
DC norm bus 1 switching contactor
switching contactor
Battery charge Battery charge

limiter(2) limiter(2)
HOT bus 1 Hot bus 2

Battery 1 Battery 2
SHUNT
SHUNT
Figure 4.19 Single line diagram of the j150þ aircraft restructured distribution
network for a more electric aircraft (MEA) [65]
current protection and OCPDs are mainly used for single DC bus-based system [65].
In case of the ring-bus aircraft, directional protection scheme can be applicable,
because once a fault happens at a point, then current will flow to that point from both
sides. Fault current can be interrupted by fuses, but replacement would be difficult for
aircraft onboard power systems. Resettable CBs can be used, but it may not be an
effective solution to arc fault detection and interruption. Semiconductor-based pro-
tective devices have been suggested for arc fault, which also provide protection
against short-circuit fault and overloads using I 2 t operating characteristics [68].
Recently, fault current limiting CBs (FCLCBs) based on series resonance with
superior current limiting and breaking advantages have been proposed [65]. A single
line diagram of a more electric aircraft power system is shown in Figure 4.19.
4.4 Summary
DC power systems have been investigated for transmission and distribution systems
due to the advantages such as improved efficiency, reliability, and simpler con-
figuration, as well as increasing DC-based distributed generation, energy storage,
and loads. Along with comparatively short operating history and insufficient
guidelines and standards, one of the current challenges is system protection tech-
niques to deal with the highly dynamic and nonzero crossing nature of DC power.
In this chapter, various coordinated protection schemes and applications for DC
power systems have been reviewed in conjunction with the analysis of DC fault
currents dynamics and topological bus configurations.
Acknowledgment
This work was supported by the National Science Foundation (NSF) under Award
ECCS-1554626.
References
[1] T. Dragičević and F. Blaabjerg, Microgrid: advanced control methods and
renewable energy system integration. Waltham, MA: Butterworth-Heinemann,
2016, pp. 263–279.
[2] IEEE Power and Energy Society, ‘‘IEEE guide for determining fault location
on AC transmission and distribution lines,’’ IEEE Std C37.114-2004, pp. 1–36,
2005.
[3] J. D. Park, J. Candelaria, L. Ma, and K. Dunn, ‘‘DC ring-bus microgrid fault
protection and identification of fault location,’’ IEEE Transactions on Power
Delivery, vol. 28, no. 4, pp. 2574–2584, October 2013.
[4] A. Berizzi, A. Silvestri, D. Zaninelli, and S. Massucco, ‘‘Short-circuit
current calculations for DC systems,’’ IEEE Transactions on Industry
Applications, vol. 32, no. 5, pp. 990–997, 1996.
[5] W. Leterme, J. Beerten, and D. V. Hertem, ‘‘Nonunit protection of HVDC

grids with inductive DC cable termination,’’ IEEE Transactions on Power
Delivery, vol. 31, no. 2, pp. 820–828, April 2016.
[6] M. K. Bucher and C. M. Franck, ‘‘Contribution of fault current sources in
multiterminal HVDC cable networks,’’ IEEE Transactions on Power
Delivery, vol. 28, no. 3, pp. 1796–1803, 2013.
[7] D. Jovcic and K. Ahmed, High voltage direct current transmission: con-
verters, systems and DC grids. Chichester: John Wiley & Sons, 2015.
[8] J. Suonan, J. Zhang, Z. Jiao, L. Yang, and G. Song, ‘‘Distance protection for
HVDC transmission lines considering frequency-dependent parameters,’’
IEEE Transactions on Power Delivery, vol. 28, no. 2, pp. 723–732, 2013.
[9] S. Luo, X. Dong, S. Shi, and B. Wang, ‘‘A directional protection scheme for
HVDC transmission lines based on reactive energy,’’ IEEE Transactions on
Power Delivery, vol. 31, no. 2, pp. 559–567, 2016.
[10] E. Kontos, G. Tsolaridis, R. Teodorescu, and P. Bauer, ‘‘On dc fault
dynamics of MMC-based HVDC connections,’’ IEEE Transactions on
[11] D. Pieniazek, ‘‘HV substation design: applications and considerations,’’ in
IEEE CED–Houston Chapter, 2012.
[12] V. Mehta and R. Mehta, Principles of power system, New Delhi: S. Chand, 2004.
[13] A. Kwasinski, ‘‘Advanced power electronics enabled distribution architectures:
design, operation, and control,’’ in 2011 IEEE 8th International Conference on
Power Electronics and ECCE Asia (ICPE & ECCE), 2011, pp. 1484–1491.
[14] M. K. Bucher, R. Wiget, G. Andersson, and C. M. Franck, ‘‘Multiterminal
HVDC networks—what is the preferred topology?’’ IEEE Transactions on
[15] J. Yang, J. E. Fletcher, and J. O’Reilly, ‘‘Multiterminal DC wind farm col-
lection grid internal fault analysis and protection design,’’ IEEE Transac-
tions on Power Delivery, vol. 25, no. 4, pp. 2308–2318, 2010.
[16] P. Bresesti, W. L. Kling, R. L. Hendriks, and R. Vailati, ‘‘HVDC connection
of offshore wind farms to the transmission system,’’ IEEE Transactions on
Energy Conversion, vol. 22, no. 1, pp. 37–43, 2007.
[17] O. Gomis-Bellmunt, J. Liang, J. Ekanayake, R. King, and N. Jenkins, ‘‘Topol-
ogies of multiterminal HVDC-VSC transmission for large offshore wind
farms,’’ Electric Power Systems Research, vol. 81, no. 2, pp. 271–281, 2011.
[18] Y. Liu, J. Wang, N. Li, Y. Fu, and Y. Ji, ‘‘Enhanced load power sharing
accuracy in droop-controlled DC microgrids with both mesh and radial
configurations,’’ Energies, vol. 8, no. 5, pp. 3591–3605, 2015.
[19] L. Tang and B.-T. Ooi, ‘‘Protection of VSC-multi-terminal HVDC against
DC faults,’’ in 2002 IEEE 33rd Annual IEEE Power Electronics Specialists
Conference. Proceedings (Cat. No.02CH37289), vol. 2, 2002, pp. 719–724.
[20] D. Chen, L. Xu, and L. Yao, ‘‘DC voltage variation based autonomous
control of DC microgrids,’’ IEEE Transactions on Power Delivery, vol. 28,
no. 2, pp. 637–648, April 2013.
[21] N. Eghtedarpour and E. Farjah, ‘‘Distributed charge/discharge control of

energy storages in a renewable-energy-based DC micro-grid,’’ IET Renew-
able Power Generation, vol. 8, no. 1, pp. 45–57, January 2014.
[22] C. Kim, V. Sood, G. Jang, S. Lim, and S. Lee, HVDC transmission: power
conversion applications in power systems. Singapore: John Wiley & Sons
(Asia), 2009.
[23] L. Tang and B. Ooi, ‘‘Locating and isolating DC faults in multi-terminal
DC systems,’’ IEEE Transactions on Power Delivery, vol. 22, no. 3,
pp. 1877–1884, July 2007.
[24] IEEE Power and Energy Society, ‘‘IEEE application guide for AC high-
voltage circuit breakers rated on a symmetrical current basis,’’ in ANSI/IEEE
C37.010-1979, pp. 1–54, 1980.
[25] L. Tang and B. Ooi, ‘‘Protection of VSC-multi-terminal HVDC against DC
faults,’’ in 33rd Annual IEEE Power Electronics Specialist Conference,
vol. 2, pp. 719–724, November 2002.
[26] S. Foster, L. Xu, and B. Fox, ‘‘Control of an LCC HVDC system for con-
necting large offshore wind farms with special consideration of grid fault,’’
in 2008 IEEE Power and Energy Society General Meeting – Conversion and
Delivery of Electrical Energy in the 21st Century, July 2008, pp. 1–8.
[27] A. Meghwani, S. C. Srivastava, and S. Chakrabarti, ‘‘A non-unit protection
scheme for DC microgrid based on local measurements,’’ IEEE Transactions
on Power Delivery, vol. 32, no. 1, pp. 172–181, February 2017.
[28] S. Azizi, M. Sanaye-Pasand, M. Abedini, and A. Hasani, ‘‘A traveling-wave-
based methodology for wide-area fault location in multiterminal DC sys-
tems,’’ IEEE Transactions on Power Delivery, vol. 29, no. 6, pp. 2552–2560,
December 2014.
[29] K. D. Kerf, K. Srivastava, M. Reza, D. Bekaert, S. Cole, D. V. Hertem, and
R. Belmans, ‘‘Wavelet-based protection strategy for DC faults in multi-
terminal VSC HVDC systems,’’ IET Generation, Transmission Distribution,
vol. 5, no. 4, pp. 496–503, April 2011.
[30] W. Li, A. Monti, and F. Ponci, ‘‘Fault detection and classification in medium
voltage DC shipboard power systems with wavelets and artificial neural
networks,’’ IEEE Transactions on Instrumentation and Measurement, vol.
63, no. 11, pp. 2651–2665, November 2014.
[31] Y. Ji and J. Bals, ‘‘Multi-model based fault detection for the power system of
more electric aircraft,’’ in 2009 7th Asian Control Conference, August 2009,
pp. 93–98.
[32] D. Dong, Y. Pan, R. Lai, X. Wu, and K. Weeber, ‘‘Active fault-current
foldback control in thyristor rectifier for DC shipboard electrical system,’’
IEEE Journal of Emerging and Selected Topics in Power Electronics, vol. 5,
no. 1, pp. 203–212, March 2017.
[33] M. Baran and N. Mahajan, ‘‘Overcurrent protection on voltage sourced
converter based multiterminal DC distribution systems,’’ IEEE Transactions
on Power Delivery, vol. 22, no. 1, pp. 406–412, January 2007.
[34] C. Zhan, C. Smith, A. Crane, A. Bullock, and D. Grieve, ‘‘DC transmission

and distribution system for a large offshore wind farm,’’ in 9th IET Inter-
national Conference on AC and DC Power Transmission (ACDC 2010),
October 2010, pp. 1–5.
[35] J. D. Park, ‘‘Ground fault detection and location for ungrounded DC traction
power systems,’’ IEEE Transactions on Vehicular Technology, vol. 64,
no. 12, pp. 5667–5676, December 2015.
[36] J. Yang, J. E. Fletcher, and J. O’Reilly, ‘‘Short-circuit and ground fault
analyses and location in VSC-based DC network cables,’’ IEEE Transac-
tions on Industrial Electronics, vol. 59, no. 10, pp. 3827–3837, October
2012.
[37] Y. Pan, M. Steurer, and T. L. Baldwin, ‘‘Ground fault location testing of a
noise-pattern-based approach on an ungrounded DC system,’’ IEEE Trans-
actions on Industry Applications, vol. 47, no. 2, pp. 996–1002, March 2011.
[38] J. Cheng, M. Guan, L. Tang, H. Huang, X. Chen, and J. Xie, ‘‘Paralleled
multi-terminal DC transmission line fault locating method based on travel-
ling wave,’’ IET Generation, Transmission Distribution, vol. 8, no. 12,
pp. 2092–2101, 2014.
[39] X. Liu, A. H. Osman, and O. P. Malik, ‘‘Hybrid traveling wave/boundary
protection for monopolar HVDC line,’’ IEEE Transactions on Power
Delivery, vol. 24, no. 2, pp. 569–578, April 2009.
[40] Y. J. Kwon, S. H. Kang, D. G. Lee, and H. K. Kim, ‘‘Fault location algorithm
based on cross correlation method for HVDC cable lines,’’ in 2008 IET 9th
International Conference on Developments in Power System Protection
(DPSP 2008), March 2008, pp. 360–364.
[41] K. Jia, T. Bi, B. Liu, E. Christopher, D. W. P. Thomas, and M. Sumner,
‘‘Marine power distribution system fault location using a portable injection
unit,’’ IEEE Transactions on Power Delivery, vol. 30, no. 2, pp. 818–826,
April 2015.
[42] E. Christopher, M. Sumner, D. W. P. Thomas, X. Wang, and F. de Wildt,
‘‘Fault location in a zonal DC marine power system using active impedance
estimation,’’ IEEE Transactions on Industry Applications, vol. 49, no. 2,
pp. 860–865, March 2013.
[43] C. Feltes, H. Wrede, F. W. Koch, and I. Erlich, ‘‘Enhanced fault ride-through
method for wind farms connected to the grid through VSC-based HVDC
transmission,’’ IEEE Transactions on Power Systems, vol. 24, no. 3,
pp. 1537–1546, August 2009.
[44] L. Bing, J. Xu, R. Torres-Olguin, and T. Undeland, ‘‘Fault mitigation control
design for grid integration of offshore wind farms and oil and gas installa-
tions using VSC-HVDC,’’ in International Symposium on Power Electro-
nics, pp. 792–797, June 2010.
[45] J. D. Park and J. Candelaria, ‘‘Fault detection and isolation in low-voltage
DC-bus microgrid system,’’ IEEE Transactions on Power Delivery, vol. 28,
no. 2, pp. 779–787, April 2013.
[46] F. Liu, W. Liu, X. Zha, H. Yang, and K. Feng, ‘‘Solid-state circuit breaker
snubber design for transient overvoltage suppression at bus fault interruption
in low-voltage DC microgrid,’’ IEEE Transactions on Power Electronics,
vol. 32, no. 4, pp. 3007–3021, April 2017.
[47] W. Lu and B. Ooi, ‘‘DC overvoltage control during loss of converter in
multiterminal voltage sourced converter based HVDC (M-VSC-HVDC),’’
IEEE Transactions on Power Delivery, vol. 18, no. 3, pp. 915–920, July
2003.
[48] L. Lamont, D. Jovcic, and K. Abbot, ‘‘VSC transmission control under
faults,’’ Universities Power Engineering Conference, vol. 3, pp. 1209–1213,
2004.
[49] M. Monadi, C. Koch-Ciobotaru, A. Luna, J. I. Candela, and P. Rodriguez,
‘‘Multi-terminal medium voltage DC grids fault location and isolation,’’ IET
Generation, Transmission Distribution, vol. 10, no. 14, pp. 3517–3528,
2016.
[50] R. M. Cuzner and V. Singh, ‘‘Future shipboard MVdc system protection
requirements and solid-state protective device topological tradeoffs,’’ IEEE
Journal of Emerging and Selected Topics in Power Electronics, vol. 5, no. 1,
pp. 244–259, 2017.
[51] M. W. Rose and R. M. Cuzner, ‘‘Fault isolation and reconfiguration in a
three-zone system,’’ in 2015 IEEE Electric Ship Technologies Symposium
(ESTS), 2015, pp. 409–414.
[52] E. Schweitzer, A. Guzman, M. Mynam, V. Skendzic, B. Kasztenny, and
S. Marx, ‘‘Protective relays with traveling wave technology revolutionize fault
locating,’’ IEEE Power and Energy Magazine, vol. 14, no. 2, pp. 114–120,
2016.
[53] Y. M. Yeap and A. Ukil, ‘‘Wavelet based fault analysis in HVDC system,’’
in 2014 40th Annual Conference of the IEEE Industrial Electronics Society
(IECON), 2014, pp. 2472–2478.
[54] R. Hill, ‘‘Electric railway traction. part 3. traction power supplies,’’ Power
Engineering Journal, vol. 8, no. 6, pp. 275–286, 1994.
[55] M. X. Li, J. H. He, Z.-q. Q. Bo, H. Yip, L. Yu, and A. Klimek, ‘‘Simulation
and algorithm development of protection scheme in DC traction system,’’ in
PowerTech, 2009 IEEE Bucharest. IEEE, 2009, pp. 1–6.
[56] C. Chang, T. Feng, A. Khambadkone, and S. Kumar, ‘‘Remote short-circuit
current determination in DC railway systems using wavelet transform,’’ IEE
Proceedings-Electric Power Applications, vol. 147, no. 6, pp. 520–526,
2000.
[57] H.-S. Shin, S.-M. Cho, and J.-C. Kim, ‘‘Protection scheme using SFCL for
electric railways with automatic power changeover switch system,’’ IEEE
Transactions on Applied Superconductivity, vol. 22, no. 3, pp. 1–4, 2012.
[58] E. Cinieri, A. Fumi, V. Salvatori, and C. Spalvieri, ‘‘A new high-speed
digital relay protection of the 3-kVdc electric railway lines,’’ IEEE Trans-
actions on Power Delivery, vol. 22, no. 4, pp. 2262–2270, 2007.
[59] Y. Zhao, J.-F. de Palma, J. Mosesian, R. Lyons, and B. Lehman, ‘‘Line–line

fault analysis and protection challenges in solar photovoltaic arrays,’’ IEEE
Transactions on Industrial Electronics, vol. 60, no. 9, pp. 3784–3795, 2013.
[60] National Fire Protection Association (NFPA), ‘‘Article 690—solar photo-
voltaic systems,’’ National Electrical Code, 2011.
[61] K. A. Saleh, A. Hooshyar, E. F. El-Saadany, and H. H. Zeineldin, ‘‘Voltage-
based protection scheme for faults within utility-scale photovoltaic arrays,’’
IEEE Transactions on Smart Grid, 2017. Early Access.
[62] S. Dhar, R. Patnaik, and P. Dash, ‘‘Fault detection and location of photo-
voltaic based DC microgrid using differential protection strategy,’’ IEEE
Transactions on Smart Grid, 2017. Early Access.
[63] Z. Yi and A. Etemadi, ‘‘Line-to-line fault detection for photovoltaic arrays
based on multi-resolution signal decomposition and two-stage support vector
machine,’’ IEEE Transactions on Industrial Electronics, vol. 64, no. 11,
pp. 8546–8556, 2017.
[64] G. Wawrzola, ‘‘Challenges of DC data center power distribution protec-
tion,’’ in 13th International Conference on Development in Power System
Protection 2016 (DPSP), March 2016, pp. 1–6.
[65] H. Radmanesh and A. Kavousi, ‘‘Aircraft electrical power distribution sys-
tem protection using smart circuit breaker,’’ IEEE Aerospace and Electronic
Systems Magazine, vol. 32, no. 1, pp. 30–40, 2017.
[66] J. Chen, X. Zhang, and C. Wen, ‘‘Harmonics attenuation and power factor
correction of a more electric aircraft power grid using active power filter,’’
IEEE Transactions on Industrial Electronics, vol. 63, no. 12, pp. 7310–7319,
2016.
[67] M. Terorde and D. Schulz, ‘‘New real-time heuristics for electrical load
rebalancing in aircraft,’’ IEEE Transactions on Aerospace and Electronic
Systems, vol. 52, no. 3, pp. 1120–1131, 2016.
[68] D. Izquierdo, A. Barrado, C. Fernández, M. Sanz, and A. Lázaro, ‘‘SSPC
active control strategy by optimal trajectory of the current for onboard
system applications,’’ IEEE Transactions on Industrial Electronics, vol. 60,
no. 11, pp. 5195–5205, 2013.
Chapter 5
Energy management systems for dc microgrids
Amjad Anvari-Moghaddam1, Tomislav Dragičević1,
and Marko Delimar2
5.1 Introduction
Over the past few years, large-scale utilization of renewable energy sources
together with the high penetration of low-to-medium voltage dc sources into the
generation mix of energy networks have increased attention toward deployment
of dc microgrids for improving system operation, reducing environmental foot-
prints, mitigating congestion and power losses through the network, increasing
energy efficiency and potentially boosting system economics. Moreover, with the
increasing share of dc appliances and loads, especially in residential and office
buildings, as well as the technological advancements in ac-to-dc rectifiers, dc
microgrids have become the point of interest for many research scientists and
engineers worldwide and have become a hot subject area for a large number of
research, development and demonstration activities. On the other hand, average
energy consumption distribution in different sectors at worldwide level illustrates
that more than one-third of the total consumption in each society is used by the
residential sector [1,2]. Although the increasing rate of energy consumption in
this sector has decelerated recently thanks to energy-aware solutions, energy
prices are still rising [3]. In this regard, many ongoing research activities are
directed toward energy efficiency in residential and office buildings and its
impact on the environment and climate [4–10]. Accordingly, a large number of
initiatives have been taken to promote sustainable consumption and green living
patterns as well as to promote delivery of affordable, clean and reliable power to
customers that, in turn, necessitates the use of smart automation and energy
management solutions. Although ‘‘energy management’’ is a term that could
mean differently, in this chapter, we are mainly concerned about the process of
energy saving and optimal scheduling of energy sources in a given system here
called microgrids.
1
2
Department of Energy and Power Systems, Faculty of Electrical Engineering and Computing,
University of Zagreb, Croatia
5.2 DC microgrid operation and control fundamentals

5.2.1 Power/energy management schemes
For optimal operation of a microgrid system in different working modes (i.e. in
grid-connected mode or islanded mode), it is crucial to properly design and
implement energy management systems (EMSs). These system optimizers nor-
mally determine the best possible operating scheme at the supply and demand sides
in terms of optimized set-points for controllable units and then send them as control
signals into the dedicated control system of interfacing converters. Generally, there
are two types of energy/power management strategies which are used in microgrid
applications; interactive schemes based on information sharing mechanisms and
passive schemes based on self-autonomy [11].
5.2.1.1 Interactive power/energy management strategies

In any given interactive power (IP)/EMS, local and global system information
(such as line currents, nodal voltages, frequency and powers) is communicated
in the microgrid and exchanged between corresponding nodes in order to determine
the operation point of each controllable distributed generation (DG) or consump-
tion unit. These strategies also benefit from a sort of intelligence in the integration
of the computing and communications technologies which help them to define and
develop the communication structure based on the computation burden of each
node and other related system’s objectives and constraints [12]. In this regard, three
different communication schemes can be realized for an IP/EMS: centralized,
decentralized and hybrid. In each of these schemes, different communication
technologies such as microwave (mW), power line communication (PLC), fiber-
optics, infrared and/or wireless radio networks (such as global system for mobile
communications and code division multiple access) can be effectively used and
integrated into existing infrastructures [13,14].
● Centralized P/EMS
In a centralized P/EMS, also known as a supervisory scheme, there is a
centralized entity or a control center that monitors the system’s behavior,
collects information from different parts of the network, makes decisions based
on the observations and accordingly updates set-points for the controllable
units in the supply/demand sides [15–19]. In other words, a centralized P/EMS
acts as a master unit while other local controllers (LCs) within the system are
treated as slaves to follow the reference signals coming from the master unit, as
shown in Figure 5.1. To improve the effectiveness of a P/EMS, it is also very
important to clearly define a system’s objectives and constraints. These
objectives (such as operating cost minimization, emission mitigation, power
loss reduction, etc.) together with the constraints might be conflicting in some
cases, which, in turn, makes the optimal decision-making process a difficult or
even impossible task. Different examples of centralized P/EMS for microgrids
can be found in literature [10,20–24]. As an illustration, an expert centralized
energy management approach is presented in [22] for optimal operation of a
Energy management systems for dc microgrids 93
Operator’s input, measurements,

objectives, constraints
Centralized P/EMS
Local controller Local controller Local controller
Energy source Energy source Energy source

(DG Unit 1) (DG Unit 2)
... (DG Unit n)
Fixed loads Utility Controllable loads

(grid)
Figure 5.1 Block diagram of a centralized P/EMS
given microgrid with different micro-sources (such as microturbines, fuel cells

and batteries) and renewables. In this work, the optimal management problem
is formulated as a nonlinear constraint multi-objective optimization problem to
minimize the total operating cost and the simultaneous net emissions inside the
microgrid.
The advantages of a centralized scheme mainly lie within the simplicity of
implementation and globality of optimal solution, however it brings two dis-
advantages: single point of failure which implies that a centralized P/EMS
must be securely designed with appropriate redundancy inbuilt, and massive
communication expenditure. The latter is not a challenging problem in small-
scale networks but could be problematic for larger systems as the complexity
of the centralized optimization grows exponentially with the number of units
(control variables) in the system.
● Distributed P/EMS
Distributed P/EMS is the second interactive scheme for management of
microgrids in which there is no central supervisory unit, but all the LCs are
connected and communicate with one another through a communication bus
[25,26]. In this way, each controller not only captures local measurements, but
also receives information from neighboring nodes which helps in decision-
making process according to different optimization objectives [27,28]. In this
Energy source Energy source ... Energy source

(DG Unit 1) (DG Unit 2) (DG Unit n)

(grid)
Figure 5.2 Block diagram of a decentralized P/EMS
scheme, intelligent algorithms are often used for better exploration/exploita-

tion of the environment in order to find the optimal operation point. Figure 5.2
shows the block diagram of a decentralized P/EMS.
A distributed scheme has some advantages over a centralized one. Firstly,
it supports a scalable structure with a plug-and-play feature for newly added/
removed energy sources or load blocks. Secondly, the computation burden of
each LC is mitigated which, in turn, reduces the required communication
bandwidth. Finally, a distributed P/EMS could improve the redundancy and
modularity of the system where this is needed. However, there is still a pro-
blem if a communication link fails in the system. This failure would not end in
a total system collapse, but the performance of the system would no longer
be optimal. Also, a distributed P/EMS suffers from degradation of performance
for small/medium networks, increased use of database space, and complex use
and administration. A multi-agent system (MAS) is one of the best illustrations
of a distributed scheme [28]. Considering a MAS-based structure, there are
several interacting intelligent agents that make decisions based on predefined
objectives within the environment. The agent’s intelligence may include some
methodical, functional, procedural approaches, algorithmic searching or rein-
forcement learning [29–32].
● Hybrid P/EMS
A hybrid scheme for power/energy management in microgrids can be realized
as another interactive structure that is based mainly on a combination of cen-
tralized and distributed schemes. In a hybrid structure, LCs, which are used for
operation management of different energy sources, are divided into groups
[33]. Within each group, a centralized scheme is used to control and optimize
the performance of LCs. On a higher level, a distributed scheme is utilized to
coordinate the operation of centralized controllers in different clusters for
global optimization [34]. Such a hybrid strategy can be seen in Figure 5.3.
Distributed control
Centralized control
Centralized P/EMS Centralized P/EMS
Local controller Local controller Local controller Local controller
Energy source Energy source ... Energy source Energy source

(DG Unit i) (DG Unit j) (DG Unit k) (DG Unit n)
Loads Utility Loads Utility

(grid) (grid)
Figure 5.3 Block diagram of a hybrid P/EMS
It is notable that a hybrid P/EMS scheme is normally implemented for large-

scale networks such as interconnected energy systems or microgrids, where the
optimal operation of the entire system depends on cooperation and coordination of
different control layers over a time period. Moreover, by doing this hybridization,
it is possible to improve system reliability and resiliency for long-run operations,
thanks to the unique features inherently existent in decentralized/decentralized
schemes [35,36].
5.2.1.2 Passive power/energy management strategies

Self-autonomy of operation for a LC without having information from neighboring
nodes is the main idea of a passive power/energy management scheme (PP/EMS).
In this structure, it is assumed that making an information sharing mechanism is
too costly or not viable, thus independent operation of energy sources is required.
Moreover, it is necessary to clearly define the control objective of each energy
source to assure reliable operation of the system. A block diagram for such a
power/energy management scheme is shown in Figure 5.4.
Among the existing methods for PP/EMS, droop-based control strategy is
regarded as a dominant method [37–43]. This control methodology adopts the
behavior of synchronous machines in responding to the changes in voltage and
frequency and applies similar rules in the operation management of converters in
ac/dc sides. The droop-based control strategy works based on the assumption that
the output impedance of a controllable unit (such as a micro-source) is mainly
inductive, and it utilizes the droop characteristics of voltage amplitude and
Energy source Energy source Energy source

...
(DG Unit 1) (DG Unit 2) (DG Unit n)

(grid)
Figure 5.4 Block diagram of a PP/EMS
frequency of each controllable unit to control its output [41]. In the case of a dc
microgrid, bus voltages, and in the case of an ac microgrid, the system voltage and
frequency are the information sensed by each local droop controller and used
subsequently to adjust output active (and/or reactive) power as a generation unit.
Since there is no communication requirement for fulfilling the control objectives,
this control strategy is highly reliable. Moreover, this control structure could be
easily extended to different energy sources while enabling true plug-and-play fea-
tures [39]. Apart from these benefits, there are several issues in such a power/
energy management strategy. First, nonlinear current sharing among parallel units
cannot be addressed directly by this method. Second, low X/R line impedance ratio
may result in active and reactive power coupling and instability issues in low-
voltage microgrid systems and cause power sharing errors for generation units
[44,45]. Recently, work has been carried out to improve the performance of a
conventional droop-based control method by implementing the droop in virtual
frames [44, 46], adding virtual impedance in control loops [47,48], or adjusting the
output voltage bandwidth [44]. However, without a coordinating unit, such as a
central controller (CC) or a system optimizer, it would be a challenging task to
optimally manage the operation of a microgrid system with PP/EMS.
As another type of PP/EMS, maximum power point tracking (MPPT) control
methodology is also applied in microgrids to maximize power extraction from
renewable energy sources (mainly wind turbines (WTs) and photovoltaic (PV)
solar systems) under all conditions [49,50]. In such a power management techni-
que, the unit’s voltage and current are sampled frequently, and the duty ratio of the
interfaced converter is adjusted accordingly. However, it should be noted that in
islanded renewable-based microgrids, which are controlled based on MPPT prin-
ciples, energy storage systems (ESSs) must also be dispatched to provide voltage
and frequency regulation services [51].
Considering the drawbacks of IP/EMS and PP/EMS, it seems that a combined

P/EMS structure (e.g. a consensus-based droop framework [52] or a droop-based
distributed cooperative control [53]) could not only address reliability issues, but
also enhance control performance of the system both in grid-connected and stand-
alone modes.
5.2.2 Control schemes

In order to guarantee stable and efficient operation of a dc microgrid, effective
control strategies should be developed. The structure of a dc microgrid control
system can be depicted as shown in Figure 5.5.
In general, a microgrid consists of several parallel converters that ideally work
in harmony. Local control functions of these converters typically cover the fol-
lowing: (1) current, voltage and droop control for each unit; (2) source-dependent
functions, e.g. MPPT for PV modules and WTs or a state of charge (SoC) esti-
mation for ESSs; (3) decentralized coordination functions, such as local adaptive
calculation of virtual resistances, distributed dc bus signaling (DBS) or power line
signaling (PLS).
At a global microgrid level, a dedicated communication-based coordinated
control can be implemented to achieve advanced energy management functions.
It can be realized either in a centralized or a distributed fashion, via a CC or sparse
communication network, respectively [28,38]. In case of a distributed control,
variables of interest are only exchanged between LCs. A consensus algorithm can
then be used to calculate either the average of all the variable values in distributed
LCs (DCLs) or the exact value of any variable present in a specific LC. Some of the
functionalities that can be accomplished by using DCLs include secondary/tertiary
control, real-time optimization, unit commitment and internal operating mode
changing (see Figure 5.6 for more details) [54].
Small wind PN
turbine Boost
converter
PV
AC
load
Battery
stack
DC
load
Figure 5.5 Single-bus dc microgrid with direct battery connection

Source side
PV Converter and filter stage Microgrid side
DC/DC Filter
Coordinated
WT Local control control
commands
Local consumers
AC/DC Converter to converter
communication DC/AC
DC/DC Filter iMG

ESS AC utility
Pulse width modulation

isource iL
Fuel cell vsource Droop Voltage/current vPCC
ωrot control control
iMG vPCC Rd1, …, Rdn
Solar irradiation Basic functions
Wind speed
Charging SoC DCL-based coordinated control
MPPT
/discharging algorithm estimation Consensus Real time
Source depending functions Coordinated algorithm
optimization
control Unit
commands commitment
Adaptive Rd DC Bus Power line Secondary/
calculation signaling communication Operating
modes tertiary control
Decentralized coordination
Local control
Figure 5.6 General control structures in dc microgrids

Decentralized comm. network
LC 1 LC 2 LC n
…
DG 1 DG 2 DG n
Physical connection
Centralized comm. network Distributed comm. network
CC
LC 1 LC 2 LC n LC 1 LC 2 LC n
… …
DG 1 DG 2 DG n DG 1 DG 2 DG n
Physical connection Physical connection
Figure 5.7 Differences between decentralized, centralized and distributed

control architectures
From the communication perspective, overall control of dc microgrids can be

divided into the following three categories:
● Decentralized control: DCLs do not exist and power lines are used as the only
channel of communication.
● Centralized control: data from distributed units are collected in a centralized
aggregator, processed and feedback commands are sent back to them via
DCLs.
● Distributed control: DCLs exist but are implemented between units and coor-
dinated control strategies are processed locally.
The basic configuration of each of these control structures is depicted in Figure 5.7.
A more detailed overview of the significant features of local and coordinated
control strategies is provided in the following sections.
5.2.2.1 Local control functionalities

As previously mentioned, the control framework of a dc microgrid consists of local
and coordinated control levels. In this section, a local control level is discussed in
detail. Basic functions which include current, voltage and droop control are
reviewed. A review of MPPT and charging algorithms has been omitted here. More
details on charging algorithms for batteries can be found in [55], while an in-depth
analysis of MPPT algorithms has been presented in a number of references, e.g. in
[56]. As a backbone of a dc microgrid, the interface converters play an important
role in efficient and reliable operation of the overall system. In order to ensure not
only proper local operation, but also to enable coordinated interconnection between
Local terminal n
+
Power
DG n
– converter n
vdci Power-based droop
+
Power mp2
DG 2 2 vdc*
– converter 2
s
bu
ad
+
mp1
Lo
Power
DG 1
converter 1 1
–
vo1 io1 Po1 Po2 Po
PWM
Power calculation Low-pass
d1 + low-pass filter filter
vdci
Current Po1 Io1 Current-based droop
control Inner loops Power-based Current-based vdc* mc2
– droop droop
iin1 – iin1*
Voltage
control Droop controller mc1
vo1*
Converter control 1
Io1 Io2 Io
Figure 5.8 Inner control loops for dc microgrids
different modules in a dc microgrid, flexible local current and voltage control

should be employed and accurate power sharing among parallel connected con-
verters should be achieved.
The basic local control diagram is shown in Figure 5.8, including local current
and voltage controllers, and a droop control loop. For local dc current and voltage
control systems in dc microgrids, proportional–integral (PI) controllers are com-
monly used since they introduce zero steady-state error, can be easily tuned and are
highly robust. However, the use of other types of controllers, such as proportional
derivative (PD), fuzzy and boundary controllers, has also been reported [57–59].
PD controllers can be used to improve the phase margin (PM) of the system, but
they do not eliminate steady-state error and, also, need to have high-frequency
poles to attenuate the high-frequency noise. Hence, rather than appearing in a pure
PD form, the derivative term in a PD controller is usually replaced by a high-pass
digital filter. By combining the beneficial effects of PI and PD controllers, pro-
portional-integral-derivative (PID) controllers can be employed. Fuzzy control is
designed to emulate a human being’s conclusion deduction process based on the
stimulus he/she gets from the environment and his/her own embedded knowledge.
In the engineering world, this can be defined as a knowledge-based control method
that can simultaneously take advantage of both static and dynamic properties of the
system. It has therefore good adaptability in environments with varying conditions.
For the purpose of local voltage and current regulation, fuzzy controllers can
either be used as principal regulators that process the error signal or in a series with
feedback loops. To ensure fast convergence and extreme robustness, nonlinear
control strategies based on state-dependent switching can be employed. They pre-
sent simple implementation, but their detailed performance analysis can be quite
complex. It should be noted that alternative control methods for dc microgrids have
recently drawn a lot of attention in academic circles. However, their practical
application should be elaborately justified by performing modeling, analysis,
simulation, implementation, as well as a full cost-benefit analysis. For instance,
increased production cost and lead time often proves to be too large an obstacle for
their deployment.
Droop control is commonly installed on top of inner loops, primarily for cur-
rent sharing purposes. Figure 5.8 demonstrates that either output power or output
current can be selected as the feedback signal in droop control [60,61]. For dc
microgrids with a power-type load, output power can be used as droop feedback, as
shown in (1). On the other hand, when current signal is used, as shown in (2), droop
coefficient mc can be regarded as a virtual internal resistance. In this case, the
implementation and design of the parallel converter system in a dc microgrid can
be simplified to some extent as the control law is linear [60]. The principle of
current-based droop control was also extensively used in distributed power systems
for paralleling multiphase converters that supply computer CPUs. Here, droop
control is commonly known as adaptive voltage positioning [62–64]. The calcula-
tions of references for voltage controller, in the two aforementioned cases, are as
follows:
vdci ¼ vdc mp Poi (5.1)

vdci ¼ vdc mc ioi (5.2)
where vdci is the output of the droop controller, i.e. the reference value of the dc
output voltage of converter i, vdc is the rated value of the dc voltage; mp and mc are
the droop coefficients in power- and current-based droop controllers, while Poi and
ioi are the output power and current, respectively, of converter i.
5.2.2.2 Coordinated control

Although the local interface converter control is an essential part of a dc microgrid
which provides regulation of physical quantities, coordinated control is needed to
achieve extended objectives. As already mentioned, depending on the means of
communication between the interface converters, coordinated control should be
implemented to achieve an intelligent control system with extended objectives.
● Centralized coordination
Centralized coordination control can be implemented in dc microgrids by
employing a CC and a dedicated communication network to connect it with
sources and loads, as shown in the DCL-based coordination control window of
Figures 5.6 and 5.7. For small-scale dc microgrids, each unit can be directly
controlled by the CC that employs a high bandwidth communication using a
master/slave approach [65,66]. However, for larger scale dc microgrids, hier-
archical control is often a preferred choice since it introduces a certain degree
of independence between different control levels. This control is more reliable
as it can be designed to continue the operation even in a case of failure of
centralized control. Hierarchical control is achieved by simultaneously using

local converter control and DCL-based coordinated control, which are sepa-
rated by at least an order of magnitude in the control bandwidth [60]. Coor-
dinated functions can include secondary/tertiary regulation of the dc voltage,
power flow control and different grid-interactive control objectives such as
unit commitment, changing operating modes, global optimization aimed at
maximizing efficiency, minimizing operating cost, etc.
It should be noted that centralized control provides the best foundation for
the employment of advanced control functionalities since all relevant data can
be collected and processed in a single controller. However, the most obvious
disadvantage of this strategy is that it has a single point of failure. In particular,
if the CC or any key communication link fails, the commands from/to the
controller will not be transmitted and corresponding control objectives will
likely not be achieved. For mission critical applications, redundant commu-
nication systems can be installed in order to reduce the possibility of failure,
but this needs to be justified by a cost–benefit analysis. Another option to
increase the reliability of the system is to combine decentralized and cen-
tralized control methods into a hierarchical control structure [67–72]. In this
case, basic functions of the dc microgrids can be retained even if the cen-
tralized controller fails.
● Decentralized coordination
Decentralized coordination strategies are achieved exclusively by LCs, as
shown in Figures 5.6 and 5.7. The most common types of decentralized control
are DBS, adaptive adjustment of droop coefficients and PLS. While their
advantage is in the simplicity of control and independence from dedicated
communication technology, they inherently have performance limitations
due to lack of information from other units. Moreover, as these methods are
invariably based on the interpretation of the voltage in the common dc bus, the
accuracy of voltage sensors impacts their effectiveness and reliability.
Originally proposed in [73] and the follow-up in [74], DBS is the most
prominent decentralized coordination method for dc microgrids. By using the
DBS approach, coordinated operation of different units in dc microgrids is
realized by imposing and identifying variations in the common dc bus voltage.
The transitions between different modes are triggered by different preset dc
bus voltage values.
The method of adaptive droop coefficient calculation has been mainly used
for power balancing of distributed ESSs, as shown above. The main limitation
of the method is the potential instability induced by improperly designed droop
curves. To that end, there always exists a trade-off between the permissible
voltage deviation and stability properties of the system, i.e. higher voltage
deviation is associated with the higher PM. It should also be noted that some of
the SoC balancing methods mentioned above [75,76] were originally proposed
for ac microgrids. However, since the same principle can apply to dc micro-
grids (in ac microgrids, the frequency is normally used as a system level
coordinating signal), they are also included here.
PLS is another decentralized method that can be deployed for coordinated

control. In particular, sinusoidal signals of a specific frequency are injected
through amplifiers into the dc bus, allowing each device to send and receive
information on its status, performance, history or internal operational mode.
Although PLS relies on dedicated communication, here it is categorized as
decentralized since the power network is the only communication medium. It
should be noted that in power systems’ literature, this particular way of com-
munication is sometimes also referred to as PLC [77]. REbus, an open standard
for dc electricity distribution in homes, commercial buildings, campuses and
other settings, uses PLS as a primary communication carrier [78]. Alternative
methods that exchange information between devices without using dedicated
amplifiers were presented in [79, 80]. Signals are generated by the PWM of
dc–dc converters in those works. They are injected in open loop in [80],
whereas dedicated proportional-resonant controllers are used during injection
periods in [79] to avoid the steady-state error in the dc bus [81].
PLS is generally more complex to implement compared to other decen-
tralized methods. Moreover, it is commonly only used for changing operating
modes or shutting down corrupted components of the system, and it is not
suitable for power sharing. However, as opposed to permanent voltage devia-
tion in the common dc bus which is inherent for DBS and adaptive droop
methods, sinusoidal signals are only periodically injected into the system.
Therefore, the quality of the voltage waveform can be considered, to some
extent, improved compared to other methods.
● Distributed coordination
Distributed control indicates the control principle where a central control unit
does not exist and LCs communicate only among themselves through dedi-
cated DCLs, as shown in Figure 5.7. The main advantage of this approach is
that the system can maintain full functionality, even if the failure of some
communication links occurs, provided that the communication network
remains connected.1 Therefore, distributed control is immune to a single point
of failure. The functionalities that can be achieved by this approach resemble
those of centralized control and are also represented in the DCL-based coor-
dination control window of Figure 5.6. However, in order to enable these
functionalities, the information exchanged through DCLs first needs to be
appropriately processed. In particular, information directly exchanged between
LCs can contain only locally available variables. In other words, if the two
units are not connected by a DCL, they do not have direct access to each
other’s data and their observation of the system is quite limited. In order to
circumvent this problem and to make the level of awareness of an LC similar
to that of a CC, a consensus algorithm can be used. In its basic form, a con-
sensus algorithm is a simple protocol installed within every LC which con-
tinuously adds up all algebraic differences of some variable(s) of interest
1
For the exact definition and a more in-depth discussion of the connectivity of communication net-
works, please refer to [83].
present in a given LC and those present in LCs adjacent to it. If we look at

LC i, this definition can also be expressed by the following equation:
X
x_ i ðtÞ ¼ xi ðtÞ xj ðtÞ þ bi ðtÞ (5.3)
j2Ni
where xi(t) and xj(t) are the values of variables of interest in LC i and LC j,
respectively. Here, j is iterated through all of the neighbors of LC i, represented
by Ni. Finally, bi(t) is an optional input bias of LC, which can be used to
declare it as a virtual leader. It can be seen from (5.3) that xi(t) is interactively
adapted with respect to the values of its neighboring units. Likewise, variables
in any other controller adapt with respect to the values of their own neighbors.
Consequently, it can be analytically proved that, if the communication network
is connected, all variable values will converge to a common average after a
certain amount of time [82,83]. Another option is to use a nonzero input bias in
one of the LCs. In that case, variables of all other LCs will converge to their
respective bias [84]. In either case, the ability of a consensus to share infor-
mation in such a manner has wider applicability than simple data averaging.
For instance, if every LC has information on the number of other active LCs,
an exact value of any specific variable can be calculated directly from the
average.
The collective dynamics of a communication system realized via consensus
protocol can be represented by the following equation:
x_ ðtÞ ¼ Q xðtÞ (5.4)
where Q ¼ [qij] is the graph Laplacian of the network whose elements are
defined as follows:

1 j 2 Ni
qij ¼ (5.5)
jNi j j ¼ i
where |Ni| denotes the number of neighbors of node i. The topology of the
communication network is explicitly reflected by graph Laplacian, and it is
also possible to design weights of the respective matrix to control the con-
vergence speed [85]. Figure 5.9 shows the configurations of an exemplary
power electronics-based microgrid and its sparse communication network
In summary, it can be concluded that distributed control can achieve information
awareness comparable to that of centralized control. Therefore, objectives such as
output current sharing, voltage restoration, global efficiency enhancement, SoC
balancing and others can be easily realized. In this sense, distributed control offers
much wider functionalities than decentralized control, but remains protected from
the single point of failure. Its main limitation is the complexity of analytical per-
formance analysis, i.e. assessment of convergence speed and stability margins,
Converter n
Converter i
Converter 2
Converter 1
Converter n
.
.
. DC
microgrid Converter i
Converter 2
Converter 1
Figure 5.9 Configurations of exemplary physical and communication networks
especially in nonideal environments characterized by communication time delays

and measurement errors.
5.3 Interfacing converter control strategies for power/energy

management purposes
It was previously discussed that the primary goal of a power/EMS is to determine
the optimal operating point of controllable units (such as active/reactive power
references) with regard to the system’s objective(s) and constraints (e.g. device
technical specifications, permissible voltage and frequency deviations, demand–
supply balance, etc.). To effectively track optimal set-points (reference signals)
coming from a system-level optimizer while regulating current and the voltage,
advanced control strategies must be applied in interfacing converters. These control
strategies are generally categorized as voltage control mode (VCM) (also known
as grid-forming mode) and current control mode (CCM) (also known as grid-
following mode) [86]. In the next section, a brief description is presented for each
of the interfacing converter control modes.
5.3.1 Voltage control/grid-forming mode

By mimicking a synchronous generator/condenser behavior for adjusting the grid’s
voltage, VCM can be applied for a converter-based DG unit to regulate the output
power of the unit and reach the same objective. Droop control concepts can also be
used for efficient implementation of such control strategy. Let us consider the
problem of complex power flow through a line as shown in Figure 5.10. Without
loss of generality, the line could be considered as an R–L circuit (RþjX) with the
vi = E∠φ
Zl = Z∠θ vi = V∠0
Grid
Sij = Pij + jQij
DG unit PCC
Figure 5.10 Equivalent circuit of DG unit connected to the common ac bus
voltages at the terminals of the line being held constant. Thus,

vi vj E
Sij ¼ Pij þ jQij ¼ vi ¼ jq E Vejj (5.6)
Zl Ze
Considering that a converter-based DG unit is connected through a highly inductive

line to a node and the phase angle j between the converter and the point of cou-
pling is small, it can be concluded that
EV X Pij
Pij ffi sinðjÞ ! j ffi (5.7)
X EV
E2 E V X Qij
Qij ffi cosðjÞ ! E V ffi (5.8)
X X E
meaning that the active and reactive power outputs of a given DG unit are propor-
tional to the phase angle difference j and the voltage magnitude difference (EV),
respectively. In other words, by controlling the real power, it would be possible to
adjust the power angle and, through reactive power control, voltage regulation can be
achieved. In a droop-based control, the same procedure is adopted with the exception
that each unit uses the frequency (instead of the power angle or phase angle), for
controlling active power flow.
Figure 5.11 shows the block diagram of the VCM control strategy which can
be applied in both grid-tied and islanded operating modes of microgrids and makes
the operation mode transition easy and smooth. As observed from the same figure,
active and reactive power references are provided by the P/EMS to meet certain goals
such as voltage profile smoothing. The lower level controllers embedded in the VCM
structure (named as active and reactive power controllers in Figure 5.11) assure that
the output voltage phase angle and magnitude are of the right orders and that they can
be designed as proportional controllers for realizing active power–frequency droop
(P–w) and reactive power–voltage magnitude droop (Q–E) as follows:

w ¼ wref m Pmeas Pref (5.9)

V ¼ Vref n Qmeas Qref (5.10)
In this strategy, inner current loops can be assigned to the voltage closed-loop
control system to improve the transient and stability performance [87].
Vref + Iref
Voltage + Current
_ controller _ controller
Vmeas Imeas
Pref + Active Space PCC

ω 1 θ
power vector
_ controller S Vref
+ Voltage
controller PWM ZL Grid
_
Pmeas
abc
Qref + Reactive V Vmeas
power
_ controller
Qmeas
Figure 5.11 Voltage-based real and reactive power control strategy
Vdc,ref
+ DC voltage Id,ref
_ controller
Vdc,meas
PCC
Pref + Active Id,ref + Current
power _ controller dq
_ controller PWM ZL Grid
abc
Pmeas
Qref Reactive Iq,ref +

+ Current θ
power controller PLL VPCC
_ controller _
Qmeas Id
dq
Ig
abc
Iq
Figure 5.12 CCM/grid-following control strategy
More complex controllers can also be integrated into the VCM to closely mimic
the behavior of a synchronous machine equipped with governor and excitation
systems [88].
5.3.2 Current control/grid-following mode

Figure 5.12 shows the schematic of a current control/grid-following control strat-
egy for power flow management. As can be observed from the figure, there are
several current-based controllers in a CCM that track and regulate the active and
reactive powers in the closed-loop manner. The outer active and reactive power
control loops are in charge of producing d-axis and q-axis reference currents (based
on the received set-points), respectively. It should be noted that the reference power
signals can be generated by the system optimizer (such as an EMS) or through

any other control system such as a MPPT unit. It is also noteworthy that for
synchronizing the interfacing inverter output current with the grid voltage in grid-
connected applications, CCM control strategy uses the grid voltage angle infor-
mation coming from a phase-locked-loop block. In dc networks where there is no
reactive power flow, reference current signals (only available in d-axis) are gen-
erated using a dc voltage controller. In this regard, CCM is mainly executed
through the upper control leg shown in Figure 5.12. The same procedure can be
adopted in cases where two-stage power conversion is needed (such as dc/dc–dc/ac
or ac/dc–dc/ac). In this situation, the first level of conversion is controlled by an
active power controller, while the next level is controlled through a dc voltage
controller in which the power difference between the inverter output and the input
stage can be used to charge or discharge the dc link capacitor [89]. Taken as a
whole, CCM/grid-following mode is mainly used for power flow control purposes
in grid-connected conditions where the utility (i.e. main grid) plays the key role
in supporting the system frequency and regulating the common bus voltage while
other controllable units can be operated/dispatched based on an optimal control
strategy enforced by the CCM. On the other hand, the CCM-based power control
mode cannot be used as the only control strategy in an off-grid or islanded opera-
tion of microgrids as it is not capable of regulating the microgrid voltage and
frequency in different working conditions. Thus, complementary actions based on
VCM control strategy are needed for at least one controllable generation unit or
energy storage option.
5.4 Illustrative example
A block diagram of a given residential microgrid system is shown in Figure 5.13.

As can be seen in the figure, the main actors on the generation side include the
utility (power grid) and the controllable DGs such as micro-combined cogeneration
systems micro-combined heat and power (mCHP), while on the demand side,
schedulable appliances serve as actors.
In this structure, the leading actor is the residential smart EMS (RSEMS) that
manages the operation of in-home devices and appliances optimally in coordination
with the microgrid CC taking into account users’ objectives and existing technical/
operational constraints. RSEMS also provides required information such as set-
points and reference signals for LCs when is needed. While the RSEMS is designed
as a soft real-time system to improve and optimize system operation during each
decision period, lower level controllers at the supply/demand sides are modeled as
hard real-time systems to maintain stability. The two aforementioned systems have
different time granularities according to types of controlled objects and perfor-
mance requirements. In the examined microgrid, sustainable energy sources such
as PV systems and WTs are exploited as nondispatchable prime movers of the
mentioned system. The power grid is also used as a sustained power supply to
support required point of common coupling (PCC) voltage regulation and system
Residential microgrid
Energy Information
RSEMS
Smart Home
meter scheduler Local
controllers
Distributed MGCC
generation
Local controller
µCHP
EV Water tank Loads
Radiant floor heating/cooling system

Residential buildings
PCC
Power grid
Figure 5.13 Integrated building and microgrid system
frequency. Other dispatchable dispersed generators within the residential section

adjust their behaviors according to the power demand and reference signals
received from RSEMS [90].
The lower level control structure (here named as the ‘‘master-slave’’) of the
existing DG sources can be implemented in a way depicted in Figure 5.14. Based
on this control structure, each controllable DG is treated as a ‘‘grid-following’’ unit
and operated in CCM/power control mode to follow a power reference (P*), while
the utility (which is the master unit) is operated in grid-forming mode (VCM) to
maintain the voltage and the frequency within the range. In this regard, droop
control is implemented based on (5.11) and (5.12) in the primary loop for the
master unit, and the reverse droop controller is utilized for the slave units as
expressed in (5.13):
w ¼ w md P (5.11)
E ¼ E nd Q (5.12)

1
Q ¼ ðE Eg Þ (5.13)
nr
PCC
Master unit (VCM) iL Lin Lo i
o
Vdc
ω C
Vc
Vref + Iref + dq d
PI _ PI PWM
_ abc
abc ω
Fixed
dq
loads
abc ω
Inner-loop control dq
E_*
ON/OFF
Qmaster q Vc
Voltage E + nd PQ
LPF Power
reference Pmaster io
θ md p calculation
generator 1/S _ LPF
+
ω*
Droop control
Schedulable
load
Slave unit (PCM) iL Lf Vg
Vdc
ωg
Vg
Iref + + + dq d
_ PI PWM
abc
abc
dq ωg
PLL
Inner-loop control
ωg
Vgd Vgq E*
Current Qslave + _ Eg E'g
1/nr LPF
reference
RSEMS
generator Pslave = P*
Reverse droop control
Figure 5.14 Master–slave control of distributed energy sources
It should be mentioned that within the mentioned structure, the inner loop control
of the VCM aims to achieve good output voltage regulation with respect to a
determined capacitor voltage reference while the inner loop control of the PCM
regulates the output power. For both inner loops, typical PI controllers are appro-
priately designed and used. In the primary control loops, low-pass filters are also
applied to limit the loop bandwidth, so that the primary control can be separately
designed and the inner loop of the VCM and CCM can be considered as ideal
voltage and current sources, respectively.
5.5 Conclusions
With the application of dc microgrids at different scales and topologies, it is pos-
sible to integrate different energy sources (such PV systems, fuel cells and bat-
teries) into the energy mix of a larger system with lower conversion requirements.
Moreover, it is much easier and cheaper to convert ac power into dc power which,
in turn, helps economic operation of future energy systems. By using dc microgrids,
it is also possible to achieve energy savings up to a great extent (~15%) and to
improve system reliability by reducing the number of devices required and the total
points of failure. The dc architecture also enables cost-effective and green solutions
for operation and control of zero-net energy residential/office buildings as well as
data centers. However, such optimal performance mainly depends on the proper
design and application of EMSs which effectively manage the process of energy
production and consumption based on predefined objectives and constraints. There
are, definitely, a number of challenges in this regard which have to be suitably
addressed. Currently, there is a lack of approved standards and technical codes for
dc equipment and distribution networks at low voltage. There is also a lack of
approved and recognized dc architectures at low-to-medium voltage levels which in
turn necessitates different safety and protection practices in comparison with con-
ventional ac systems. Last but not least, there is a strong need for upgrading the
existing infrastructure to accommodate dc systems and interfaces.
References
[1] American Physical Society, ‘‘Buildings,’’ in ENERGY FUTURE: Think
Efficiency. American Physical Society, 2008, pp. 52–85.
[2] International Energy Agency, Transition to Sustainable Buildings – Strategies
and Opportunities to 2050. Paris: OECD/IEA, 2013.
[3] N. Dempsey, C. Barton, and D. Hough, ‘‘Energy Prices,’’ in The House of
Commons Library, 2016, no. 4153, pp. 1–18.
[4] A. Mousavi, C. W. Yang, C. Pang, and V. Vyatkin, ‘‘Energy efficient
automation model for office buildings based on ontology, agents and IEC
61499 function blocks,’’ in 19th IEEE Int. Conf. Emerg. Technol. Fact.
Autom. ETFA 2014, 2014, pp. 1–7.
[5] C. Delmastro, G. Mutani, M. Pastorelli, and G. Vicentini, ‘‘Urban mor-
phology and energy consumption in Italian residential buildings Urban
morphology and energy consumption in Italian residential buildings,’’ in 4th
International Conference on Renewable Energy Research and Applications,
2015, vol. 5, no. Jan. 2016, pp. 1603–1608.
[6] S. I. Symposium, ‘‘Simulation-based optimization in energy efficiency ret-
rofit for office building,’’ in Proc. 2014 IEEE/SICE Int., 2014, pp. 222–227.
[7] K. Vatanparvar, Q. Chau, A. Faruque, et al., ‘‘Home energy management as
a service over networking platforms,’’ in IEEE PES Conference on Innova-
tive Smart Grid Technologies (ISGT), Washington, DC, 2015, pp. 1–5.
[8] A. Gligor, H. Grif, and S. Oltean, ‘‘Considerations on an intelligent buildings
management system for an optimized energy consumption,’’ in 2006 IEEE
Int. Conf. Autom. Qual. Testing, Robot. AQTR, 2006, no. 1, pp. 280–284.
[9] A. Anvari-Moghaddam, H. Monsef, and A. Rahimi-Kian, ‘‘Optimal Smart
Home Energy Management Considering Energy Saving and a
Comfortable Lifestyle,’’ IEEE Trans. Smart Grid, vol. 6, no. 1, pp. 324–332,
2015.
[10] A. Anvari-Moghaddam, H. Monsef, and A. Rahimi-Kian, ‘‘Cost-Effective
and Comfort-Aware Residential Energy Management under Different Pricing
Schemes and Weather Conditions,’’ Energy Build., vol. 86, pp. 782–793,
2015.
[11] Y. Li and F. Nejabatkhah, ‘‘Overview of Control, Integration and Energy
Management of Microgrids,’’ J. Mod. Power Syst. Clean Energy, vol. 2,
no. 3, pp. 212–222, Aug. 2014.
[12] A. Ghasemkhani, A. Anvari-Moghaddam, J. M. Guerrero, and B. Bak-
Jensen, ‘‘An efficient multi-objective approach for designing of commu-
nication interfaces in smart grids,’’ in IEEE PES Innovative Smart Grid
Technologies (ISGT 2016), 2016, pp. 1–6.
[13] S. Z. Islam, N. Mariun, H. Hizam, et al., ‘‘Communication for distributed
renewable generations (DRGs): A review on the penetration to smart grids
(SGs),’’ in PECon 2012 – 2012 IEEE Int. Conf. Power Energy, 2012,
no. December, pp. 870–875.
[14] K. M. Liyanage, M. A. M. Manaz, A. Yokoyama, Y. Ota, H. Taniguchi, and
T. Nakajima, ‘‘Impact of communication over a TCP/IP network on the
performance of a coordinated control scheme to reduce power fluctuation
due to distributed renewable energy generation,’’ in 2011 6th Int. Conf. Ind.
Inf. Syst. ICIIS 2011 – Conf. Proc., 2011, pp. 198–203.
[15] A. Anvari-Moghaddam, G. Mokhtari, and J. M. Guerrero, ‘‘Coordinated
Demand Response and Distributed Generation Management in Residential
Smart Microgrids,’’ in Energy Management of Distributed Generation
Systems, 1st ed., Dr. Eng. L. Mihet, Ed. InTech, 2016, pp. 27–57.
[16] M. Sechilariu, B. Wang, and F. Locment, ‘‘Power management and optimi-
zation for isolated DC microgrid,’’in 2014 International Symposium on
Power Electronics, Electrical Drives, Automation and Motion, Ischia, 2014,
pp. 1284–1289.
[17] F. Nejabatkhah and Y. W. Li, ‘‘Overview of Power Management Strategies
of Hybrid AC/DC Microgrid,’’ IEEE Trans. Power Electron., vol. 30, no. 12,
pp. 7072–7089, 2015.
[18] D. E. Olivares, S. Member, and C. A. Cañizares, ‘‘A Centralized Energy
Management System for Isolated Microgrids,’’ IEEE Trans. Smart Grid,
vol. 5, no. 4, pp. 1864–1875, 2014.
[19] L. E. Zubieta, ‘‘Power Management and Optimization Concept for DC
Microgrids,’’ in 2015 IEEE First International Conference on DC Micro-
grids (ICDCM), 2015, pp. 81–85.
[20] M. Parvizimosaed, F. Farmani, and A. Anvari-Moghaddam, ‘‘Optimal Energy
Management of a Micro-Grid with Renewable Energy Resources and Demand
Response,’’ J. Renewable Sustainable Energy, vol. 5, no. 5, pp. 31–48, 2013.
[21] A. Anvari-Moghaddam, A. Seifi, and T. Niknam, ‘‘Multi-Operation Man-
agement of a Typical Micro-Grids Using Particle Swarm Optimization:
A Comparative Study,’’ Renewable Sustainable Energy Rev., vol. 16, no. 2,
pp. 1268–1281, 2012.
[22] A. Anvari-Moghaddam, A. Seifi, T. Niknam, and M. Pahlavani, ‘‘Multi-

Objective Operation Management of a Renewable MG (Micro-Grid) with
Back-up Micro-Turbine/Fuel Cell/Battery Hybrid Power Source,’’ Energy,
vol. 36, no. 11, pp. 6490–6507, 2011.
[23] Q. Jiang, M. Xue, G. Geng, ‘‘Energy Management of Microgrid in Grid-
Connected and Stand-Alone Modes,’’ IEEE Trans. Power Syst., vol. 28,
no. 3, pp. 3380–3389, 2013.
[24] D. J. Becker and B. J. Sonnenberg, ‘‘DC microgrids in buildings and data
centers,’’ in Telecommunications Energy Conference (INTELEC), 2011
IEEE 33rd International, 2011, pp. 1–7.
[25] A. Tani, M. B. Camara, and B. Dakyo, ‘‘Energy Management in the
Decentralized Generation Systems based on Renewable Energy—
Ultracapacitors and Battery to Compensate the Wind/Load Power Fluctua-
tions,’’ IEEE Trans. Ind. Appl., vol. 51, no. 2, pp. 1817–1827, 2015.
[26] Z. Wang, B. Chen, J. Wang, and J. Kim, ‘‘Decentralized Energy Manage-
ment System for Networked Microgrids in Grid-Connected and Islanded
Modes,’’ IEEE Trans. Smart Grid, vol. 7, no. 2, pp. 1097–1105, 2016.
[27] G. Mokhtari, A. Anvari-Moghaddam, and G. Nourbakhsh, ‘‘Distributed
Control and Management of Renewable Electric Energy Resources for
Future Grid Requirements,’’ in Energy Management of Distributed Gen-
eration Systems, 1st ed., no. Jul., Dr. Eng. L. Mihet, Ed. InT, 2016, pp. 1–24.
[28] A. Anvari-Moghaddam, J. J. M. Guerrero, A. Rahimi-Kian, and M. S. Mirian,
‘‘Optimal real-time dispatch for integrated energy systems: An ontology-based
multi-agent approach,’’ in 7th International Symposium on Power Electronics
for Distributed Generation Systems (PEDG’16), 2016, pp. 1–7.
[29] C. Dou, D. Yue, S. Member, X. Li, and Y. Xue, ‘‘MAS-Based Management
and Control Strategies for Integrated Hybrid Energy System,’’ IEEE Trans.
Industrial Informatics, vol. 12, no. 4, pp. 1332–1349, 2016.
[30] J. Yu, S. Member, C. Dou, and X. Li, ‘‘MAS-Based Energy Management
Strategies for a Hybrid Energy Generation System,’’ IEEE Trans. Industrial
Electronics, vol. 63, no. 6, pp. 3756–3764, 2016.
[31] Y. S. F. Eddy, S. Member, H. B. Gooi, S. Member, and S. X. Chen,
‘‘Multi-Agent System for Distributed Management of Microgrids,’’
IEEE Trans. Power Systems, vol. 30, no. 1, pp. 24–34, 2015.
[32] M. Mao, P. Jin, and N. D. Hatziargyriou, ‘‘Multiagent-Based Hybrid Energy
Management System for Microgrids,’’ IEEE Trans. Sustainable Energy,
vol. 5, no. 3, pp. 938–946, 2014.
[33] C. X. Dou, Z. S. Duan, and B. Liu, ‘‘Two-Level Hierarchical Hybrid Control
for Smart Power System,’’ IEEE Trans. Autom. Sci. Eng., vol. 10, no. 4,
pp. 1037–1049, 2013.
[34] C. Dou, X. Jia, Z. Bo, D. Liu, and F. Zhao, ‘‘Hybrid Control for Micro-Grid
based on Hybrid System Theory,’’ in 2011 IEEE Power and Energy Society
General Meeting, San Diego, CA, pp. 1–11, 2011.
[35] T. Ma, M. H. Cintuglu, S. Member, and O. A. Mohammed, ‘‘Control of
Hybrid AC/DC Microgrid Involving Energy Storage and Pulsed Loads,’’
IEEE Trans. Ind. Appl., vol. 53, no. 1, pp. 567–575, 2017.
[36] T. Lu, Z. Wang, Q. Ai, and W.-J. Lee, ‘‘Interactive Model for Energy
Management of Clustered Microgrids,’’ IEEE Trans. Ind. Appl., vol. 53,
no. 3, pp. 1739–1750, 2017.
[37] E. Rodriguez, J. C. Vasquez, M. Josep, et al., ‘‘Multi-level energy manage-
ment and optimal control of a residential DC microgrid,’’ in 2017 IEEE
International Conference on Consumer Electronics (ICCE), Las Vegas, NV,
2017, pp. 312–313.
[38] A. Anvari-moghadam, Q. Shafiee, J. C. Vasquez, and J. M. Guerrero,
‘‘Optimal adaptive droop control for effective load sharing in AC micro-
grids,’’ in 42nd Annual Conference of the IEEE Industrial Electronics
Society (IECON’16), 2016, pp. 2–7.
[39] A. Tah and D. Das, ‘‘An Enhanced Droop Control Method for Accurate
Load Sharing and Voltage Improvement of Isolated and Interconnected DC
Microgrids,’’ IEEE Trans. Sustainable Energy, vol. 7, no. 3, pp. 1194–1204,
2016.
[40] P. H. Huang, P. C. Liu, W. Xiao, and M. S. El Moursi, ‘‘A Novel Droop-
Based Average Voltage Sharing Control Strategy for DC Microgrids,’’ IEEE
Trans. Smart Grid, vol. 6, no. 3, pp. 1096–1106, 2015.
[41] A. Maknouninejad, Z. Qu, F. L. Lewis, and A. Davoudi, ‘‘Optimal, Non-
linear, and Distributed Designs of Droop Controls for DC Microgrids,’’
IEEE Trans. Smart Grid, vol. 5, no. 5, pp. 2508–2516, 2014.
[42] A. P. N. Tahim, D. J. Pagano, E. Lenz, and V. Stramosk, ‘‘Modeling and
Stability Analysis of Islanded DC Microgrids under Droop Control,’’ IEEE
Trans. Power Electron., vol. 30, no. 8, pp. 4597–4607, 2015.
[43] K. J. Bunker and W. W. Weaver, ‘‘Multidimensional Droop Control for
Wind Resources in dc Microgrids,’’ IET Gener. Transm. Distrib., vol. 11,
no. 3, pp. 657–664, 2017.
[44] Y. Li and Y. W. Li, ‘‘Power Management of Inverter Interfaced Autonomous
Microgrid based on Virtual Frequency-Voltage Frame,’’ IEEE Trans. Smart
Grid, vol. 2, no. 1, pp. 18–28, 2011.
[45] Y. W. Li and C. N. Kao, ‘‘An Accurate Power Control Strategy for Power-
Electronics-Interfaced Distributed Generation Units Operating in a Low-
Voltage Multibus Microgrid,’’ IEEE Trans. Power Electron., vol. 24, no. 12,
pp. 2977–2988, 2009..
[46] Y. Guan, J. M. Guerrero, X. Zhao, J. C. Vasquez, and X. Guo, ‘‘A New
Way of Controlling Parallel-Connected Inverters by Using Synchronous-
Reference-Frame Virtual Impedance Loop – Part I: Control Principle,’’
IEEE Trans. Power Electron., vol. 31, no. 6, pp. 4576–4593, 2016.
[47] J. Matas, M. Castilla, L. G. de Vicuña, J. Miret, and J. C. Vasquez, ‘‘Virtual
Impedance Loop for Droop-Controlled Single-Phase Parallel Inverters Using
a Second-Order General-Integrator Scheme,’’ IEEE Trans. Power Electron.,
vol. 25, no. 12, pp. 2993–3002, 2010.
[48] Y. Zhang and Y. Wei Li, ‘‘Energy Management Strategy for Supercapacitor
Virtual Impedance,’’ IEEE Trans. Power Electron., vol. 32, no. 4, pp.
2704–2716, 2017.
[49] I. Tank, ‘‘Renewable based DC microgrid with energy management sys-

tem,’’ pp. 1–5, 2015.
[50] A. Anvari-Moghaddam, A. Seifi, T. Niknam, and M. R. Alizadeh Pahlavani,
‘‘Multi-Objective Operation Management of a Renewable MG (Micro-Grid)
with Back-up Micro-Turbine/Fuel Cell/Battery Hybrid Power Source,’’
Energy, vol. 36, no. 11, pp. 6490–6507, 2011.
[51] A. Anvari-Moghaddam, T. Dragicevic, J. C. Vasquez, and J. M. Guerrero,
‘‘Optimal utilization of microgrids supplemented with battery energy storage
systems in grid support applications,’’ in IEEE First International Con-
ference on DC Microgrids (ICDCM) 2015, 2015, pp. 57–61.
[52] L. Lu and C. Chu, ‘‘Consensus-Based Distributed Droop Control of Syn-
chronverters for Isolated Micro-Grids,’’ IEEE Trans. Power Syst., vol. 30,
no. 5, pp. 914–917, 2015.
[53] J. Lai, H. Zhou, X. Lu, X. Yu, and W. Hu, ‘‘Droop-Based Distributed
Cooperative Control for Microgrids With Time-Varying Delays,’’ IEEE
Trans. Smart Grid, vol. 7, no. 4, pp. 1775–1789, 2016.
[54] T. Dragicevic, J. C. Vasquez, J. M. Guerrero, and D. Skrlec, ‘‘Advanced
LVDC Electrical Power Architectures and Microgrids: A Step toward a New
Generation of Power Distribution Networks,’’ IEEE Electrif. Mag., vol. 2,
no. 1, pp. 54–65, 2014.
[55] D. Linden and T. B. Reddy, Handbook of Batteries, 3rd ed. New York:
McGraw-Hill, 2002.
[56] E. Koutroulis, K. Kalaitzakis, and N. C. Voulgaris, ‘‘Development of a
Microcontroller-based, Photovoltaic Maximum Power Point Tracking Con-
trol System,’’ IEEE Trans. Power Electron., vol. 16, no. 1, pp. 46–54, 2001.
of DC Microgrids with Instantaneous Constant-Power Loads,’’ IEEE Trans.
Power Electron., vol. 26, no. 3, pp. 822–834, Mar. 2011.
[58] N. L. Diaz, T. Dragicevic, J. C. Vasquez, and J. M. Guerrero, ‘‘Intelligent
Distributed Generation and Storage Units for DC Microgrids – A New
Concept on Cooperative Control Without Communications Beyond Droop
Control,’’ IEEE Trans. Smart Grid, vol. 5, no. 5, pp. 2476–2485, 2014.
[59] C. N. Onwuchekwa and A. Kwasinski, ‘‘Analysis of Boundary Control for
Buck Converters With Instantaneous Constant-Power Loads,’’ IEEE Trans.
Power Electron., vol. 25, no. 8, pp. 2018–2032, 2010.
[60] J. M. Guerrero, J. C. Vasquez, J. Matas, L. G. de Vicuna, and M. Castilla,
‘‘Hierarchical Control of Droop-Controlled AC and DC Microgrids—A
General Approach Toward Standardization,’’ IEEE Trans. Ind. Electron.,
vol. 58, no. 1, pp. 158–172, Jan. 2011.
[61] X. Lu, K. Sun, J. M. M. Guerrero, J. C. C. Vasquez, and L. Huang, ‘‘Double-
Quadrant State-of-Charge-Based Droop Control Method for Distributed
Energy Storage Systems in Autonomous DC Microgrids,’’ IEEE Trans.
Smart Grid, vol. 6, no. 1, pp. 147–157, 2015.
[62] M. Lee, D. Chen, K. Huang, C.-W. Liu, and B. Tai, ‘‘Modeling and
Design for a Novel Adaptive Voltage Positioning (AVP) Scheme for
Multiphase VRMs,’’ IEEE Trans. Power Electron., vol. 23, no. 4, pp. 1733–
1742, 2008.
[63] C.-J. Chen, D. Chen, C.-S. Huang, M. Lee, and E. K.-L. Tseng, ‘‘Modeling and
Design Considerations of a Novel High-Gain Peak Current Control Scheme
to Achieve Adaptive Voltage Positioning (AVP) for DC Power Converters,’’
[64] H.-H. Huang, C.-Y. Hsieh, J.-Y. Liao, and K.-H. Chen, ‘‘Adaptive Droop
Resistance Technique for Adaptive Voltage Positioning in Boost DC–DC
Converters,’’ IEEE Trans. Power Electron., vol. 26, no. 7, pp. 1920–1932, 2011.
[65] F. Valenciaga and P. F. F. Puleston, ‘‘Supervisor Control for a Stand-Alone
Hybrid Generation System Using Wind and Photovoltaic Energy,’’ IEEE
Trans. Energy Convers., vol. 20, no. 2, pp. 398–405, Jun. 2005.
[66] F. Valenciaga and P. F. Puleston, ‘‘High-Order Sliding Control for a Wind
Energy Convers. System Based on a Permanent Magnet Synchronous Gen-
erator,’’ IEEE Trans. Energy Convers., vol. 23, no. 3, pp. 860–867, 2008.
[67] T. Dragicevic, J. M. Guerrero, J. C. Vasquez, and D. Skrlec, ‘‘Supervisory
Control of an Adaptive-Droop Regulated DC Microgrid With Battery
Management Capability,’’ IEEE Trans. Power Electron., vol. 29, no. 2,
pp. 695–706, 2014.
[68] L. Che and M. Shahidehpour, ‘‘DC Microgrids: Economic Operation and
Enhancement of Resilience by Hierarchical Control,’’ IEEE Trans. Smart
Grid, vol. 5, no. 5, pp. 2517–2526, 2014.
[69] X. Lu, J. M. M. Guerrero, K. Sun, J. C. C. Vasquez, R. Teodorescu, and
L. Huang, ‘‘Hierarchical Control of Parallel AC–DC Converter Interfaces for
Hybrid Microgrids,’’ IEEE Trans. Smart Grid, vol. 5, no. 2, pp. 683–692,
2014.
[70] B. Wang, M. Sechilariu, and F. Locment, ‘‘Intelligent DC Microgrid With
Smart Grid Communications: Control Strategy Consideration and Design,’’
IEEE Trans. Smart Grid, vol. 3, no. 4, pp. 2148–2156, 2012.
[71] C. Jin, P. Wang, J. Xiao, Y. Tang, and F. H. H. Choo, ‘‘Implementation of
Hierarchical Control in DC Microgrids,’’ IEEE Trans. Ind. Electron., vol. 61,
no. 8, pp. 4032–4042, 2014.
[72] Q. Shafiee, T. Dragicevic, J. C. Vasquez, and J. M. Guerrero, ‘‘Hierarchical
Control for Multiple DC-Microgrids Clusters,’’ IEEE Trans. Energy Con-
version, vol. 29, no. 4, pp. 922–933, 2014.
[73] J. Schonberger, R. Duke, and S. D. Round, ‘‘DC-Bus Signaling: A Dis-
tributed Control Strategy for a Hybrid Renewable Nanogrid,’’ IEEE Trans.
Ind. Electron., vol. 53, no. 5, pp. 1453–1460, 2006.
[74] Y. Gu, X. Xiang, W. Li, and X. He, ‘‘Mode-Adaptive Decentralized Control
for Renewable DC Microgrid With Enhanced Reliability and Flexibility,’’
[75] A. Gkountaras, S. Dieckerhoff, and T. Sezi, ‘‘Performance analysis of hybrid
microgrids applying SoC-adaptive droop control,’’ in Power Electronics and
Applications (EPE’14-ECCE Europe), 2014 16th European Conference on,
2014, pp. 1–10.
[76] E. Liegmann and R. Majumder, ‘‘An Efficient Method of Multiple

Storage Control in Microgrids,’’ IEEE Trans. Power Syst., vol. 30, no. 6, pp.
3437–3444, 2015.
[77] H. C. Ferreira, L. Lampe, J. Newbury, and T. G. Swart, Power Line
Communications: Theory and Appl. for Narrowband and Broadband
Communications over Power Lines. Hoboken, NJ: John Wiley & Sons, 2010.
[78] ‘‘REbus DC Microgrid,’’ 2013. [Online]. Available: http://rebuspower.com/.
[Accessed: 08 Jun. 2014].
[79] T. Dragicevic, J. M. Guerrero, and J. C. Vasquez, ‘‘A Distributed Control
Strategy for Coordination of an Autonomous LVDC Microgrid Based on
Power-Line Signaling,’’ IEEE Trans. Ind. Electron., vol. 61, no. 7, pp. 3313–
3326, 2014.
[80] W. Stefanutti, S. Saggini, P. Mattavelli, and M. Ghioni, ‘‘Power Line
Communication in Digitally Controlled DC–DC Converters Using Switch-
ing Frequency Modulation,’’ IEEE Trans. Ind. Electron., vol. 55, no. 4,
pp. 1509–1518, Apr. 2008.
[81] T. Dragicevic, A. Anvari-Moghaddam, J. C. Vasquez, and J. M. Guerrero, ‘‘DC
Distribution Systems and Microgrids,’’ in Large Scale Grid Integration of
Renewable Energy Sources, A. Moreno-Munoz, Ed. IET, 2017, pp. 211–236.
[82] R. Olfati-Saber, J. A. Fax, and R. M. Murray, ‘‘Consensus and Cooperation
in Networked Multi-Agent Systems,’’ Proc. IEEE, vol. 95, no. 1, pp. 215–
233, 2007.
[83] R. Olfati-Saber and R. M. Murray, ‘‘Consensus Problems in Networks of
Agents with Switching Topology and Time-Delays,’’ IEEE Trans. Autom.
Control, vol. 49, no. 9, pp. 1520–1533, 2004.
[84] Z. Meng, W. Ren, Y. Cao, and Z. You, ‘‘Leaderless and Leader-Following
Consensus with Communication and Input Delays under a Directed Network
Topology,’’ IEEE Trans. Syst. Man, Cybern. Part B Cybern., vol. 41, no. 1,
pp. 75–88, Feb. 2011.
[85] L. X. L. Xiao and S. Boyd, ‘‘Fast linear iterations for distributed averaging,’’ in
42nd IEEE Int. Conf. Decis. Control (IEEE Cat. No. 03CH37475), 2003, vol. 5.
[86] Y. Li and F. Nejabatkhah, ‘‘Overview of Control, Integration and Energy
Management of Microgrids,’’ J. Mod. Power Syst. Clean Energy, vol. 2,
no. 3, pp. 212–222, 2014.
[87] L. Yun Wei, ‘‘Control and Resonance Damping of Voltage-Source and
Current-Source Converters with LC Filters,’’ IEEE Trans. Ind. Electron.,
vol. 56, no. 5, pp. 1511–1521, 2009.
[88] Q. C. Zhong and G. Weiss, ‘‘Synchronverters: Inverters that Mimic Synchronous
Generators,’’ IEEE Trans. Ind. Electron., vol. 58, no. 4, pp. 1259–1267, 2011.
[89] F. Blaabjerg, R. Teodorescu, M. Liserre, and A. V. Timbus, ‘‘Overview of
Control and Grid Synchronization for Distributed Power Generation Sys-
tems,’’ IEEE Trans. Ind. Electron., vol. 53, no. 5, pp. 1398–1409, 2006.
[90] A. Anvari-Moghaddam, J. M. Guerrero, J. C. Vasquez, H. Monsef, and
A. Rahimi-Kian, ‘‘Efficient Energy Management for a Grid-Tied Residential
Microgrid,’’ IET Gener. Transm. Distrib., vol. 11, no. 11, pp. 2752–2761, 2017.
Chapter 6
Control of solid-state transformer-enabled
DC microgrids
Xu She1, Alex Huang2, Xunwei Yu3, and Yizhe Xu4
6.1 Introduction
With the world-wide energy shortage and deterioration of existing power grid, micro-
grid becomes one of the hottest research directions in the power engineering area.
Considering DC nature of many key components in the smart grid, such as photovoltaic
(PV), battery, fuel cell, super capacitor, etc., as well as many DC type loads, such as
light-emitting diode, DC microgrid has received more attention recently since it brings
the opportunity for boosting the efficiency by eliminating the unnecessary power
conversion stages. However, the existing DC microgrid can only interface with the
distribution system by using a heavy and bulky passive line frequency transformer plus
a rectifier, which has large space and heavy weight. Developing a more compact and
active grid interface to enable an intelligent DC microgrid system is still a research
focus. In this chapter, the solid-state transformer (SST)-enabled DC microgrid is pre-
sented. In addition, two system control strategies, namely, the centralized power
management and hierarchical power management strategies, are proposed. In addition,
an improved control strategy is proposed for increasing the penetration of distributed
renewable energy resources (DRER) integration, which controls SST-enabled DC
microgrid as a solid-state synchronous machine (SSSM). With the proposed control
concept, frequency and voltage stability are improved in case of high-power inter-
mittence at either DC load or DRER side. Design examples are given to illustrate the
main characteristics of the presented system and control schemes.
6.2 Solid-state transformer-based microgrid:

architecture and benefits
The SST is a power electronic device that replaces the traditional 50/60 Hz power
transformer by means of high-frequency transformer isolated AC–AC conversion
1
Electric Power, GE Global Research, United States
2
Electric and Computer Engineering, University of Texas Austin, United States
3
Intersil Corporation, United States
4
Grid Bridge Inc., United States
Switching Switching
Source network Transformer network Load
Solid state transformer
Figure 6.1 Configuration of solid-state transformer (SST) from source to load
technique, which is represented in Figure 6.1 [1]. The basic operation of the SST is
firstly to convert the 50/60 Hz AC voltage to a high-frequency one (normally in the
range of several to tens of kilohertz), then this high-frequency voltage is stepped
up/down by a high-frequency transformer with significantly decreased volume and
weight, and finally shaped back into the desired 50/60 Hz voltage to feed the load.
Therefore, the first advantage that the SST can offer is its reduced volume and
weight compared with traditional transformers. It is further seen from the config-
uration of the SST that some other potential functionalities that are not owned by
the traditional transformers may be obtained. First, the use of solid-state semi-
conductor devices and circuits makes the voltage and current regulation a possi-
bility. This brings promising features such as power flow control, voltage sag
compensation, fault current limitation, and others, which are not possible for tra-
ditional transformers. Second, voltage source converters connected from the sec-
ondary terminal of the SST could readily support a regulated DC bus, which could
be connected to DC microgrid, enabling this new microgrid architecture.
Figure 6.2 demonstrates the proposed SST-based DC microgrid system [2].
The AC loads are connected to the AC mains of SST, and the DC sources and loads
are connected to the DC mains for minimizing the conversion stages. The high-
voltage side of the SST is connected to the distribution system; thus, the conventional
transformer is eliminated. The presence of SST can also isolate the distribution
system from the residential side; therefore, the fault on one side will not affect
another. Furthermore, the high-voltage AC/DC converter in the distribution system
side can realize power factor regulation, enabling the Var compensation capability.
Therefore, the SST-based microgrid is more compact and shows superior character-
istics over conventional AC and DC microgrids.
Similar like the traditional microgrids, different power management strategies
could be applied. In the following sections, both the centralized and hierarchical
power management strategies are presented. In addition, SSSM control concepts
are also proposed to enhance the system stability at high-power intermittence
condition. Design examples are given to demonstrate the feasibility of the proposed
SST-enabled DC microgrid.
Control of solid-state transformer-enabled DC microgrids 121
Solid-state AC
transformer load
Grid
DC
load
AC
DESD DC/DC DC
SST-based
microgrid
DRER DC/DC
Figure 6.2 SST-based microgrid architecture: DESD and DRER. DESD,

distributed energy storage device; DRER, distributed renewable
energy resource
6.3 Centralized power management of solid-state

transformer-based DC microgrid
6.3.1 Power management strategy
In the centralized power management strategy, an intelligent energy management
(IEM) algorithm should be developed. The operating modes of DC microgrid can
be defined as shown in Figure 6.3, where three modes are identified, namely, active
grid interaction mode, passive grid interaction mode, and islanding mode.
In the passive grid interaction mode, SST controls the DC voltage. Although the
DC microgrid is interfaced with the distribution system, it balances the power within
the network without any power being exchanged with SST. In the active grid inter-
action mode, due to the limited power balancing capability of the DC microgrid
under certain operating conditions, additional power is transferred between SST and
DC microgrid. In the islanding mode, SST stops working and the battery regulates
the DC bus. The DC microgrid supplies the additional power for the load at AC side
whenever extra power is available. The assumption is made in this chapter that the
SST, as the backup of DC microgrid, can always supply enough power under the grid
interaction mode by an optimized design. The whole power management algorithm
aims at maximizing the utilization of PV and battery and minimizing the burden of
existing AC grid, as well as ensuring a high reliability of the power system.
Passive grid
interaction
mode
Active grid
Islanding
interaction
mode
mode
Figure 6.3 Operating modes and transitions of SST-enabled microgrid
If divided in detail, there are ten possible operating modes for the presented
system, and they are shown in Figure 6.4. The nomenclature in the flowchart is
defined in Table 6.1.
The division of the modes mainly depends on the operating status of the dis-
tribution system, battery state of charge (SOC), and battery power. These modes are
summarized and explained as below.
Mode 1: PV operates in maximum power point tracking (MPPT) mode, battery
stops working, and SST supplies the additional power for DC load.
Mode 2: PV operates in MPPT mode, battery operates in discharging power
limitation mode, and SST supplies the additional power for DC load.
Mode 3: PV operates in MPPT mode, battery balances the power within DC
microgrid, and SST only supplies power for AC load.
Mode 4: PV operates in MPPT mode, battery operates in charging power
limitation mode, and SST absorbs the additional power from DC microgrid.
Mode 5: PV operates in MPPT mode, battery stops working, and SST absorbs
the additional power from DC microgrid.
Mode 6: Part of loads is shed and the system operates in either Mode 8 or
Mode 9, depending on the operating condition.
Mode 7: Part of loads is shed and the system operates in Mode 8, Mode 9, or
Mode 10, depending on the operating condition.
Mode 8: PV operates in MPPT mode and battery operates in voltage tracking
mode.
Mode 9: PV operates in power tracking mode for supplying the total load
power. In this mode, the power reference of PV converter is set to
Ppv ¼ Pload DC þ Pload DC þ Pb max (6.1)
Intelligent energy
management
P* = Ppv-Pload_DC No Yes P* = Ppv-Pload_DC-Pload_AC

Fault?
No Yes No Yes
P*>0 P*>0
Yes No No Yes Yes No No Yes

SOC<SOCmin SOC>SOCmax SOC<SOCmin SOC>SOCmax
Mode1 Yes No No Yes Mode5 Mode6 Yes No No Yes Mode10

Abs(P)*>Pbmax P*>Pbmax Abs(P)*>Pbmax P*>Pbmax
Mode2 Mode3 Mode4 Mode7 Mode8 Mode9
Grid interaction mode Islanding mode
Figure 6.4 Intelligent energy management system diagram of SST using PV and battery
Table 6.1 Nomenclature for IEM system
Ppv Power of PV
Pload_DC Power of DC load
Pload_AC Power of AC load
SOCmin Minimum allowed SOC
SOCmax Maximum allowed SOC
Pbmax Maximum allowed battery power
The battery operates in voltage tracking mode, and its absorbed power is
fixed to charging power limitation automatically.
Mode 10: PV operates in power tracking mode for supplying the total load power
and the battery controls the DC bus voltage while no power is delivered to the
load.
As it can be seen above, Modes 1–5 belong to the active grid interaction mode.
Mode 3 belongs to the passive grid interaction mode. Modes 6–10 belong to the
islanding mode. In the presented power management strategy, the matched battery
capacity needs to be designed, and the system should operate in Mode 3 for most
of the time. Only when the local power balancing capability is limited, a transition
to other modes occurs. Thus, the presented IEM system can maximize the utili-
zation of PV and battery and minimize the effect to the existing AC power grid
architecture.
6.3.2 Case study

To verify the presented intelligent power management algorithm, a simulation
platform is established by using the average modeling technique. The average
model instead of the switching model is preferred for this large power system since
it speeds up the simulation without losing the capability of capturing the main
characteristics of the system. In the developed simulation platform, the power rat-
ing of SST is set to 20 kVA, the maximum power of PV under the irradiation of
1 kW/m2 is set to 8 kW, and the maximum charging/discharging power of the
battery pack is set to 6 kW. Different load profiles are adopted to test different
cases. Only parts of the cases are tested for demonstrating the main characteristics
of the system due to page limitations.
6.3.2.1 Passive grid interaction (Mode 3)

To emulate the operating Mode 3, which is the passive grid interaction mode, the
AC load for SST is set to 10 kW, and DC load is set to 4 kW. The key operating
waveforms are shown in Figure 6.5.
The irradiation of PV is gradually changing from 300 W/m2 to 1 kW/m2,
and then back to 300 W/m2 to emulate the irradiation of a day, as shown in
Figure 6.5(a). Figure 6.5(b) shows the power distribution of the presented system.
1,200
1,100
1,000
Irradiation (W/m2)
900
800
700
600
500
400
300
200
0.5 1 1.5 2 2.5 3 3.5 4 4.5
(a) Time (s)
15,000
SST power, AC load power

10,000
PV power
Power (W)
DC load power
5,000
Battery power
–5,000
0.5 1 1.5 2 2.5 3 3.5 4 4.5
(b) Time (s)
600
240 V AC voltage 400 V DC voltage 120 V AC voltage
400
Voltage (V)
200
–200
–400
2.5 2.55 2.6 2.65 2.7
(c) Time (s)
Figure 6.5 Key operating waveforms in Mode 3 of the system in Figure 6.2:
(a) irradiation of PV panel, (b) power distribution of the system,
(c) low-voltage terminal waveforms, (d) high-voltage terminal
waveforms, (e) SOC of the battery
1.5
Distribution voltage (7.2 kV base) Input current (3 A base)
1
Voltage (V) Current (A)
0.5
–0.5
–1
–1.5
2.5 2.55 2.6 2.65 2.7
(d) Time (s)
80.04
80.03
State of charge (%)
80.02
80.01
80
79.99
79.98
0.5 1 1.5 2 2.5 3 3.5 4 4.5
(e) Time (s)
Figure 6.5 (Continued )
The power of the PV has the same trend with the irradiation. SST only supplies
the AC load power, which is 10 kW. Since the power of the battery is within its
limitation, it can balance the power between DC load and PV. Figure 6.5(c)
shows the regulated low-voltage terminal waveforms, including 400 V DC and
120/240 V AC. In Figure 6.5(d), the SST input voltage and current are depicted.
The current is in phase with voltage, indicating a unity power factor operation.
The battery SOC is described by Figure 6.5(e), where the initial SOC is set to
80%. The trend of SOC depends on the direction of current. A positive battery
current discharges the battery and a negative battery current charges the battery in
the presented system in Figure 6.2.
6.3.2.2 Transition from passive grid interaction mode to active grid

interaction mode (Mode 2 to Mode 3)
The transition from passive grid interaction mode to active grid interaction mode is
demonstrated. Mode 2 and Mode 3 are chosen to verify the operation, in which the
AC load is set to 10 kW and the DC load is set to 1 kW. Key operating waveforms
× 104
1.5 600
SST power AC load power 240 V AC voltage 400 V DC voltage 120 V AC voltage
1 400
PV power
Voltage (V)
Power (W)
0.5 200
DC load power
0 0
Battery power
–0.5 –200
Passive grid interaction mode Active grid interaction mode Passive grid interaction mode
–1 –400
0.5 1 1.5 2 2.5 3 3.5 4 4.5 2.5 2.55 2.6 2.65 2.7
(a) Time (s) (b) Time (s)
1.5 80.1
1 80.08
State of charge (%)

0.5
Current (A)
80.06
0
80.04
–0.5
80.02
–1
Passive grid interaction mode Active grid interaction mode Passive grid interaction mode
–1.5 80
0.5 1 1.5 2 2.5 3 3.5 4 4.5 0.5 1 1.5 2 2.5 3 3.5 4 4.5
(c) Time (s) (d) Time (s)
Figure 6.6 Key operating waveforms during mode transition (Mode 2 to Mode 3): (a) power distribution of the system,
(b) low-voltage terminal waveforms, (c) high-voltage terminal current (3 A base), (d) SOC of the battery pack
are shown in Figure 6.6. Figure 6.6(a) shows the power distribution of the system.
At the beginning, the power of battery pack is within the limitation and the system
operates in the passive grid interaction mode (Mode 3). With the irradiation of PV
panel increasing, the charging power of the battery also increases. Then the system
transits into the active grid interaction mode (Mode 2) when battery power reaches
the 6 kW limitation. The additional power of DC microgrid flows into the SST and
supplies to the AC load. After that, the PV power decreases to a certain value and
the system operates back to passive grid interaction mode. Figure 6.6(b) depicts the
well-regulated AC and DC voltages. Figure 6.6(c) demonstrates the input current of
SST, which shows the same trend with the SST power since the input voltage is
fixed. Figure 6.6(d) shows the SOC of the battery and it increases continuously.
As it can be seen, the transition between these two modes is smooth and occurs
automatically.
6.3.2.3 Islanding mode (Mode 8)

The microgrid transits into the islanding mode when fault occurs in the distribution
system. How to detect the grid fault is not addressed here. For the demonstration
purpose, operation of Mode 8 is chosen and the waveforms are shown in Figure 6.7,
where the AC load is set to 4.5 kW and the DC load is set to 1 kW.
Power distribution of the system is shown in Figure 6.7(a), where the battery
balances the power between PV and load (both AC and DC load). Figure 6.7(b)
shows the key waveforms at the low-voltage terminal. It can be seen that the DC
bus voltage is regulated well by the battery converter. Figure 6.7(c) shows the SOC
of the battery pack. When the battery current is positive, the SOC decreases.
Otherwise, the SOC increases.
6.3.2.4 Islanding mode transition (Mode 8 to Mode 9)

Finally, the mode transition in the islanding mode is studied. In the study, the AC
load is set to 1 kW, and the DC load changes from 2 to 0 kW at 2.7 s. The key
operating waveforms are shown in Figure 6.8. Figure 6.8(a) shows the power dis-
tribution in this case. Before the DC load changes, the system operates in mode 8
and the battery can balance the power between the PV and the load. Thus, the PV
operates in MPPT mode for extracting the maximum power available. When the
DC load is cut at 2.7 s, the power command of the battery is larger than 6 kW and
battery then operates in charging power limitation mode. The PV no longer oper-
ates in MPPT mode and transits to power tracking mode. Figure 6.8(b) illustrates
the 400-V DC voltage during the mode transition. Due to the cutting down of the
DC load, a voltage overshoot occurs and recovers within 0.05 s. The SOC of the
battery is given in Figure 6.8(c), and the battery is charged continuously. Clearly,
the transition between different islanding modes is smooth.
6.3.3 Summary
A centralized power management strategy is presented for SST-enabled microgrid
system with PV and battery. It renders the coordinate management of power
10,000
PV power
8,000
6,000 AC load power

Power (W)
4,000
DC load power
2,000
0
Battery power
–2,000
–4,000
0.5 1 1.5 2 2.5 3 3.5 4 4.5
(a) Time (s)
600
240 V AC voltage 400 V DC voltage 120 V AC voltage
400
Voltage (V)
200
–200
–400
2.5 2.55 2.6 2.65 2.7
(b) Time (s)
80.01
80.005
State of charge (%)
80
79.995
79.99
79.985
79.98
0.5 1 1.5 2 2.5 3 3.5 4 4.5
(c) Time (s)
Figure 6.7 Mode 8 key operating waveforms: (a) power distribution of the system,
(b) low-voltage terminal waveforms, (c) SOC of battery pack
between DC microgrid and distribution system. In addition, the impact of the

established DC microgrid to the existing power grid can be minimized, which
promises a better reliability and stability. The presented strategy is verified through
some simulation studies.
× 104
1
PV power
0.5
DC load power
Power (W)
0
AC load power
–0.5
Battery power
Mode 8 Mode 9
–1
0.5 1 1.5 2 2.5 3 3.5 4 4.5
(a) Time (s)
500
450
Voltage (V)
400
DC voltage dynamics during transition

350
300
2.6 2.65 2.7 2.75 2.8 2.85 2.9 2.95 3
(b) Time (s)
80.08
80.06
State of charge (%)
80.04
80.02
80
79.98
0.5 1 1.5 2 2.5 3 3.5 4 4.5
(c) Time (s)
Figure 6.8 Key operating waveforms during mode transition (Mode 8 and
Mode 9): (a) power distribution of the system, (b) 400-V DC bus
voltage dynamics, (c) SOC of battery pack
6.4 Hierarchical power management of solid-state

transformer-enabled DC microgrid
6.4.1 Power management strategy

The hierarchical control is a hybrid method, which combines the advantages of both
centralized and distributed control and can be served as standard method for the
microgrid [3]. As shown in Figure 6.9, the system control strategy frame is categorized
into three layers: primary control, secondary control, and tertiary control. The primary
control is the distributed control, which ensures that the microgrid system can operate
without communication. Therefore, the primary control usually takes effect at the
micro-second level, which is basically the same level as the converter control. All
the local information, including the voltage, current, SOC, etc., are sent to the upper
controller, which implements the tertiary control and secondary control through a
bidirectional communication link. Here, the DC microgrid is enabled by the SST, and
therefore, the SST controller is used as the upper controller [4]. The objective of the
secondary control is to recover the microgrid bus voltage to achieve seamless transfer
as the system switches from islanding mode to SST-enabled mode. The time scale of
the secondary control is on the order of milliseconds to seconds.
6.4.1.1 Primary control algorithm

When the system operates in islanding mode, the battery needs to regulate the DC
bus voltage alone. Therefore, (6.2) can be obtained.
dVDC
Ceq VDC ¼ ðPDC Load PBattery PPV Þ (6.2)
dt
Tertiary control
Battery charge
and discharge
control at Secondary control
hours level
Voltage recover
second level
Primary control
Voltage and current control

microsecond level
Individual converter
Figure 6.9 Hierarchical control frame

Since PPV is equal or larger than zero, the power direction of battery is determined
by the difference between the PPV and PDC_Load. For example, the power of the
battery will be larger than zero when PDC_Load is larger than PPV, while the power
of the battery will be less than zero when PDC_Load is smaller than PPV.
Basically, the primary control algorithm is based on DC bus signal control, and
therefore, the distributed control can be achieved. In this control algorithm, the DC bus
voltage is no longer a fixed value, but varies in a certain range in the islanding mode.
The DC bus voltage range is defined as from 360 to 400 V in the presented system
design. In islanding mode, only the battery is used to regulate the DC bus voltage.
Droop control is adopted for battery control with the expression shown in (6.3):
V b ¼ V 0 R b Ib (6.3)
where Vb is the reference for the DC bus voltage, Vo is bus voltage value without
load and it is set to 380 V, Rb is the virtual output impedance, and Ib is battery
output current. For the PV module, it always operates in MPPT mode to deliver
power to the system. Noted here, the load management is involved in the primary
control. The load is divided into the critical load and the noncritical load. When the
PV has no output power to the system, the bus voltage will drop. The battery has to
source more power to the load and the battery SOC will decrease rapidly. To ensure
that the DC microgrid can operate as long as possible in islanding mode, the battery
SOC has to be considered. Therefore, to avoid battery SOC dropping fast, the
noncritical load will be shed from the system when the bus voltage decreases below
some preset threshold (370 V). When the bus voltage recovers to a certain value,
380 V for instance, the noncritical load can be reconnected to the system. The
primary control diagram is shown in Figure 6.10.
Battery control
v0 Vb jb
Droop Voltage Current
curve control control
Vbus ib
ib
PV control
MPPT mode
ipv
vref Voltage Current jPV
MPPT
control control
vpv
vbus
vbus Load control signal

Filter 1
0 370 380
Figure 6.10 Primary control diagram of the DC microgrid system

6.4.1.2 Secondary control algorithm

Since the DC bus voltage is regulated by droop control in islanding mode, the bus
voltage might deviate from the SST DC terminal output voltage. To achieve the
seamless transfer from islanding mode to SST-enabled mode, the secondary control
is adopted. Specifically, when the DC microgrid needs to connect to the SST, the
secondary control-enabled signal is triggered. The SST DC terminal voltage is
sensed and compared with the DC microgrid bus voltage. Then the voltage dif-
ference is processed by a compensator and the output is sent to the battery unit via
the low-bandwidth communication to restore the DC microgrid bus voltage (see
Figure 6.11). Since the PV and load control do not contribute to regulating the DC
bus voltage, their control remains unchanged as shown in Figure 6.10. Due to the
space limitations, only the battery module control is depicted in Figure 6.11.
The controller can be represented as
Vs ¼ ksp ðVSST dcout Vbus Þ þ ðksi ðVSST dcout Vbus Þ=sÞ (6.4)
where ksp and ksi are the control parameters for the secondary control. Therefore,
the droop control of the battery becomes
V b ¼ V 0 Rb Ib þ Vs (6.5)
Based on (6.5), the key function of the secondary control is to shift the droop curve
(see Figure 6.12). When the SST’s DC output voltage is higher than DC microgrid
Secondary control
v0 Battery control
enable signal
VSST_dcout
Secondary Vb Current jb
Saturator Droop Voltage
control Vs curve control
control
ib
ib
Vbus
Figure 6.11 Secondary control diagram of the DC microgrid system
c
d V0 b
Vs < 0 Vs > 0
a
0 I
Figure 6.12 Droop curve shift with secondary control

bus voltage, Vs is larger than zero and the droop curve is shifted upward (battery
operation point from a to b), and vice versa (battery operation point from c to d).
It is noticed that since the low bandwidth communication is adopted, the
secondary controller’s time step should agree with that of the communication to
prevent secondary controller saturation.
6.4.1.3 Tertiary control algorithm

An appropriate SOC value is necessary for the battery to supply the DC bus voltage
while operating in islanding mode. Therefore, SOC is included in the tertiary
control. When the system operates in SST-enabled mode, the DC bus voltage will
be regulated by the SST, and therefore, SST DC output is a stiff bus as seen from
the DC microgrid. Then the battery charge or discharge can be determined by the
SST DC voltage and battery droop curve. There are two options for managing the
battery SOC in SST-enabled mode. The first is to change the SST DC terminal
output voltage, and the second is to shift the battery droop curve, both of which are
detailed below.
Change SST DC terminal output voltage
Since the PV and load control do not contribute to regulate the DC bus voltage,
their control is the same as shown in Figure 6.10. Only the battery module control is
depicted in Figure 6.13. As shown in Figure 6.13, the SOC is evaluated and pro-
cessed by the current reference selection block. If the SOC reaches the HSOC (upper
limit for SOC) or LSOC (lower limit for SOC), Iref equals Id (discharging current for
battery) or Ic (charging current for battery). As soon as the SOC reaches the limit
value, the tertiary control-enabled signal is triggered and tertiary control starts to
operate. However, based on Figure 6.12, the secondary control achieves voltage
recovery by sourcing or sinking more current, which will cause the battery SOC to
reach the SOC limit faster. Therefore, the secondary control will be disabled when
tertiary control begins to operate. The tertiary controller can be represented as
kcp ðIref Ib Þ þ kci ðIref Ib Þ
Vc ¼ (6.6)
s
Secondary control v0
enable signal Battery control
VSST_dcout
Secondary Vb Voltage jb
Saturator Droop Current
control curve control control
Vs Vbus ib
Vbus
Tertiary control Vbus

enable signal
I
Iref Vc jSST
Id Tertiary Voltage
SOC SOC Saturator
LSOC HSOC control regulator
Ic
ib VSST_dcref
Current reference selection
Figure 6.13 First tertiary control diagram of the DC microgrid system

V V
g
V0 VBus_ref
V0
VBus_ref e
e
f
0 0 I
I
(a) SOC to HSOC (b) SOC to LSOC
Figure 6.14 SST DC voltage shift in the tertiary control shown in Figure 6.13:
(a) SOC to higher boundary and (b) SOC to lower boundary
where kcp and kci are the tertiary control parameters. The bus voltage reference can
be as shown in (6.7):
VBus ref ¼ VSST dcref Vc (6.7)
As shown in Figure 6.14(a), for example, if Iref equals Id, the bus voltage operation
point will shift from point e to point f. Here, the bus voltage is less than the V0
for battery droop curve. Thus, the battery will switch to discharging mode auto-
matically. When SOC reaches LSOC, the SST DC output voltage will be higher than
V0 (point e to point g) and the battery will be charged [see Figure 6.14(b)].
Shifting battery droop curve
As in the previous tertiary control scheme, the current reference is determined by
the current reference selection block based on the SOC. If the SOC reaches the
HSOC, Iref equals Id. Or if the SOC reaches the LSOC, Iref equals Ic. In contrast to
the aforementioned tertiary control scheme, the battery droop curve is changed
instead of the SST DC bus voltage. Then the tertiary controller can be represented
as following:
ktp ðIref Ib Þ þ kti ðIref Ib Þ

Vt ¼ (6.8)
s
where Ktp and kti are the control parameters for the tertiary controller. Therefore,
for battery droop control, (6.5) becomes
V b ¼ V 0 Rb Ib þ Vt (6.9)
The secondary control will be disabled at the time which or before the tertiary
control starts. The key point for this tertiary control scheme is to shift the droop
curve as shown in Figure 6.15. When the battery SOC is low, the droop curve will
shift down, and vice versa. Since the PV and load control do not contribute to
regulate the DC bus voltage, their control is the same as shown in Figure 6.10. Due
to the space limitations, only the battery module control is depicted in Figure 6.16.
V0
SOC reaches SOC reaches

LSOC HSOC
0 I
Figure 6.15 Droop curve shift with second tertiary control shown in Figure 6.13
Secondary control v0
enable signal Battery control
VSST_dcout
Secondary Vb jb
Saturator Droop Voltage Current
control curve control control
Vs
Vbus ib
Vbus Vt
I
Id I
SOC SOC ref Tertiary Saturator
LSOC HSOC control
Ic
ib Tertiary control
Current reference selection enable signal
Figure 6.16 Second tertiary control diagram shown in Figure 6.13
6.4.2 Case study of a small-scale DC microgrid

To verify the primary control algorithm shown in Figure 6.10, an experiment is
carried out on a lab test-bed. A 500-W battery module and a 200-W PV module are
constructed to constitute the DC microgrid. The input voltage of the SST is 1.2 kV,
the low voltage DC link is 380 V, and the low voltage AC link is 240-V AC. When
the PV generation unit is operating with MPPT control, the maximum power is
100 W and the corresponding MPPT voltage is about 48 V. The nominal battery
voltage is 48 V and the capacity is 20 A h.
6.4.2.1 Case I: Primary control of the SST-enabled

DC microgrid system
To verify the primary control algorithm, an experiment is carried out in the test-
bed. The DC microgrid operates in islanding mode and the result is depicted in
Figure 6.17. Initially, only the battery supplies power to the load (2 kW). Then the
PV connects to the system and supply the power to the load. Since the PV’s output
Noncritical
load shedding
Noncritical
PV on load shedding
Noncritical
load on
1) Load control signal 2 V/div

PV off PV on
2) PV output current 5 A/div
3) Battery current 2 A/div Noncritical
Time 4 s/div
4) Bus voltage 200 V/div load back
Figure 6.17 Waveforms from the primary control
power is larger than the load, the battery switches to the charging mode auto-
matically and the bus voltage is higher than 380 V. When the noncritical load
(1 kW) connects to the system, the bus voltage drops and the battery switches back
to the discharging mode. Since the bus voltage exceeds the 370 V (value for load
shedding), the noncritical load remains connected to the bus. After the PV dis-
connects from the system, the battery has to supply the power to the load alone and
the bus voltage decreases. After 0.5 s when the bus voltage drops below 370 V, the
noncritical load is shed from the system, and the bus voltage can recover. When
the PV is connected back to the system the bus voltage increases. After 0.5 s when
the bus voltage again exceeds 380 V (the value for noncritical load back), the
noncritical load is reconnected to the system automatically.
6.4.2.2 Case II: Secondary control of the SST-enabled

DC microgrid system
To verify the secondary control algorithm, experimental results are shown in
Figure 6.18. The communication ports are not included in this chapter, so a
software trigger is used to start the secondary control; the time step for secondary
control is 1 ms.
Figure 6.18(a)–(c) shows the waveforms with the secondary controller.
Figure 6.18(a) shows the SST waveforms, including the input voltage and current,
high voltage DC of one AC/DC power stage, and low-voltage DC. In Figure 6.18(b),
before the secondary control loop starts, the DC microgrid bus voltage is less than
380 V because the battery and the PV have been supplying the power to the load.
As soon as the secondary control loop begins to operate, the microgrid bus voltage
increases with the secondary control time step until it reaches 380 V. Then the DC
1) Input current 4 A/div
2) Input voltage 2 kV/div
3) VhDC 1 500 V/div
Time 1 s/div
4) VlDC 500 V/div
(a)
Connect the SST

Secondary loop starts
1) Battery output current 2 A/div

2) Bus voltage 200 V/div Time: 4 s/div
3) Voltage recovery loop flag 1 V/div
(b)

Time 1 s/div
2) PV voltage 50 V/div
(c)
Figure 6.18 Waveforms with secondary control for the system in Figure 6.2:
(a) SST waveforms, (b) battery current, DC microgrid bus voltage,
and SST AC output, (c) PV waveform
microgrid connects to the SST, and a seamless transfer is achieved. Furthermore, the
battery droop curve is lifted based on the secondary control loop as previously
described. Therefore, when the system is in SST-enabled mode, the battery still
outputs current to the load. For the PV module, there is no change in its output
voltage and current because it always operates in the MPPT mode as shown in
Figure 6.18(c).
6.4.2.3 Case III: First tertiary control

To verify the first tertiary control algorithm, experimental results are shown in
Figure 6.19. SOC estimation is beyond the focus of the chapter, and therefore, it is
not covered here. A software trigger is used to start the tertiary control, emulating
the time when the battery SOC reaches its upper limit.
Figure 6.19(a) shows the SST waveforms, including the input voltage and
current, high-voltage DC of one AC/DC power stage, and low-voltage DC. When
the tertiary control starts, the SST output DC bus voltage increases because the Vc
in (6.7) is less than zero. Figure 6.19(b) shows the SST AC-side voltage, DC
microgrid bus voltage, and the battery current. It is noted that the secondary control
is disabled before the tertiary control starts. Therefore, the battery module’s output
current is almost zero before the tertiary control starts. When the tertiary control
starts, the DC bus voltage increases until the battery output current reaches its
current reference (2 A). Since the PV is still in the MPPT mode, its curve is same
as the previous case, which is shown in Figure 6.19(c).
6.4.2.4 Case IV: Second tertiary control

To verify the second tertiary control algorithm, experimental results are shown
in Figure 6.20. Similarly, a software trigger is used to start the tertiary control
to emulate when the SOC reaches the LSOH, and the time step for tertiary control
is 0.1 s.
Figure 6.20(a) shows the SST waveforms, including the input voltage and
current, high-voltage DC, and low-voltage DC. When the tertiary control starts,
in contrast to the previous tertiary control, the SST output DC bus voltage remains
constant at 380 V. Figure 6.20(b) shows the SST AC-side voltage, DC microgrid
bus voltage, and the battery current. When the secondary control loop stops, the
battery output current drops to zero because the V0 equals 380 V (the SST DC
output voltage) without the voltage recovery item (i.e., Vs is zero). After the bat-
tery’s output current reaches zero, the tertiary control loop is triggered and begins
to operate (the green line goes to 3.3 V). Then the battery starts to charge until
the current reaches the Ic value (2 A). As with the first tertiary control method, the
PV is still in MPPT mode, and there is no change in its waveform whether the
secondary control stops or tertiary control starts, as shown in Figure 6.20(c).
6.4.2.5 Comparison of the two tertiary control methods

The first tertiary control has the following disadvantages: (1) when the DC bus
voltage changes, the load power changes; (2) the DC bus voltage range is limited
Tertiary control starts

3) VhDC 500 V/div
4) VlDC 200 V/div Time 1 s/div
(a)
1) AC output voltage 500 V/div

2) Battery output current 5 A/div Time 1 s/div
3) DC bus voltage 200 V/div
(b)

Time 1 s/div
2) PV voltage 50 V/div
(c)
Figure 6.19 Waveforms with first tertiary control principal: (a) SST waveform,
(b) battery current, DC microgrid bus voltage, and SST AC output,
(c) PV current and voltage
3) VhDC 500 V/div Time 1 s/div
4) VlDC 500 V/div
(a)
Secondary control stops

1) Battery current 5 A/div

2) Bus voltage 200 V/div
3) Tertiary control loop flag 2 V/div
4) Battery voltage 20 V/div Time 4 s/div
(b)

2) PV voltage 50 V/div Time 1 s/div
(c)
Figure 6.20 Waveforms with second tertiary control: (a) SST waveform,
(b) battery current, DC microgrid bus voltage, and control flag,
(c) PV waveform
by the SST output voltage range. When the charging and discharging current
references are large, the SST output DC voltage might reach its limitation and then
the battery’s output current cannot match with the references. Compared to the first
tertiary control method, the second tertiary control method does not have these two
problems since it just changes the battery’s droop curve. However, its disadvantage is
that communication ports need to be involved because the control output commands
need to be sent from the SST to the battery module, which will degrade the system
reliability. For the first tertiary method, the communication is exclusive since the
control method is only implemented on the SST side. Which method to select will
depend on the designer and the system requirements.
6.4.3 Summary
A hierarchical power management strategy for the SST-based DC microgrid is
proposed, which includes primary control, secondary control, and tertiary control.
The DC microgrid not only can operate more reliably in islanding mode by primary
control, but can also seamlessly transfer the DC microgrid from islanding mode to
SST-enabled mode via secondary control. The battery SOC is involved in the ter-
tiary control for battery management. Experimental results are presented to verify
the proposed schemes.
6.5 Control of SST-enabled DC microgrid as a solid-state

synchronous machine (SSSM)
6.5.1 Concept of the SSSM

The SST-enabled DC microgrid integrates DRER, distributed energy storage
devices (DESD) and DC loads, and interfaces to the legacy grid by a phase
locked loop (PLL). Traditionally, the SST is controlled as a current source and
from the grid point of view, the SST-enabled DC microgrid lacks the rotational
inertia of a traditional synchronous generator, which follows the grid frequency
and the high-power intermittence of DRER, and load is instantaneously reflected
at the point of common coupling. The concept of SSSM is proposed to over-
come the frequency and voltage instability issues in a high DRER penetration
system [5–7].
Figure 6.21 shows the overall architecture of an SST-enabled DC micro-grid
when it is controlled as an SSSM. The detailed control diagram applied in single-
phase inverter is shown in Figure 6.22. The SST interfaces the DC microgrid to
legacy grid with frequency and voltage regulation capabilities. The microgrid acts
as a voltage source which mimics the behavior of the synchronous machine to form
the grid. It improves the aforementioned SST-enabled DC microgrid with (1) a
larger inertia; (2) a simple and extendible model for large-scale power system
analysis; (3) no use of PLL during normal operation; (4) seamless transition
from grid-connection mode to islanding mode; (5) up/down reserve for frequency
regulation with the support of energy storage (battery).
ωg, δg, vg
PCC
AC grid Line impedance
Pout, Q ω, δ, E
Q ω
Frequency
Volt-Var
regulation
regulation
vg vDC (first droop)
vDC_ref
E
SST
Pin vDC DC microgrid bus
Pbat PPV Pload
vDC
Battery Second
droop PV
Load
OR unit
Pbat_ref
SOC/
Battery unit economic
Figure 6.21 Overall architecture of the SST-enabled DC microgrid as an SSSM

with the associated control strategy
Frequency vDC GLPF1(s) vDC_ f

regulation
vDC_R
+ vDC_ref – + 1 ω δ
Voltage reference generator
ωR mω – 1
+ Jvirs s
–
2E–d
Gpss(s)
Volt-Var +
QR mvg GLPF3(s) + E
regulation +
– +
Q GLPF2(s) Qf ER
Figure 6.22 Detailed control strategy diagram of the SST controlled as SSSM
6.5.2 Frequency regulation

The traditional SM rotor’s mechanical characteristic is called the swing equation
and expressed as
dwr
J þ Dðwr wg Þ ¼ Tm Te (6.10)
dt
where wr is the rotor angular frequency, wg is the grid frequency, J is the angular
momentum inertia of the rotor, D is the damping coefficient generated by the
damper windings, and Tm and Te are the input mechanical torque and load electric
torque, respectively. The dynamics of SM rotor is normally slow due to the large J.
In the SST, the dynamic equation of the DC voltage with a DC capacitor CDC can
be derived as
dvDC
CDC vDC ¼ iin vDC iout vDC ¼ Pin Pout (6.11)
dt
Equation (6.11) is similar with (6.10), where CDCvDC, VDC, Pin, and Pout are,
respectively, corresponding to J, wr, Tm, Te. Assuming that SST is lossless, Pout is
also the output active power into the AC grid. Equation (6.11) shows that the DC
capacitor has similar dynamics with the SM rotor and can be used for the inertia
emulation. But in order to get the same inertia as the SM rotor, the DC capacitor
will be very large.
The equation of the frequency regulation in Figure 6.23 is derived as
dw
Jvir þ mw ðw wR Þ ¼ GLPF1 ðsÞvDC vDC R (6.12)
dt
Equation (6.12) is also similar with (6.10), where the virtual inertia parameter Jvir
and damping coefficient mw are, respectively, corresponding to J and D. So this
regulation can also emulate the SM rotor’s characteristic without physical com-
ponent. The difference is that the torque is replaced by VDC, and the dynamic
equation (6.11) is then inserted in the swing equation. GLPF1 is a first-order low-
pass filter used to suppress the low-order harmonics for VDC.
The virtual rotor model of the SSSM concept-based SST is developed based on
(6.11) and (6.12), as shown in Figure 6.23 in s-domain. When Pin or Pout varies, the
difference between them will change VDC based on CDC, and then w based on Jvir.
This will change d and then Pout according to the power–angle relationship. Finally,
Pin vDC_R
+ –
– 1 vDC vDC_ f + 1
Pout GLPF1(s) w
CDCvDCs Jvir S + Kw
+
mwwR
Figure 6.23 Block diagram of the SST’s virtual rotor model

a new balanced working point for Pin, Pout, and d is reached, while w synchronizes
with wg again. The microgrid’s total inertia is determined by CDC, VDC, wc1, and
Jvir. When CDC is designed normally, Jvir will be an extra control parameter to
adjust the inertia to meet the stability requirement in the power system.
According to the analysis above, any power from the DC microgrid can be
transmitted to the AC grid through the SST at a given AC frequency. The dynamics
of AC output active power is slowed down and completely decoupled from the
intermittent, variable DRERs and loads at the DC microgrid. So this microgrid’s
model is simple and promising for studying the scalability and stability of the
paralleled systems. Due to its swing characteristic, the microgrid is able to respond
to the short-term requirements of the grid frequency regulation.
6.5.3 Power up/down reserve support

Different from the traditional SST control that synchronizes with power grid with
no frequency response, the SSSM-based SST will supply the power system with
up/down power reserve with its emulated angular speed w and support of energy
storage units such as supercapacitor or battery. Dual droop control, as shown in
Figure 6.24 is proposed to coordinate the SST with DESD and implement the up/
down reserve for primary frequency regulation, similar to the governor control in
synchronous machine.
1. First droop in SST: Since the energy storage units in the DC microgrid only
react to VDC, w in the SST must be linked to VDC through the active power.
The traditional droop controls for active power in AC and DC system,
respectively, are
w wR ¼ kw ðP Prate Þ (6.13)

vDC vDC R ¼ kvDC ðP Prate Þ (6.14)
vDC vDC
SSSM Battery
220 units
200
180
ω Pbat
0 wmin ωR ωmax –Pbat_max 0 Pbat_max
Figure 6.24 Detailed diagram of dual droop control for SSSM and battery units’
coordination
where Prate is the converters’ rated power, and kw and kvDC are the droop slopes
for AC and DC systems. Combining (6.13) and (6.14), the first droop in the
SST can be rewritten as
kvDC
vDC vDC R ¼ ðw wR Þ ¼ mw ðw wR Þ (6.15)
kw
In (6.15), mw is determined by the allowable ranges of w and VDC. Typically,
1/kw is proportional to the SST’s rated power. The DC voltage ranges are
designed to be the same for different SSTs, then 1/kvDC is also proportional to
their rated power, and the resulting mw will be the same for different SSTs. For
the frequency regulation control in Figure 6.23, this droop control is naturally
integrated in it, and mw is also the coefficient for the part of the damping.
2. Second droop in energy storage units: The second droop in the energy storage
units is a traditional V–P droop, and mvDC is its droop slope. According to the
first droop in the SST, Pbat should be controlled based on VDC. This control is
shown in the right column in Figure 6.24. The hierarchical power management
discussed in Section 6.4 for battery is still applicable as an offset power
reference.
3. Coordination of dual droop: Figure 6.25 shows the coordination diagram of the
dual droop control. VDC is controlled according to w through the first droop.
This voltage is a command, which indicates the energy storage units how much
power SST requires corresponding to the second droop curve. Then energy
storage units will output the corresponding power through the second droop
and inject it into the AC grid via the SST. This is similar to the traditional
active power control in the SM, and finally, the SST with the DC microgrid
will have the w–P droop characteristic.
The dual droop control is distributed and autonomous without communication. In
steady state, the SSTs connected to the same AC grid will synchronize with it, share
and balance the power via the AC frequency. So the SST can provide up/down
reserve to the grid when wg varies in a desirable region, with the support from
energy storages. The power in the DC microgrid is also shared and balanced based
on the second droop among paralleled energy storage units. So the power man-
agement of the AC grid and DC microgrid are achieved simultaneously either in
PPV + Pload
+ Pin
Pbat + Power flow
Battery Pout
Pbat_ref SSSM vDC_R

w
+ –
Second vDC vDC_ref – +
DC mw wR
droop
microgrid
Figure 6.25 Coordination diagram of a dual droop control

grid-connection or islanding modes. During the dynamics, the dual droop control
will also support the demanded inertia at a given wg. It is achieved by commanding
energy storages to provide temporary energy to the SST through VDC, when w and
VDC vary due to the disturbances from the AC grid or DC microgrid.
6.5.4 Voltage regulation

In Figure 6.22, a simple Volt-Var droop control (shown in Figure 6.26) is adopted
to mimic the excitation system of SMs, where mvg is its droop slope. In SMs, the
exciter is usually a proportional controller, and the field circuit flux is produced
based on it after a long-time delay. To emulate this, a first-order LPF can be
adopted, shown as GLPF3(s) in Figure 6.22.
6.5.5 Case study

To verify the primary control algorithm, a low-power experiment is carried out in
the lab test-bed. A 500-W battery module, a 500-W PV module, and up to 500-W
DC load are constructed to constitute the DC microgrid. The grid voltage of the
1,200-W SST is 120 V, and the DC link voltage is 200 V. When the PV generation
unit is operating with MPPT control, the maximum power is 500 W, and the cor-
responding MPPT voltage is about 28 V. The nominal battery voltage is 12 V and
the capacity is 20 A h. mw ¼ 31.83 is chosen so that VDC works from 180 to 220 V,
when w works from 59.9 to 60.1 Hz; mvg ¼ 0.005 is chosen so that E will work
from 114 to 126 V when the virtual synchronous machine based interface (VSMBI)
provides 100% to 100% reactive power to the grid. The constant mvDC ¼ 25 is
chosen, so that battery will operate from 180 to 220 V with 500-W discharge to
500-W charge power.
Q1 (V1,Q1)
Dead band
Capacitive
V1 100%
(nominal voltage)
% Available VARS (Q)
Percent
(V3,Q3) voltage
(V2,Q2) Increase
Inductive
V4
Q4 (V4,Q4)
Figure 6.26 Volt-Var droop control of SSSM applied in Figure 6.22

6.5.5.1 Case I: Load change

In this case, an 80-W resistance load is disconnected first, and then reconnected.
Figure 6.27 shows the results when Jvir is separately set at 1.06 and 4.24. The
system will take about 1.7 and 2 s to reach new operating points in Figure 6.27(a)
and (b), compared with 2 and 2.1 s in Figure 6.27(c) and (d). These results verify
the effectiveness and functions of the SST with the DC microgrid as SSSM in
generator mode.
6.5.5.2 Case II: Source change

In this case, PV output current changes from 2.5 to 0 A first, and then changes back.
Figure 6.28 shows the results when Jvir is set at 1.06 and 4.24. The system takes
about 1.6 and 1.2 s to reach new stable working points in Figure 6.28(a) and (b),
compared with 2.6 and 2 s in Figure 6.28(c) and (d). The system’s functions in
motor mode are verified.
6.5.5.3 Case III: Power up/down reserve

Figure 6.29 shows the results for 200 s when everything is in default settings.
w varies with the time, because wg is varying and w is synchronizing with it. The
frequency variation is normal in utility grid due to the customers’ load change, and
Pout Pout
CH1: 0 W CH1: 0 W
Load reconnected
CH2: ω CH2: ω
120π rad/s 120π rad/s
vDC vDC
CH3: 200 V CH3: 200 V
Load disconnected
CH4: 0 A CH4: 0 A
ibat ibat
(a) (b)
Time (s) Time (s)
Pout Pout
CH1: 0 W CH1: 0 W
Load reconnected
CH2: ω CH2: ω
120π rad/s 120π rad/s
vDC vDC
CH3: 200 V CH3: 200 V
Load disconnected
CH4: 0 A CH4: 0 A
ibat ibat
Figure 6.27 Experimental results of Case I: (a) Jvir ¼ 1.06, 80-W load changes to
no load; (b) Jvir ¼ 1.06, no load changes to 80-W load; (c) Jvir ¼
4.24, 80-W load changes to no load; (d) Jvir ¼ 4.24, no load changes
to 80-W load. CH1: Pout (500 W/div); CH2: w (0.08 p rad/s/div);
CH3: vDC (10 V/div); CH4: ibat (2 A/div); X-axis: time t (400 ms/div)
CH1: 0 W CH1: 0 W
Pout Pout
CH2: CH2:
120π rad/s ω 120π rad/s ω
vDC vDC
CH3: 200 V
CH3: 200 V
Source reconnected
CH4: 0 A CH4: 0 A
Source disconnected ibat ibat
CH1: 0 W CH1: 0 W
Pout Pout
CH2: CH2:
vDC vDC
CH3: 200 V CH3: 200 V
Source reconnected
CH4: 0 A CH4: 0 A
Source disconnected ibat ibat
Figure 6.28 Experimental results of Case II: (a) Jvir ¼ 1.06, 2.5-A source changes
to 0 A; (b) Jvir ¼ 1.06, 0-A source changes to 2.5 A; (c) Jvir ¼ 4.24,
2.5-A source changes to 0-A; (d) Jvir ¼ 4.24, 0-A source changes
to 2.5 A. CH1: Pout (500 W/div); CH2: w (0.08 p rad/s /div); CH3:
vDC (10 V/div); CH4: ibat (2 A/div); X-axis: time t (400 ms/div)
Pout
CH1: 0 W
CH2:
120π rad/s ω
vDC
CH3: 200 V
CH4: 0 A
ibat
Time (s)
Figure 6.29 Experimental results of Case III: CH1: Pout (500 W/div); CH2:
w (0.08 p rad/s /div); CH3: vDC (10 V/div); CH4: ibat (2 A/div);
X-axis: time t (20 s/div)
CH1: 0 W CH1: 0 W Pout

Pout
CH2: CH2:
vDC Soft-start vDC
CH3: 200 V CH3: 200 V
CH4: 0 A CH4: 0 A
Soft-start ibat ibat
Figure 6.30 Experimental results of Case IV: (a) reconnecting with battery
unit and load and (b) reconnecting with battery unit and current
source. CH1: Pout (500 W/div); CH2: w (0.08 p rad/s /div); CH3:
VDC (10 V/div); CH4: ibat (2 A/div); X-axis: time t (400 ms/div)
it is the reason why sometimes w and VDC do not work at the rated values in
Figures 6.27 and 6.28. Since the proposed system can synchronize its own fre-
quency with wg with a high degree of accuracy, it can also work as a phasor
measurement unit in the power systems. In Figure 6.29, VDC and ibat vary and
follow w according to the dual droop control. This verifies the effectiveness and
functions of the dual droop control in the proposed concept, supplying the power
system with up/down reserve when the frequency deviates from the rated 60 Hz.
6.5.5.4 Case IV: Islanding and reconnection

In this case, at first the AC grid is disconnected, SST stops working and the DC
microgrid keeps working. Figure 6.30(a) shows the transition for the system
reconnecting and synchronizing with the AC grid when only battery unit and
400-W load are present, and Figure 6.30(b) shows the transition when only battery
unit and 2-A DC source are present. Initially, ibat is about 1.95 and 2.1 A,
respectively, and VDC is about 185.5 and 218.5 V; the system takes about 1.4 and
1.3 s to reconnect, as shown in Figure 6.30. The achievement of power manage-
ment in DC microgrid when SST is not working and the proposed concept’s
synchronization capability are verified.
6.5.6 Summary
An SSSM concept is proposed to integrate the DC microgrid into legacy AC grid
through an SST. The benefits are (1) integrates the DERs, loads, and energy
storages in the DC microgrid into SSSM; (2) achieves power management for the
AC grid and DC microgrid, and enable the DERs and loads to output any power at a
given AC frequency; (3) introduces inertia, provides up-reserve and down-reserve
for the grid with support from practical energy storages, and decouples the AC
output from the fast responding DERs and loads. Experimental results verify the
effectiveness and the benefits of the presented method.
6.6 Conclusion
The SST-enabled DC microgrid concept is introduced in this chapter. Compared to
the traditional microgrids, the presented system enables a compact and integrated
interface for the DC and AC grids. Centralized power management strategy, hier-
archical power management strategy, and SSSM concept are proposed to control
the operation of the presented system. Simulation and experimental results are
given to demonstrate the feasibility of the proposed system.
References
[1] X. She, A. Q. Huang, and R. Burgos, ‘‘Review of Solid-State Transformer
Technologies and Their Application in Power Distribution Systems,’’ IEEE
J. Emerg. Sel. Topics Power Electron., vol. 1, no. 3, pp. 186–198, September
2013.
[2] X. She, A. Q. Huang, S. Lukic, and M. Baran, ‘‘On Integration of Solid-State
Transformer with Zonal DC Microgrid,’’ IEEE Trans. Smart Grid, vol. 26,
no. 12, pp. 3778–3789, June 2012.
[3] J. M. Guerrero, J. C. Vasquez, J. Matas, L. Vicuna, and M. Castilla,
‘‘Hierarchical Control of Droop-Controlled ac and dc Microgrids – A General
Approach Toward Standardization,’’ IEEE Trans. Ind. Electron., vol. 58,
no. 1, pp. 158–172, May 2011.
[4] X. Yu, X. She, X. Ni, and A. Q. Huang, ‘‘System Integration and Hierarchical
Power Management Strategy for a Solid-State Transformer Interfaced
Microgrid System,’’ IEEE Trans. Power Electron., vol. 29, no. 8, pp.
4414–4425, August 2014.
[5] M. Ashabani, and Y. A.-R. I. Mohamed, ‘‘Novel Comprehensive Control
Framework for Incorporating VSCs to Smart Power Grids Using Bidirectional
Synchronous-VSC,’’ IEEE Trans. Power System, vol. 29, no. 2, pp. 805–814,
March 2014.
[6] Q. Zhong, and G. Weiss, ‘‘Synchronverters: Inverters that Mimic Synchro-
nous Generators,’’ IEEE Trans. Ind. Electron., vol. 58, no. 4, pp. 1259–1267,
April 2011.
[7] D. Chen, Y. Xu, and A. Q. Huang, ‘‘Integration of DC Microgrids as Virtual
Synchronous Machines into the AC Grid,’’ IEEE Trans. Ind. Electron.,
vol. 64, no. 9, 7455–7466, September 2017.
Chapter 7
The load as a controllable energy
asset in dc microgrids
Robert S. Balog1,2, Morcos Metry1,2,
and Mohammad Shadmand1,2
7.1 Introduction
The electric power grid is the backbone of modern societies and their economies by
interconnecting electrical energy generation to loads via a vast transmission and
distribution network. Heralded as the greatest engineering achievement of last
century [1], the power grid in the United States and other countries is deteriorating
due to age and usage stresses [2]. In fact, it experiences a major outage every
decade that costs more than $2 billion, and on any given day, 500,000 customers
are without power for 2 h or more [3]. The traditional grid was not designed with
the renewable energy sources in mind and the reverse power flows that occur and
thus does not easily accommodate the transformation from consumer to prosumer.
Simply patching the old system beyond its original design and safety margins will
not alleviate the problems of power delivery congestion, feeder system voltage
profile regulation, reactive power, harmonic power, and the other challenges of the
way in which the grid is now being used.
One solution that does not require a $5 trillion complete overhaul of the
existing $1.5–$2 trillion [4] US electricity grid is to separate the functions of bulk
power delivery from point-of-load (POL) control. In this paradigm, we interconnect
loads and distributed energy sources and storage into a local area power and energy
system (LAPES) [5], sometimes called a microgrid [6], which is then connected to
the utility power grid or to other microgrids [7]. The Department of Energy defines
a microgrid as ‘‘a group of interconnected loads and distributed energy resources
within clearly defined electrical boundaries that acts as a single controllable entity
with respect to the grid’’ [8]. A classical perspective of ac power systems is a top-
down, generator-to-load flow of power in which utilities control the generation and
infrastructure and are required by a regulatory body to deliver power to the load.
1
Department of Electrical and Computer Engineering Texas, A&M University, USA
2
Department of Electrical and Computer Engineering, Kansas State University, USA
The utility has no control of the load, which imposes operational constraints on the
systems. In a LAPES, the load as well as distributed sources and storage can all be
controlled, which changes the paradigm of generation following load into a para-
digm in which control of the load itself can be thought of as a controllable energy
asset [9]. The paradigm further changes as the control moved from centralized to
distributed.
In the past, some utilities have had load-shed programs in which customers
relinquish some limited control of their load. Such a system was most commonly
used for peak load shaving in which customer-owned air-conditioning systems were
fitted with cut-off switches that responded to a command from the grid or system
operator, as explained by Commonwealth Edison Company of Illinois [10]. Thus,
shedding individual load to match available generation capabilities or congestion
constraints to avert a blackout is one way the load can be considered as an energy
asset [11]. Load shed has also been considered for ac systems where both voltage and
frequency stability were a concern and implemented with frequency-activated load
shedding [12]. Less aggressive approaches that have been considered include con-
servation voltage reduction in which the voltage on a feeder system is lowered to
reduce power consumption, which can work to some degree when the loads are
mainly resistive in nature [13]. Market-based incentive programs take a slightly
different approach and attempt to encourage customers to modify their power con-
sumption based on adjusting the price of electricity. In some cases, a smartphone app
is used to signal cost savings that may be possible by voluntarily reducing con-
sumption, or offer an incentive, such as a coupon, for volunteering to shed load
[14,15]. However, such initiatives depend on the active participation of the customer.
While it raises public awareness of energy consumption, it may not be reliable for
peak-time load shaving to avert a crisis if the customer does not behave as expected.
Fundamentally, the challenge in any electrical power system is that of just-in-
time delivery: supply must balance consumption. If the system is congested, or
generation is already at peak, one way to relax the constraint is to introduce energy
storage. Pumped hydro can be cost effective but depends on geography and thus is
not a universal solution, just like compressed air storage which depends on vast
underground storage. Battery-based storage can be more arbitrarily located in the
utility system as needed. Examples of grid-scale storage include the Younicos cube
or the Tesla power pack [16,17]. Another approach is to tap into a growing market of
residential-scale storage element such as the Tesla power wall located in individual
houses [18]. However, while adding storage can decouple supply from demand, it
also adds cost.
7.1.1 Local area power and energy system

Consider the LAPES microgrid shown in Figure 7.1. From the perspective of the
legacy power system, a LAPES would exist at the grid-edge and would incorporate
distributed renewable energy sources such as solar photovoltaics and wind, energy
storage, and load. The LAPES would also provide a point of common coupling
(PCC) for all of the individual components to the grid-connection. The LAPES
The load as a controllable energy asset in dc microgrids 155
Electric
power grid
High-level Load and PV module
Wind PCC
controller
generators
(e.g., Energy PEI
management)
PEI
PV PEI Local area
modules power and
energy system PEI
Source-side
energy storage
Electric PEI
vehicles PEI PEI
PEI Power electronic interface Distributed

energy sources
Energy line Control line
Central energy
PCC Point of common coupling storage
Figure 7.1 One-line system diagram for a local area power electronic system
(LAPES) that integrates community-scale energy storage with
distributed solar generation colocated with the electrical load. The
system is interconnected with the electric utility at a single point of
common coupling (PCC). Sizing and control of the energy storage
unit, the ‘‘Central Energy Storage’’ can enable arbitrary power flow
profiles at the PCC to the grid
would not merely supply power to the load but would also interact with the
loads, sources, and storage. In such a paradigm, the operation of the load, dis-
tributed sources, and distributed storage can respond to the LAPES system to
ensure stable operation and even participate in energy market opportunities
upstream in the utility power grid by controlling power flow at the PCC. Without
loss of generality, examples of a LAPES include a neighborhood or subdivision of
houses, an industrial park, a shopping center, a school campus. In these examples,
the central energy storage shown in Figure 7.1 would be a common resource used
by the entire system, much like a community lake is a common resource used for
firefighting or a retention basin is used for flood control.
In this chapter, we consider a LAPES as a dc microgrid system. Virtually, all
modern loads are dc, and dc systems offer efficiency and control advantages over
ac for distribution networks and integration of renewable sources and distributed
storage [5]. The focus of this chapter is on control of the load so that the system can
view the load as an energy asset rather than a liability. In other words, view the load
as an integral participant in the overall power and energy balance of the microgrid.
This is done by exploring how POL controllers combined with modulation of the
bus voltage can control the power and energy consumption of the load in the system
without needing expensive communications.
7.2 Why control the load?

7.2.1 Benefit of load control
The challenge with quality assurance is that everyone wants it but nobody
wants to pay for it. [19]
The practicality for any engineering solution is determined by the ability to find a
trade-off between cost and benefit. Power interruptions inevitably do occur in the
microgrid, and most of the discussed solutions to these problems are profoundly
centered around adding generation and storage capacity to guarantee unin-
terruptible power supply. The approach in this chapter considers the load as a
controllable energy asset that is included within the cost and benefit trade-off for
system design and operation. Are all loads always necessary? Can some loads be
reduced instead of being turned off completely? The answers to these questions
provide an economic justification and benefit.
A Pareto Frontier is a tool that is usually used in engineering and economic
applications to visualize the most efficient set of solutions to a multiobjective
optimization problem [20]. Consider a design in which there is a trade-off between
two parameters, such as converter efficiency and cost. When we consider all of the
possible designs in the design space of these two parameters, we find the Pareto
Frontier as the set of nondominated designs. In other words, for a give value of
parameter 1 (efficiency, for example), the nondominated design is the one that
achieves this efficiency at the lowest cost. Thus, all points on a Pareto Frontier are
considered to be Pareto Efficient for the particular objective function. Pareto
Efficient points are the only solutions that guarantee satisfying all the assigned
objectives without making any particular objective worse off [21].
The benefits of load control can be observed by considering the set of equally
good solutions, the Pareto Frontier, for the optimal multicriteria design of the
hybrid microgrid energy system in Figure 7.1. The Pareto Frontier can be generated
by using optimization methods for hybrid energy systems such as nondominated
sorting genetic algorithm (NSGA-II) [22]. Figure 7.2 illustrates the set of optimal
solutions given by the Pareto Frontier to maximize the energy availability and
minimize the cost associated with energy storage requirement of the hybrid
microgrid. The energy availability, the fraction of the time when energy is avail-
able, is a key figure of merit of a hybrid microgrid system. It is important to make a
clear distinction between reliability and energy availability. The reliability is the
ability of the system to operate without failure; energy availability is the ability of
the system to supply power to the load. As an example, a highly reliable renewable
energy system, where the components are not prone to failure, can have low energy
availability if there is insufficient energy storage to support the power requirements
of the load during the night or during an overcast day. Thus, a specified level of
energy availability can be achieved with many configurations of a microgrid sys-
tem while minimizing the system size to have a financially feasible system. This is
the equilibrium point of the techno-economic optimization.
Modulated Modulated
Load at: Load at:
30%
20%
10%
91 91
30%
20%
10%
0%
0%
Reduction in
92 92 energy availability
due to failure in
93 93 portion of energy
Energy availability (%)
Energy availability (%)

94 94 storage bank
Energy storage size/cost Modulating the load to
95 95
reduction by increase the energy
96 modulating the load to 96 availability for reduced
maintain energy energy storage size
97 97 B
availability
98 D C B A 98 A
B’
99 99 B”
B”’
100 100
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1
Required energy storage (p.u.) Required energy storage (p.u.)
(a) (b)
40
Load modulation (%)
30 D
C Energy storage capital

20 cost reduction by
modulating the load
10 B
0 A
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1
(c) Capital expense (p.u.)
Figure 7.2 Pareto-frontier reveals the set of optimal solutions for the hybrid
system [22]. (a) Load control benefits to design: energy storage size/
cost reduction by moving from point A to D (modulating the load from
0% to 30% while maintaining the energy availability of the
microgrid), (b) load control benefit to control: portion of the energy
storage bank is failed, thus energy availability will be decreased by
moving from point A to B due to this failure; the energy availability for
the reduced available energy storage size can be increased by moving
from point B to B000 (load modulation from 0% to 30%), (c) load
control benefit to capital cost of the energy storage system
Consider a LAPES that contains PV generation and storage. The amount of PV

generation and energy storage determines the energy availability for particular load
[23]. Figure 7.2 illustrates the trade-off between the cost and energy availability.
Consider that individual loads can be controlled to reduce power consumption.
Without loss of generality, one may think of dimming a light or slowing a heating,
ventilation, and air conditioning (HVAC) blower motor. The base-case is the micro-
grid system without load control, illustrated as 0% load modulation. This is compared
to systems in which the load can be modulated to reduce power consumption from
10% to 30%. This can be accomplished by sending a signal to each dc/dc or dc/ac POL
converter. Two scenarios are considered that reveal the benefit of load control
(modulation) in a microgrid system: one scenario encompasses the gained benefits
from proper storage and load modulation ratios of the microgrid systems, while the
other displays strategies for failures in portion of energy storage bank.
The first scenario is illustrated in Figure 7.2(a). This figure shows that if
point A is selected as the final optimal solution of the base-case based on particular
preferences to achieve 98% energy availability, the required energy storage will be
0.85 p.u. This configuration requires bulky and expensive energy storage to com-
pensate the stochastic behavior of renewable energy resources and/or to meet the
peak load at 0% load modulation (an uncontrolled load). The load modulation
could refer to the fraction of total load that is shed or the percentage of power
reduced in individual load for a period of time. For example, dimming a light or
changing the set point on a thermostat represents load power modulation that is a
different control paradigm than simply turning off the load or HVAC system which
may balance providing benefit to stabilize the grid while also minimizing impact to
the end-user. By modulating the load from 0% to 30% (i.e., moving from point A to
D), the required energy storage system requirement is reduced significantly from
0.85 to 0.55 p.u. as shown in Figure 7.2(a), while maintaining the same energy
availability levels. This scenario demonstrates the benefits of the load control to the
design and sizing for the microgrid system.
The second scenario is illustrated in Figure 7.2(b) where the system was
initially designed for operation at point A. Then, portion of the energy storage bank
is failed, causing reduction of energy availability from 98% to 97% (i.e., moving
from point A to B). The advantage of the load modulation comes into play in this
scenario; by modulating the load at 10%, the same energy availability of 98% could
be maintained with a failed storage element. Further energy availability improve-
ment could be achieved by modulating the load further. For example, at 30%
modulation, the energy availability reaches 99.99% (shown by moving from point B
to B000 ) at an energy storage size of 0.72 p.u. Thus, the function of load modulation, in
this example, increased the energy availability by 1.9% and decreased the energy
storage size by 0.13 p.u. in comparison with the originally designed system that did
not have controllable load considered initially. Thus, the load control can improve
the system controllability for maximizing the energy availability as well as com-
pensating the consequences of energy storage device failures in a typical microgrid
system.
It is worth mentioning that moving from point A to D in Figure 7.2(a) is a
capital expense saving at the design stage because less energy storage is required.
The effect of load modulation on the capital cost of the energy storage system is
illustrated in Figure 7.2(c). As is illustrated, for example, if lights are dimmed to
modulate the load to 30%, the capital cost of the system will be reduced to 0.55 p.u.
at point D. However, if the loads are not modulated, the capital cost of the system is
0.85 p.u. at point A. The capital cost of 0.85 p.u. demonstrates the cost of an
optimal system design at 0% load modulation, and the capital cost of 0.55 p.u.
demonstrates the cost of an optimal system design at 30% load modulation. This
demonstrates the impact of load modulation on the end user in terms of pro-
ductivity, capability, and economic model because the end user would expect
financial compensation for the reduced performance or reduced capability.
7.2.2 Is load modulation practical?

Figure 7.3 shows the generation inertia of the ERCOT (Electric Reliability Council
of Texas) utility grid for the state of Texas during the month of April 2014 along
with the corresponding load demand [24]. For the most part, the inertia seems to
meet and surpass the actual load for that month. This is because nuclear and coal
generation inertia is scheduled based on historical demand and forecasted weathers
and fast peaking supply, along with spinning reserves ensure rapid response to
unpredictable changes in load. However, microgrids that rely on renewable energy
sources, particularly solar energy, along with other sources may not have such
flexibility, particularly during times of peak load. During peak loads, microgrids
have multiple options: generating more energy, importing energy from the main
grid or from other microgrids, shedding the load, or modulating the load. There are
many scenarios; the first option of generating more energy is not possible physi-
cally because the system is already at maximum generation capacity. The second
option, importing energy from the grid or other microgrids, may be unfavorable
economically because of the exorbitant cost from purchasing power on the volatile
spot market. The third option, load shed, may be unfavorable since customers
would be inconvenienced from the power interruption. Load modulation, on the
other hand, requires a closer look at the characteristics of loads in the microgrids.
Let us consider lighting, HVAC and electric/plug-in hybrid vehicles as examples of
controllable loads.
2016 April
Inertia provided by Nuclear Inertia provided by Coal Inertia provided by Simple Cycle Inertia provided by Combined Cycle Load
300,000 60,000
250,000 50,000
200,000 40,000
Inertia (MW*s)
Load (MW)
150,000 30,000
100,000 20,000
50,000 10,000
0 0
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
-0 01 00
16 0
00
20 :0
03 4-2 6 0:
04 4-2 6 0:
05 4-2 6 0:
06 4-2 6 0:
07 4-2 6 0:
08 4-2 6 0:
09 4-2 6 0:
10 4-2 6 0:
11 4-2 6 0:
12 4-2 6 0:
13 4-2 6 0:
14 4-2 6 0:
15 4-2 6 0:
16 4-2 6 0:
17 4-2 6 0:
18 4-2 6 0:
19 4-2 6 0:
20 4-2 6 0:
21 4-2 6 0:
22 4-2 6 0:
23 4-2 6 0:
24 4-2 6 0:
25 4-2 6 0:
26 4-2 6 0:
27 4-2 6 0:
28 4-2 6 0:
29 4-2 6 0:
30 4-2 6 0:
0:
4- 6 0
-0 01
02 4-2
-0
01
Figure 7.3 Example of generation inertia and load consumption data available
from ERCOT for the state of Texas. The figure shows the daily cycle
which changes over the month of April in 2014 [24]
Consider the peak-time load consumption which occurs during daylight hours.
Since lighting loads represent 30% of the electrical consumption in commercial
buildings, modern buildings are designed with controllable lighting systems for
energy conservation and adjust artificial lighting based on the amount of sunlight
entering the building [25], occupancy, and usage requirements. More windows
unshaded means a lower illuminosity foot-candles (ftcd) is required from the artifi-
cial lighting system. Standards recommend office spaces to have an illuminosity of
46 ftcd [26]. Hence, sunlight exposure from the office windows could reduce light
consumption. While this might seem like a good idea in principle, tenants of the
office complain about sudden switch off of lighting during direct sunlight, delayed
turn on of the lighting when the room is darker, and a continuous switching on and
off of lights on cloudy days [27]. Such concerns mean that in many workspaces,
these ‘‘intelligent’’ systems are bypassed; hence, lights remain at their full illu-
minosity the whole day, impeding the energy saving opportunity and missing the
opportunity to control the load as an energy source. As many lightbulbs come now
with dimming features, lighting modulation could be achieved by adjusting the
brightness rather than turning on or off certain light fixtures. This is achieved by
giving some load modulation control to the grid. After all, a user can barely distin-
guish the difference between a 46-ftcd and a 35-ftcd room [28] yet when taken at the
scale of an entire building over a period of time can provide enough energy to have a
meaningful impact on the stability and energy availability of the microgrid. Hence,
lighting in commercial buildings is well suited as a controllable energy asset.
Consider a building heating, ventilation, and air conditioning system (HVAC).
A building has a large thermal mass so adjusting blower speeds or temperature set
points on a short enough time bases may have minimal impact on the occupancies
perception but frees up energy. Typical HVAC systems have a control approach
similar in principle to the bang-bang controller, in that there is a 2 C upper and lower
limits around the set point and the HVAC compressor and fan will work when the
temperature exceeds the upper limit and turns off when the temperature goes under the
lower limit. Such systems could be thought of as a manageable load asset, in that
the occupants of a room would barely be able to tell if the temperature went up to 3 C
above the upper limit during times of contingencies and would not notice the lower fan
speed intensity. Such initiative takes place in the middle of the day, during peak
consumption time, assuming the HVAC system has been on all day and has already
cooled the space. Many companies have considered demand side energy management
in which the HVAC unit could be controlled by the grid. Edison Commonwealth of
Illinois [10] has the initiative of Central AC Cycling for homeowners, which offers the
installation of direct load control switches or utilizing a smart thermostat to cycle the
load during peak time by turning off the compressor when needed, up to 15 min every
30-min interval. Applying such system on large commercial and government build-
ings such as schools and offices could add a flexibility of manageable load assets.
Consider electric vehicles (EV) or plug in hybrids (PHEV): if all the owners
care about is having a fully charged battery ready for the next trip, he or she could
set the expected time the car will next be needed by using a smart phone app [29]
when parking the car. From the driver’s perspective, the rate at which energy is
delivered to the car battery is meaningless as long the battery is fully charged and
ready for service when the user returns [30]. Thus, EVs become a controllable
energy asset by scheduling unnecessary EV charging off peak times. In 20 years or
so, when EV become the mainstream technology, utilizing such load could be
significant. For example, a parking garage next to a hospital with a 1,000 cars
capacity could have enough energy storage, using vehicle-to-grid protocols [31],
for one of the hospital’s building segments (e.g., operation rooms or intensive care
units) during emergency power deficiencies.
Numerous application examples can be envisioned, ranging from civilian to
military, terrestrial to outer space, to personal power networks, in which load can
participate in the overall control and operation of power and energy management of
the dc microgrid system. It may be easy to visualize the imperative for load control
if the microgrid is stand-alone and energy constrained. The need also exists, how-
ever, for grid-connected dc microgrids, like the one in Figure 7.1, in which the
operator desires to shape the load profile at the PCC for reasons of economics or
utility-side stability. For example, the microgrid can take advantage of market
clearing prices to control local energy assets [32]. This may be obvious for the
cycling of energy stored in battery, but we can also modulate a load to achieve
similar effects. Therefore, there are many considerations to be taken into account
during the design of the microgrid. During the design stage for the energy storage,
sizing of the storage is determined by assuming fully modulated load to size the
storage based on worst case availably requirement. During operation, energy sto-
rage must be first used as primary energy asset and then the load modulation is
secondary. Energy tapping opportunities from the load should be categorized in a
way that ensures the minimum cost. For example, tapping into EV batteries could
potentially result in higher battery maintenance cost than reducing the fan speed of
an HVAC system which could only cost a little convenience.
7.3 Time-scale of energy requirements

Not all power interruptions have the same impact and may require different strategies
to mitigate or minimize disturbance to the system. Figure 7.4 illustrates a time con-
tinuum of events that may be experienced on an electrical grid The timescale could be
as fast as thousandths of a second for power electronics stability, subsecond for elec-
trical network stability, tens of seconds for automatic generator control, minutes to
hours for economic dispatch/optimal power flow, diurnal (daily) for solar and wind
power, and seasonal for heating and air-conditioning systems. Such continuum of
interruption timescales makes it impossible to tackle such variability with only one
strategy. Hence, it is crucial to divide the different interruptions based on the range of
their time scale (long or short) and plan for the appropriate contingency plan.
7.3.1 Short-term transients

One method to maintain stability through short-term disturbances is to reduce the
power to the load—appropriate for lighting or inertial loads where the momentary
Automatic
generator control
of grid Diurnal load
and
PV curves kW h/day
for PV
Economic
Electrical stability dispatch
of system of grid HVDC
load
Subsecond Seconds Minutes Hours Days Seasonal

duration
Power Energy
management management
Figure 7.4 An illustration of the timescales of potential load interruptions
Power buffer
Zbus
Point
+ + of
Vbus Rin – Load
– load
converter
Estorage
Figure 7.5 Power buffer for a dc system
slowing of fans and pumps or dimming of lights may be an acceptable alternative to

system-wide instability. However, many modern electronic loads and other sensi-
tive loads do not lend themselves to this technique. A more general method is to
implement an active dynamic buffer as an interface between the power system and
the POL converter as shown in Figure 7.5.
A power buffer decouples the load from the bus dynamics [33]. The rating of
the local energy storage device provides the designer with a degree of freedom to
choose the extent of the transients through which the load can be sustained. Once
the local energy has been depleted, however, continued operation of the load is no
longer possible. This gives rise to the notion of time scales based on the energy
storage in the power buffer.
The power buffer has several modes of operation. During normal system
operation, the power buffer supplies the load by drawing power directly from the
bus. When a system transient occurs, the buffer senses the system voltage sag and
presents constant impedance to the bus while continuing to supply the load with
constant power. Since less power is drawn from the bus during the constant
impedance mode, internal storage is required to maintain the power requirements of
the load. After the transient passes, the buffer returns to a power regulation mode
and draws additional incremental power to recharge the buffer capacitor. In effect,
a power buffer stretches the time scale of the transient, diminishing the impact of
tight converter regulation.
A power buffer is limited by the amount of stored energy available in the bus
capacitor. The time that a power buffer can maintain constant input impedance
while continuing to supply full power to the load is called its sustaining time and is
defined as
2 2
Cbuffer Vload0 Vload1
2
Vbus0
Tsustain ¼ 2 (7.1)
2Plaod Vbus0 Vbus1
2
where Vload0 is the nominal load voltage, Vbus0 is the nominal input voltage, Vbus1 is
the sag voltage, and Pload1 is the load power. The buffer design parameters in the
sustaining time are Vload1 (minimum allowable load voltage) and Cbuffer (energy
storage capacity).
A plot of a typical sustaining time versus voltage sag for a buffer supplying a
100 W, 400 V load from a 100 V distribution system is shown in Figure 7.6. As
long as the voltage sag magnitude and duration fall above the curve, the power
buffer can successfully ride through the transient, presenting constant impedance to
the power system bus while maintaining constant power to the load. If the transient
event begins to approach the sustaining time limit, then the local load control needs
to switch strategies to maintain stability.
7.3.2 Long-term transients

System transients that exceed the sustaining time of the power buffer (or that are
caused by topological failures such as loss of generation or a bus fault) require an
alternative technique to mitigate system instability and voltage collapse. In this
1
Ride through capability
0.8
Bus voltage [p.u.]
0.6
Insufficient energy storage
0.4
220 µF
0.2 470 µF
1,000 µF
0
10–2 10–1 100 101 102
Time (s)
Figure 7.6 Sustaining time capability

case, the only long-term strategy to stabilize an energy-constrained dc system is to

switch to another supply or to load shed.
In practice, dc–dc regulators have minimum allowable input voltages that
prevent them from operating as true constant power loads (CPLs). As the system
bus voltage decreases due to increased loading or loss of generation, these con-
verters will turn off to self-protect as the bus voltage decreases below this specified
voltage limit. In a radial system, the voltage drop along the bus due to bus impe-
dance automatically gives rise to a notion of priority to the loads located closest to
the source and hence with the highest bus voltage. Thus, the first converters shed
due to under-voltage protection (UVP) are the ones physically farthest from the
source. This behavior directly couples the system topology to the load priority. In
general, it is not desirable that the topology dictate priority. Instead, a supervisor
control at each POL converter monitors the bus voltage and turns off the converter
based on the priority of the load. Mapping the priority setting to a particular bus
voltage forces loads with the lowest priority to be shed first and allows higher
priority loads to remain energized, regardless of location in the system. Thus,
system operation is decoupled from system topology. A similar technique has been
proposed for ring-bus architectures.
7.4 Autonomous load control

Direct current power systems have long been the standard architecture for many
applications where reliability is critical, like telecommunication and naval ships.
dc Power systems allow for more straightforward integration of renewable energy
sources into the grid. Modern high frequency dc–dc converters have enabled more
efficient voltage conversion at higher power densities than the AC 60-Hz converters
could [34]. dc Systems are better fitted for energy storage integration to improve
reliability and energy availability of distributed systems. Also, many concerns that
accompanied ac systems such as synchronization requirements for multiple small
generation sources, reactive power flow, and circulating current due to differences in
voltage magnitude, phase angle, or dc offset in a multilayer system, do not exist in dc
systems [35]. The paradigm shift from the common ac system to a dc system facil-
itates easier control of individual load performance and coordination, especially as
energy allocation priorities change to match load priority requirements.
7.4.1 Control
The generally adopted control hierarchy of the dc microgrid is of close resemblance
to that of the conventional grid, in that it is stratified into tertiary, secondary, and
primary control levels [36]. The highest level controller, the tertiary level,
responsible for economic dispatch and coordination, assigns the microgrid voltage
that is appropriate for power exchange between the microgrid and the main grid
[37]. Such function is communicated to the PCC, as shown in Figure 7.1. The
secondary controller updates the voltage set points for the primary controllers to
match the voltage demands of the secondary controller. Like the tertiary controller,
the secondary controller is carried out at the high-level controller. Then, the signals
from the secondary controller are sent to all the power electronics interface (PEI)
units as illustrated in Figure 7.1.
Implemented locally at each PEI, the primary controller regulates the voltages
for the individual converters to facilitate load sharing among sources. The control
structure of the primary controller based on distributed control [35] is inherently
different from the secondary and tertiary controllers that are based on centralized
control. Accordingly, secondary and tertiary controller units are prone to higher
risks of failure and instability in the event of losing any links in the topology [34].
On the other hand, distributed control provides higher reliability as each PEI unit
controller is independent in operation from failures in other units. It also means that
systems based on distributed control are easily scalable [38]. A detailed illustration
on the differences between centralized and distributed control is illustrated when
studying the difference between active current sharing (centralized control) and
droop control (distributed control) in Table 7.2.
7.4.2 Architecture
As has been shown in Figure 7.1, a LAPES microgrid system incorporates renew-
able energy generation systems such as solar photovoltaics and wind along with
storage elements and EV. A LAPES microgrid is not merely a supplier of load to its
load but goes beyond by using residential scale generation on the load when
available. Without loss of generality, the loads could be residential houses, office
buildings, hospitals, university campuses, etc.; within each building, there could be
several distribution and load systems as illustrated in Table 7.1.
A high-level energy management system provides signals to the buildings on
power and energy usage/setting, and then the PEI for the building turns that setting
into a control response. A case scenario could be to balance generation to meet the
load; however, there is a point when generator control is no longer possible to
supply loads from generation alone. Hence, the need to tap into the energy
resources of the load itself. Consider a case study of a residential house with an ac
power system and that has different types of load as shown in Figure 7.7. One could
observe the energy storage potential in some of the loads (i.e., the UPS of a com-
puter system) and utilize that for the overall system stability. Moreover, a load
like residential lighting may be modulated/dimmed to reduce load demands while
keeping the lights to a comfortable illumination level. Such control actions may
Table 7.1 Summary of the different loads used in different buildings
Residential Office space Retail outlets Hospitals University Public spaces

campus/schools
Lighting Lighting Lighting Lighting Lighting Street lighting
HVAC HVAC HVAC HVAC HVAC Security
Consumer Consumer Refrigeration Medical Security
electronics electronics devices and safety
Kitchen Telecomm/ Telecomm/ Telecomm/ Telecomm/
devices datacom datacom datacom datacom
Fluorescent
Micro M lighting
turbine
Consumer
AC
ac electronics AC
ac AC
ac
Water DC
dc HVAC DC
dc DC
dc
Electronic
heater motor ballast
DC
dc DC
dc drive DC
dc DC
dc
AC
ac AC
ac AC
ac AC
ac
60 Hz
ac
AC
ac AC
ac AC
ac AC
ac
DC
dc DC
dc DC
dc DC
dc
PCC
UPS
AC DC
dc
grid dc
DC
Solar Solid-state
Electronics
lighting
dc
DC
AC
ac
AC
ac
DC
dc
Computer
Figure 7.7 An example of a residential ac building with different loads. Utilizing

different energy storage potential in the load could be used for load
modulation
be achieved using a configuration like that of the single bus radial system of Fig-
ure 7.8 connected to the LAPES microgrid of Figure 7.1. The details of the control
methods are explained in the following subsections.
7.4.3 Strategy for controlling the load to be an energy asset

Control strategy for each POL bus interface when the system is stressed depends on
the load priority [9]. High-priority loads are required to continue operation while
lower priority loads can be turned off to preserve the systems voltage stability.
Also, the timeframe of the need for energy determines the type of control used:
short term or long term. For short term, a power buffer decouples the dynamics of
the load from the dynamics of the dc bus. For long term, modulating the load
consumption balances energy in the system. A power buffer on the higher priority
loads eases the burden on the bus by presenting constant input impedance instead of
a CPL during a buffering event. In this mode, current is still drawn from the bus and
internal energy storage satisfies the load requirements. Figure 7.9 illustrates the
decision logic for a complete POL power unit.
+ Input
bus
V dc–dc
– bus
R
filter
us
POL converter #3
Lb
+ Input
bus
V dc–dc
– bus
R
filter
us
POL converter #2
Lb
+ Input
bus
V dc–dc
us
R – bus filter
POL converter #1
Lb
Dc source +
converter –
Figure 7.8 Single bus radial system with three loads [39]
Normal operation
Unbuffered load dynamics, monitor

bus voltage for transient event
Transient event occurred
High Low
Load priority
Medium
Power buffer Power buffer Load shed
Supply the load until the Supply the load until the Monitor the bus to see if the
local energy reserve is local energy reserve is load can be turned back on
depleted depleted or the bus
continues to deteriorate
Startup
Control input impedance to

Replenish energy storage limit inrush current
Monitor bus voltage for

transient event
Figure 7.9 Local control strategy for each POL supervisor
During normal operation, the buffer transfers the load power demand from the
bus, while maintaining its internal energy storage. This mode completely couples
the load and bus dynamics. When the bus experiences a transient that triggers a
protection event, load priority determines the course of action to maintain system
stability. The highest priority loads remain connected to the bus until the energy is
depleted. Lesser priority loads continue to monitor the bus and disconnect if the bus
is sensed to become worse according to some metric.
For high-priority buffer loads, the wellbeing of the load is favored; therefore,
the buffer will attempt to maintain the load until its internal energy storage is
depleted, while presenting a constant input impedance to the bus. When the internal
energy reaches a given set-point, the POL switches to a load-shed strategy. For a
medium-priority buffer load, the loads welfare is favored less. Throughout a pro-
tection event, the input impedance remains constant, while stored energy supple-
ments the load power. However, if during the protection event, the bus condition
becomes worse, the strategy is switched to a load shed.
When the bus recovers from a transient, the high and medium priority buffers
change to replenish energy storage mode in anticipation of the next event. While
drawing full load power from the bus to supply the needs of the load, additional power
is drawn to recharge the buffer energy storage capacitor. If during the replenish cycle
the bus experiences another transient, then the buffer reenters a constant input impe-
dance mode. Since the replenish cycle was interrupted, the sustaining time for the
latest transient will be diminished as there is less stored energy. If the replenish cycle
finishes uninterrupted by a bus transient, then the buffer returns to normal operation.
When the load is low priority, or the buffer energy storage has been depleted in
a high or medium buffer, load shed is implemented. This entails shutting down the
load in a manner that will cause the least inconvenience for startup and minimize
the impact on the load. The strategy is a function of the nature of the load.
When the protection event has cleared, a load-appropriate startup strategy is
implemented. To minimize the chance of triggering further system transients, the bus
power to the load should be minimized during startup, so-called soft-start. When the
load is high priority, power is immediately delivered to the load. When a low or
medium-priority load is started, the buffer energy storage capacitor is precharged
before turning on the load. During startup, load priority determines if the load is
immediately connected or if the power buffer starts up first. Highest priority loads
immediately connect to the bus, while lower priority loads first allow their power
buffer to soft-start, eliminating inrush current and gracefully loading the bus with an
initial constant impedance to help stabilize the system. After a time, the power buffer
will revert to normal operation. If a bus transient occurs during charging of the
buffer, then the charging cycle ends until the bus returns to its nominal state.
7.5 Droop control for stability and information communication
It is common in dc microgrid systems to have multiple generation sources, such as

photovoltaics, and multiple storage elements, such as batteries, supplying one dc bus.
Thus it is critical to have techniques to manage current sharing. Ideally, supply
converters will identically share the current amongst them; however, mismatches in
components and feedback networks as well as different impedances at different
locations on the dc bus can cause imbalance in current sharing. Such imbalance,
when significant, could result in overload and thermal stresses that jeopardize the
system reliability [40]. The common methods of current sharing used in the literature
fall into two general groups: active sharing and droop control [41].
Active current-sharing techniques involve a control structure and a method of
programming individual converters with a reference current. One implementation is
to use a master/slave configuration such that one dc source is designated as the
master and is used to control the bus voltage. The remaining dc sources, designated
as slaves, operate as current sources. This strategy produces a stiff bus voltage and
controlled load dispatch at each source. There are two main limitations of this
technique: high-speed communication is required and a single point failure can
disable the entire system. In practice, active current sharing techniques are best
suited for physically small systems, such as paralleled voltage regulator module
applications. If the topology were fixed and known a priori, more sophisticated
controls such as interleaving can be used to reduce ripple. In droop control [42], the
output voltage of the source drops as current increases. This is a form of local control
since converters autonomously share load current by sensing the local bus voltage.
Droop control can be as simple as a series resistance or a more efficient closed-loop
controller such as a phase-angle controller in a rectifier source converter [32]. This
scheme has been proposed for use in large-scale distributed systems with dynami-
cally changing topologies since it supports plug-and-play reconfiguration and system
scaling and is robust to component failures. A summary of the differences between
active current sharing and droop control techniques is as shown in Table 7.2.
7.5.1 Constant power load and its deleterious effect on dc systems

Similar to the ac grids, dc microgrids encounter stability issues that arise from the
voltage regulation of PEIs which have become a necessity in the integration of
Table 7.2 Comparison between load sharing techniques: active current sharing
and droop control
Active current sharing Droop control

Involves a current structure and a method The output voltage of the source drops
of programming individual converters as current increases, making it a local
with a reference current autonomous control
Requires communication channels Does not require communication channels
Principle: Master/slave Principle: Virtual resistance
One dc source acts as a master to control Implemented as a series resistance or a
bus voltage, while the rest act as slaves more efficient closed loop controller
operating as current sources
Plug and play reconfiguration not possible Plug and play reconfiguration possible
Single point of failure Robust distributes system
Suitable for small-scale embedded systems Suitable for large-scale distributed systems
sources, loads, and storage devices. In fact, switching dc power converters is inher-
ently nonlinear, that interconnecting a large number of them poses stability issues to
the whole microgrid. POL converters act as instantaneous CPLs, as they regulate the
load voltage too tightly [43]. The tight POL voltage regulation results in negative
dynamic input resistance which leads to destabilizing effects during a voltage sag.
CPLs have become very common in various applications related to microgrid inte-
gration such as in data centers, telecommunication, wireless communication base
stations. Hence, the need for an effective solution to the destabilizing effect during
voltage sags, which are quite common due to the events of loss of generation or
increase in load. Since finite dc systems are inherently weak, such CPL systems are
subject to extreme voltage sags or even voltage collapse.
Many solutions are proposed in the literature [43,44], which include the addition
of filters, the addition of bulk energy storage devices that are directly connected to
the system main bus, load shedding, the addition of linear controllers and the addi-
tion of boundary controllers. This section reviews the benefits of droop control and
load modulation to remedy the destabilizing effects that could accompany CPLs.
7.5.2 Steady state stabilization

In droop control, the output voltage drops as the current increases. Converters share
load current by sensing the bus voltage and increasing current as the bus voltage drops.
The model for droop-control based on sensed bus voltage, taken from, is shown
in Figure 7.10 and incorporates the effects of finite bus impedance Zbus and source
output-capacitance Cs. Variables Is and Vs are the current and voltage at the output
terminals of the converter, respectively. The voltage at the output, the bus voltage, is
low-pass filtered and used to close a feedback loop. The droop gain K converts the
voltage error into a current command for the source converter. Assuming the converter
current perfectly tracks the reference current, the steady-state droop relationship is

Is;ref ¼ K Vref vs;sense (7.2)
This scheme has been proposed for use in large-scale distributed systems since it
does not require any communication between the dc sources. The distributed sys-
tem is inherently robust because droop control automatically shares current among
1
Zbus + Zload
Is,ref Is –
Converter 1
Vref + K + Vs
– Gconv(s) sCS
ωLP
s + ωLP
Figure 7.10 Droop-controlled voltage source

V V V V V
Voc
Vop1
Vop2
Vop3
I I I I I
Iop1 Iop2 Iop3 Imax Iop1 Iop2 Iop3 Imax Iop1 Iop2 Iop3 Imax Iop1 Iop2 Iop3 Imax Iop1 Iop2 Iop3 Imax
Figure 7.11 Current sharing using droop-control
Table 7.3 Table of operating points
Number of sources
5 4 2
Vop (V) 47.75 46.54 38.97
Iop (A) 5.672 7.273 17.38
Ibus (A) 28.36 29.09 34.75
Pload (W) 1,354 1,354 1,354
the available converters without the need for a central controller to redispatch the
source converters. If a converter turns off or fails, the remaining converters sense a
decrease in bus voltage and increase their respective output current to compensate
for the lost source.
Consider a system with five sources on a common dc bus supplying 1,354 W of
total load. Each dc source converter has a load-line that describes the v–i terminal
characteristics, as shown in Figure 7.11. Assuming negligible bus impedance
between the five converters, the solution to the base case (where all converters are
operational) results in the bus voltage Vop1 with each converter supplying Iop1
current. The analytical solution for the operating point is found by solving the load-
flow equations for n source converters and m constant-power loads (Table 7.3):
1
Voc;n In ¼ Vbus ; 8n
Kn
X X Pm (7.3)
In ¼
n m
Vbus
It is observed that the droop gain is the slope of the v–i curve and modifies the
actual source impedance. Thus, a simple model for a source converter under droop
control is a constant voltage behind impedance:
1
Vs ¼ Voc is (7.4)
K
where K is the droop gain and can be defined in terms of a resistance:

1
K¼ (7.5)
Rdroop
Although droop control can be as simple as a series resistance, a more energy effi-
cient choice is a closed-loop controller such as a phase-angle controller in a rectifier
source converter. For an arbitrary source converter, the permissible droop resistance
is lower bounded by the actual source resistance of the converter:
Rs Rdroop (7.6)
In the previous example, the source converters are assumed to be identical with
identical droop characteristics. Thus, the total load current is shared equally.
In general, however, each converter can have an arbitrary droop characteristic
representing its operating parameters, power limits, or preferred dispatch:
1
Rdroop ¼ Rs ; where 0 l < 1 (7.7)
1l
Thus, droop control programs the effective output impedance of the source-converter
and has been shown to result in current sharing. It also has direct implications to
system dynamic behavior.
7.5.3 Dynamic stabilization

The closed-loop reference-to-output transfer function for the converter model in
Figure 7.10, assuming an ideal and lossless bus, is

K s þ wlp Zload
Gvg ðsÞ ¼ 2 (7.8)
s þ 1 þ Cs wlp Zload =Cs Zload s þ wlp ðKZload þ 1Þ =Cs Zload
After writing the characteristic polynomial in the usual way and substituting (7.5),
the expression for damping is
Cs wlp Zload þ 1
z ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi (7.9)
2 Cs wlp Zload Zload =Rdroop þ 1
This analysis suggests that droop control can be used to dynamically stabilize a
system [45]. Consider the three-bus system from Figure 7.8. The system is simu-
lated in DYMOLA using linearized averaged-model buck converters with damped
input filters. The source is modeled as an ideal voltage. The system is stable,
albeit with very light damping, z ¼ 107 106, as shown by a transient response in
Figure 7.12. The system eventually reaches steady state with each load-converter
supplying 144 W of load (about 3 A of current from the bus).
The system has reached steady state prior to 5 s when the load on bus 1
experiences a step-change in output power to 576 W. Figures 7.13 and 7.14 show
that the increase in current at bus 1 depresses the voltages at each bus in the system
51
50.5
Voltage [V] 50
49.5
49
48.5
48
0 0.25 0.5 0.75 1 1.25 1.5

Time [s]
Figure 7.12 Initial system transient response for a lightly damped system
30
25
Bus 1
20
Current [A]
15
10 Bus 1
Bus 2 Bus 2
5
Bus 3 Bus 3
0
4.8 4.9 5 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 6
Time [s]
Figure 7.13 Bus current of each POL converter. At 5 s, the load increases at
bus 1. The previously stable system becomes unstable with growing
oscillations. At 5.5 s, the droop resistance at the source converter
is adjusted to restabilize the system
70
60
Bus1 [V]
50
40
30
70
60
Bus2 [V]
50
40
30
70
60
Bus3 [V]
50
40
30
4.8 4.9 5 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 6
Time [s]
Figure 7.14 Bus voltage at each POL converter. At 5 s, the load increases at
bus 1. The previously stable system becomes unstable with growing
oscillations. At 5.5 s, the droop resistance at the source converter
is adjusted to restabilize the system
and that the previously stable system becomes unstable with growing oscillations.
At 5.5 s, droop control is enabled on the source converter. The droop control results
in an equivalent resistance of 0.10 W. The droop control increases the effective
system damping to z ¼ 5.64 103 and the unstable system is stabilized.
A potential drawback to droop control is that there is a stationary error intro-
duced in the bus voltage as seen in Figure 7.14. In the literature, a second control
loop with low-pass filtering and PI control is proposed to increase the internal
open-circuit source voltage VOC and compensate for this stationary error. Since in
this example, droop control is only used to stabilize the system, not share load, the
stationary error can be minimized by adaptively programming the minimum droop
necessary to achieve the desired damping.
The drop in bus voltage, however, carries useful information about the health
of the system. The drop can be due to a partial loss of generation and indicate that
the system may be energy constrained, as shown in Figure 7.11, or the drop can be a
result of needing increased damping which would indicate that the system is
nearing a region of instability as shown in Figure 7.14. Load-side controls that
sense this bus voltage drop will not know why the drop occurred but can take
mitigating action such as switching supply buses, activating a power buffer, or
invoking load interruption to alleviate system stress.
7.6 Voltage-based load interruption
Autonomous local control performs dynamic load interruption based on the sensed
bus voltage at the POL converter and the priority of that load. Each of the loads in
the radial system shown in Figure 7.8 is assigned a priority based on the explana-
tion of Section 7.4.3. Table 7.4 gives an example of the priority assignment that
will be used in this section. By selecting the highest priority load to be the one
furthest from the source, the example will demonstrate that priority need not be
constrained by the topology.
The control strategy in (7.10) is to turn off a particular load when the sensed
bus voltage drops below a lower threshold and turn the converter back on when the
sensed voltage exceeds an upper threshold. The control is complicated when the on
and off control action of the converter causes its input voltage to drop below lower
threshold during turn-on and to rise above the upper threshold during turn-off.
Hysteresis is useful to provide a dead band to prevent this type of chattering. In a
larger system, however, the possibility arises of not having a large enough hyster-
esis band. In this situation, additional information and controls are needed to pre-
vent the chattering (Figure 7.15).
Turn on if : Vinput > Vupper limit

(7.10)
Turn off if : Vinput < Vupper limit
One method to obtain these set points is to perform an exhaustive contingency load
flow analysis. Each contingency represents a particular combination of operating
points for the system POL converters. For a small system, this offline process is
straightforward and computationally fast. The resulting set of power flow solutions
is processed by a search algorithm to find the upper and lower voltage limits for
each load converter.
Table 7.4 Load priority assignment

in the three-bus test-bed
Converter Priority
POL 1 Semiessential (SE)
POL 2 Nonessential (NE)
POL 3 Essential (E)
49
48
47
46
Vbus
45
44
43
42
Load shed
1
POL 1 0
1
POL 2 0
1
POL 3 0
1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4
Time [s]
Figure 7.15 Control introduces chattering into the system
IBus1 IBus2 IBus3

V1 V2 V3
Z1 Z2 Z3
I1 P1 I2 P2 I3 P3
Vs +
−
POLC1 POLC2 POLC3
PL1 PL2 PL3
L1 L2 L3
Figure 7.16 Model of a radial dc power system with a single source and three
point-of-load converters
7.6.1 Power flow analysis

Power flow analysis is used to determine a priori the values of the voltage set points
for under-voltage load interruption and restoration. The MATLAB Power System
Toolbox contains a suitable power flow algorithm for the test system, redrawn in
Figure 7.16. The radial design results in rapid convergence of the power flow
algorithm. The POL converter is a closed-loop buck converter designed to have
Start
Initialize all bus

voltages to Vs
Compute efficiency
for each converter
Updated bus
voltages
Run PST load flow,

obtain new bus
voltages
No
Bus voltages
converged?
Yes
End
Figure 7.17 Outer loop of power flow algorithm incorporating converter

efficiency
good line and load regulation to ensure constant output power given a fixed resis-
tive load. Input power to the converter depends on the efficiency of the converter,
which is a function of the input voltage and the output power. Thus, the power flow
equations are complicated by the nonlinear efficiency of the dc-to-dc converters.
To improve the power flow results, the efficiency of the converters was
included in the power flow algorithm, shown in Figure 7.17. This added an outer
iteration loop to the power flow algorithm although it could be incorporated
directly into a power flow routine to increase computation efficiency. The power
flow algorithm begins by assuming that there is no voltage drop in the system and
that the voltage at each node on the bus is identical to the open-circuit source
voltage. These node voltages are then used to calculate the initial efficiency for
each dc–dc converter and hence the initial power withdrawn from each node on the
bus. The PST power flow program is then run. The result of the power flow, the
new system node voltages, is used to update the efficiency of each converter, and
the process is repeated until node voltages converge to within an acceptable bound.
7.6.2 Contingency analysis

Contingency analysis is performed to obtain the system voltages for each combi-
nation of POL converter operating points. Notation that links the power flow result
to the configuration of the converter is
V ðconverterÞstate
limit (7.11)
The parameter converter uniquely identifies the converter and takes values
converter 2 f1; 2; . . .; N g; where N is the total number of converters (7.12)
The parameter state identifies the current operating state of the converter and takes
values
state 2 foff ; on; base; highg (7.13)
The states base and high describe the output power of the converter as either nominal
rated base power or overload power. The state on is the compliment of off and
includes both the base and high states. This is useful when the actual output power is
unimportant. Thus, the cardinality of the set of unique output-power levels is
jstatej ¼ 3 (7.14)
The number of power flows that are needed to perform an exhaustive contingency
analysis for all possible configurations for N POL converters is
N jstatej (7.15)
The three-converter example in Figure 7.16 will therefore require 27 power flow
computations. The parameter limit identifies the type of voltage limit. It takes the
value
limit 2 fmin; max; UVPg (7.16)
The minimum limit is the lowest allowed input voltage before the converter turns
off, maximum is the highest input voltage before a converter turns on, and UVP
refers to the hardware UVP limit designed to self-protect.
The results of the exhaustive contingency analysis for the system in Figure 7.16
are graphically presented in Figure 7.18. The results of the 27 power flows are shown
for each of the three POL converters. Horizontal lines indicate the open circuit supply
voltage, VS, and the UVP limit, VUVP, where the converter turns off to self-protect.
Since POL3 is furthest from the source, it has the lowest bus voltages for each con-
tingency. Solid bars indicate contingencies of interest where one or more converters in
that contingency would trip off-line due to UVP.
7.6.3 Search algorithm

The upper-limit and lower-limit voltages for each POL converter can be found from
the maximum and minimum values for sets of specific contingencies. This opti-
mization is done by a search algorithm that parses the results of the contingency
23
VS 22
21
20
19
18
Input voltage
17
16
15
14
VUVP 13
12
11
10
9
1 POL1 27 1 POL2 27 1 POL3 27
27 contingencies at each point-of-load converter
Figure 7.18 Results of exhaustive contingency analysis on the radial test system
with a single source and three point-of-load converters. The load on
each converter can be either off, base load, or overload. Solid bars
indicate contingencies where a bus voltage in the system is below the
UVP for that converter
analysis. The process starts with the lowest priority converter, POL2. The lower
bound on the lower-limit is
8 9
>
> V ðNEÞUVP ; >
>
< =
on
on
V ðNEÞmin > max V ðNEÞ ; V ðEÞ V ðEÞUVP (7.17)
>
> >
>
: on ;
V ðNEÞ ; V ðSEÞ V ðSEÞUVP
The algorithm finds the voltage POL2 for all contingencies where the higher
priority loads are on and their input voltage is below the UVP limit. In this exam-
ple, this occurs for five contingencies at POL3, shown as the dark bars in
Figure 7.18. The algorithm also considers the possibility that the voltage at POL2
can fall below the UVP limit.
The upper-limit has both an upper and lower bound. Only, the essential and
nonessential loads are considered here for simplicity.
( )
high off off V ðEÞoff ;
V ðEÞ ; V ðNEÞ < V ðNEÞmax < min (7.18)
V ðEÞbase ; V ðNEÞoff
The lower bound for restoring the nonessential load occurs when the essential load
is at the maximum power and the nonessential load is off. The upper bound is found
by minimizing a set of contingencies where the essential load is off or supplying

base load power and the nonessential load is off. These combinations of con-
tingencies were chosen to ensure that restoring the nonessential load never jeo-
pardizes operation of the essential load.
7.7 dv/dt-Based dynamic load interruptions
In a power system in steady state, the supply is matched to the load and the system is
stable. However, many dc distribution systems are electrically weak and do not have
the spinning reserves or other stability mechanisms. The bus voltage can sag for many
reasons such as partial loss of generation, increase in load, or topological reconfi-
guration. Further, tight voltage regulation in dc–dc converters makes them operate as
CPL, as demonstrated in Section 7.5.1, which draw increasing current for decreasing
bus voltage, possibly leading to further voltage sag or even voltage collapse [43].
Demand-side management is a suite of techniques that control the loads so that
they become integral components in system stability. Interruptible load is one
method that provides curtailment of demand to promote system security. Autono-
mous local control is investigated to perform this load-side control and improve
system reliability.
The P–V curve is a useful tool to visualize the operation of a power system.
Figure 7.19 illustrates a family of the familiar P–V system curve. Maximum power
transmission occurs at the nose were the source impedance and load impedance are
equal. In a dc system, the bus voltage drops as the load increases due to voltage-
divider action of the source impedance and the load impedance.
A system is initially in steady state with voltage V(t1) delivering total load
power of P(t1). The system impedance suddenly increases, perhaps due to a partial
loss of generation or topological reconfiguration, and the operating point moves to
a new P–V curve at time t2. However, the voltage V(t2) is below the under-voltage
Vbus
Voc
t6 t3 t1
VUVP
t5
t2
MPT
P(t3) P(t2) P(t1) Pbus
Figure 7.19 P–V Curve showing operating points as the system impedance
increases and loads are interrupted
V(t3)
V(t6)
V(t1)
VUVP
V(t5)
V(t2)
(a)
Pload(t1)
Pload(t3)
Pload(t6)
t1 t2 t3 t4 t5 t6
(b)
Figure 7.20 Ideal bus voltage and load power as system impedance increases and
loads are interrupted to prevent voltage collapse. (a) Bus voltage
decreases in response to increased system impedance at t1 to reach
the operating point on the new P–V curve at t2. The new bus voltage
is below the UVP limit, so control action causes load to be shed,
moving to a mew operating point on the same P–V curve at t3 with a
higher bus voltage. The cycle repeats at t4. (b) Load power in the
system changes as point-of-load converters are turned-off to reduce
total system load when the bus voltage drops below the UVP
limit and load is shed, moving to a new operating point on the same P–V curve at t3.
The time-domain waveforms in Figure 7.20 reveal that these changes in operating
points do not occur instantaneously. The trajectories on the two figures, however,
are idealized to improve clarity of the system response and do not include the
dynamics associated with the inductance of the bus, the input filter, and the con-
stant-power dc–dc converters.
7.8 Load prioritization and scheduling

In an energy-constrained system, load prioritization is critical for system stability
and control because it provides a structured approach to the decision and control
process. It is likely that a particular load’s priority may need to change depending
Table 7.5 Load priority-assignment example
Load type
Status
Hotel (domestic) Lifeboats Propulsion Weaponry
In port Essential Nonessential Nonessential Nonessential
Patrol Semiessential Semiessential Essential Semiessential
General quarters Nonessential Nonessential Essential Essential
Abandon ship Nonessential Essential Nonessential Essential
on the operation of the system. In a naval ship, each load is classified as either
nonessential, semiessential, or essential. A general framework for organizing the
load priorities is a two-dimensional matrix as shown in Table 7.5. In one example,
propulsion is considered the highest priority, while hotel loads such as lighting in
crew quarters and power in the galley are less important and can be sacrificed
depending on the threat level. In another example, the launch equipment for life-
boats has a higher priority under ‘‘Patrol’’ status for safety reasons but yields
priority to other systems such as weapons and propulsion under ‘‘General Quar-
ters’’ status. If local energy is available in a power buffer, load priority is useful to
determine if the load simply turns off or the power buffer operates when trouble is
sensed on the dc bus.
Fine-tuning the performance of the system requires that each POL converter
has some information about the entire system. Low-bandwidth signaling from the
command and control center can broadcast the current state of the system, but each
POL converter ultimately decides how to use the information—unlike in a cen-
tralized control scheme where each load is directly controlled. The distributed
control strategy is inherently fault-tolerant because each controller acts indepen-
dently. If the low-bandwidth communication is compromised, each controller can
continue to operate using the last-known state or revert to fail-safe operation as
determined by the load-priority table.
7.9 Summary
In a traditional power grid system, the operator had total control of generation and
distribution assets while the load was viewed as a disturbance. Thus, planning
and operation necessitated always being prepared for unforeseen changes in the load
consumption. The result is that the US power grid is amazingly resilient, robust, and
expensive. As we consider a new paradigm of dc microgrid systems, overcapacity
may not be feasible for technological and/or economic reason. Yet, high power
quality and availability is more important than ever particularly to support the digital
economy and information age. This is compounded as renewable sources become
increasingly utilized, and the system operator no longer has total and arbitrary
control of the generation.
In this chapter, we introduced a framework for load control in a LAPES. In this

paradigm, the load is considered to be an energy asset which can be controlled not
just by the end user but also by the system. As such, a new degree of freedom is
introduced in the control problem of balancing electrical supply and demand. Key
to implementing any of these concepts is to strike the balance between the oppor-
tunity and cost, which would provide the techno-economic trade-off needed to
implement these concepts practically. Important application-specific design con-
siderations are the desired energy availability and the price willing to be paid for
that availability as well as the tolerable amount of control relinquished. Optimizing
this is not straightforward and may lead to solutions that change with time and
conditions. To anticipate this, the chapter discussed the concept of load prioritiza-
tion and a method to change the allowable control of the load.
To understand the benefits of load control, we introduced the use of a multi-
objective optimization problem, which is concerned with energy availability and
reliability. The Pareto Frontier technique was used as a way to visualize the engi-
neering trade-off and assist with the optimization. Within this chapter, we were
able to see the application and feasibility of load modulation for different time-
scale transients. Modulation was explored as an alternative to load shed for situa-
tions that could tolerate a reduction of power, such as building lighting and HVAC
systems, and PHEV charging. Within a LAPES, POL converters can be controlled,
hence modulate the load, to achieve the objective of energy availability. These
techniques range from the use of dc/dc level power buffers, load scheduling,
replenishing energy storage, and a strategy for prioritized load shedding. Using
droop control, the system can realize distribute control to eliminate a single-point
of failure of a central control architecture. A highly desirable benefit is that droop
control can also improve system stability by introducing extra damping which is
important as a power system nears a boundary of stability. Other techniques like
voltage-based interruptions and load prioritization were discussed in this chapter.
Acknowledgments
The authors wish to acknowledge the various funding sources which supported the
research behind this content of this chapter. Dr. Balog acknowledges support
received from the Qatar Foundation. This publication was made possible by NPRP
grant no. 9-204-2-103 from the Qatar National Research Fund (a member of Qatar
Foundation). The statements made herein are solely the responsibility of the
authors. This work was also supported in part by the Grainger Center for Electric
Machinery and Electromechanics at the University of Illinois. Dr. Mohamad
Shadmand acknowledges support from the Qatar Foundation. This publication was
made possible by NPRP-EP grant no. X-033-2-007 from the Qatar National
Research Fund (a member of Qatar Foundation). The statements made herein are
solely the responsibility of the authors. Mr. Morcos Metry acknowledges the
sponsorship of the Qatar Research Leadership Program (QRLP) under the Qatar
National Research Fund.
References
[1] N. A. Armstrong, ‘‘The Engineered Century: A Century Hence, 2000 May
Be Viewed as Quite a Primitive Period in Human History. It’s Something to
Hope For,’’ The Bridge, vol. 30, no. 1, pp. 14–18, Spring 2000.
[2] C. W. Gellings, ‘‘New Products and Services for the Electric Power Industry,’’
The Bridge, vol. 40, no. 1, pp. 21–28, Spring 2010.
[3] S. M. Amin, ‘‘Securing the Electricity Grid,’’ The Bridge, vol. 40, no. 1,
pp. 13–20, Spring 2010.
[4] J. D. Rhodes and The Conversation. (May 2017). The Outdated US Electric
Grid is Going to Cost $5 Trillion to Replace. Available: http://www.
businessinsider.com/replacing-us-electrical-grid-cost-2017-3.
[5] A. Kwasinski, W. Weaver, and R. S. Balog, Microgrids and Other Local
Area Power and Energy Systems. Cambridge University Press, Cambridge,
UK, 2016.
[6] R. H. Lassetter, ‘‘Smart Distribution: Coupled Microgrids,’’ Proceedings of
the IEEE, vol. 99, no. 6, pp. 1074–1082, Jun 2011.
[7] Y. Zhang, L. Xie, and Q. Ding, ‘‘Interactive Control of Coupled Microgrids
for Guaranteed System-Wide Small Signal Stability,’’ IEEE Transactions on
Smart Grid, vol. 7, no. 2, pp. 1088–1096, Nov 2015.
[8] D. T. Ton and M. A. Smith, ‘‘The U.S. Department of Energy’s Microgrid
Initiative,’’ The Electricity Journal, vol. 25, no. 8, pp. 84–94, Oct 2012.
[9] R. S. Balog, W. W. Weaver, and P. T. Krein, ‘‘The Load as an Energy Asset
in a Distributed DC SmartGrid Architecture,’’ IEEE Transactions on Smart
Grid, vol. 3, no. 1, pp. 253–260, March 2012, Art. no. TSG-00146-2011.
[10] Commonwealth Edison Co. (May 2017). Central AC Cycling. Available: https://
www.comed.com/WaysToSave/ForYourHome/Pages/CentralACCycling.aspx.
[11] J. O. Swanson and J. P. Jolliffe, ‘‘Load Shedding Program in the Pacific
Northwest,’’ Transactions of the American Institute of Electrical Engineers.
Part III: Power Apparatus and Systems, vol. 73, no. 2, pp. 1655–1668, Jan 1954.
[12] R. M. Maliszewski, R. D. Dunlop, and G. L. Wilson, ‘‘Frequency Actuated
Load Shedding and Restoration Part I—Philosophy,’’ IEEE Transactions on
Power Apparatus and Systems, vol. PAS-90, no. 4, pp. 1452–1459, Jul 1971.
[13] Z. Wang, B. Chen, J. Wang, and M. M. Begovic, ‘‘Stochastic DG Placement
for Conservation Voltage Reduction Based on Multiple Replications
Procedure,’’ IEEE Transactions on Power Delivery, vol. 30, no. 3, pp.
1039–1047, Jun 2015.
[14] H. Chen, Y. Li, R. H. Y. Louie, and B. Vucetic, ‘‘Autonomous Demand Side
Management Based on Energy Consumption Scheduling and Instantaneous
Load Billing: An Aggregative Game Approach,’’ IEEE Transactions on
Smart Grid, vol. 5, no. 4, pp. 1744–1754, Jun 2014.
[15] H. Zhong, L. Xie, and Q. Xia, ‘‘Coupon Incentive-Based Demand Response:
Theory and Case Study,’’ IEEE Transactions on Power Systems, vol. 28,
no. 2, pp. 1266–1276, May 2013.
[16] Younicos. (May 2017). Y.Cube: Our Plug-and-Play Energy Storage Solution
[Online]. Available: https://www.younicos.com/products/y-cube/.
[17] Tesla. (May 2017). PowerPack: Utility and Business Energy Storage.
Available: https://www.tesla.com/powerpack.
[18] Tesla. (May 2017). Powerwall: Reliable Power Day and Night. Available:
https://www.tesla.com/powerwall.
[19] L. Poulin, Reducing Risk with Software Process Improvement. Boca Raton,
FL: Auerbach Publications, Taylor & Francis Group, 2005.
[20] A. V. Lotov and K. Miettinen, ‘‘Visualizing the Pareto Frontier,’’ in Multi-
objective Optimization: Interactive and Evolutionary Approaches, J. Branke,
K. Deb, K. Mirettinen, and R. Slowinski, Eds.: Springer-Verlag Berlin
Heidelberg, 2008, pp. 213–243.
[21] N. Barr, Economics of the Welfare State, 5 ed. United Kingdom: Oxford
University Press, 2012.
[22] M. B. Shadmand and R. S. Balog, ‘‘Multi-Objective Optimization and
Design of Photovoltaic-Wind Hybrid System for Community Smart dc
Microgrid,’’ IEEE Transactions on Smart Grid, vol. 5, no. 5, pp. 2635–2643,
Sep 2014.
[23] J. W. Kimball, B. T. Kuhn, and R. S. Balog, ‘‘A System Design Approach
for Unattended Solar Energy Harvesting Supply,’’ IEEE Transactions on
Power Electronics, vol. 24, no. 4, pp. 952–962, Apr 2009.
[24] ERCOT. (Sep 2016, May 2017). Renewable Integration at ERCOT. Avail-
able: http://www.cigre.cl/seminarios/wp-content/uploads/2016/09/ERCOT-
DAN-WOODFIN.pdf.
[25] Lighting Research Center (LRC). (April 2010, May 2017). Demonstration
and Evaluation of Lighting Technologies and Applications: Daylight-
Harvesting Switch. Available: http://www.lrc.rpi.edu/programs/DELTA/pdf/
FTDELTA_DaylightHarvestingSwitch.pdf.
[26] The National Optical Astronomy Observatory. (May 2017). Recommended
Light Levels (Illuminance) for Outdoor and Indoor Venues. Available:
https://www.noao.edu/education/QLTkit/ACTIVITY_Documents/Safety/
LightLevels_outdoorþindoor.pdf.
[27] Occupational Safety and Health Branch. (Dec 2008, May 2017). Lighting
Assessment in the Workplace. Available: http://www.labour.gov.hk/eng/
public/oh/Lighting.pdf.
[28] R. J. Sledz, ‘‘Control Room Lighting: An Application of Human Factors Engi-
neering,’’ IEEE Power Engineering Review, vol. PAS-101, no. 8, pp. 2755–2761,
Aug 1982.
[29] J. Voelcker. (Oct 2007, May 2017). Can Plug-In Hybrid Electric Vehicles Keep
the Electric Grid Stable? Available: http://spectrum.ieee.org/transportation/
advanced-cars/can-plugin-hybrid-electric-vehicles-keep-the-electric-grid-
stable.
[30] A. Vaughan. (Mar 2017, May 2017). Charge Electric Cars Smartly to Take
Pressure off National Grid—Minister. Available: https://www.theguardian.
com/environment/2017/mar/20/electric-cars-uk-power-grids-charging-peaks-
sse-demand-side-response.
[31] S. Chakraborty, W. Kramer, B. Kroposki, et al., ‘‘Interim Test Procedures
for Evaluating Electrical Performance and Grid Integration of Vehicle-to-Grid
Applications,’’ National Renewable Energy Lab (NREL), Golden, Colorado.
Jun 2011, Available: http://www.nrel.gov/docs/fy11osti/51001.pdf.
[32] I. U. Nutkani, P. C. Loh, P. Wang, and F. Blaabjerg, ‘‘Cost-Prioritized Droop
Schemes for Autonomous AC Microgrids,’’ IEEE Transactions on Power
Electronics, vol. 30, no. 2, pp. 1109–1119, Feb 2015.
[33] L.-L. Fan, V. Nasirian, H. Modares, F. L. Lewis, Y.-D. Song, and A. Davoudi,
‘‘Game-Theoretic Control of Active Loads in DC Microgrids,’’ IEEE
Transactions on Energy Conversion, vol. 31, no. 3, pp. 882–895, Sep 2016.
[34] J. M. Guerrero, P. C. Loh, T. L. Lee, and M. Chandorkar, ‘‘Advanced
Control Architectures for Intelligent Microgrids Part II: Power Quality,
Energy Storage, and ac/dc Microgrids,’’ IEEE Transactions on Industrial
Electronics, vol. 60, no. 4, pp. 1263–1270, Apr 2013.
[35] J. M. Guerrero, M. Chandorkar, T.-L. Lee, and P. C. Loh, ‘‘Advanced
Control Architectures for Intelligent Microgrids—Part I: Decentralized and
Hierarchical Control,’’ IEEE Transactions on Industrial Electronics, vol. 60,
no. 4, pp. 1254–1262, Apr 2013.
[36] V. Nasirian, A. Davoudi, F. L. Lewis, and J. M. Guerrero, ‘‘Distributed
Adaptive Droop Control for DC Distribution Systems,’’ IEEE Transactions
on Energy Conversion, vol. 29, no. 4, pp. 944–956, Dec 2014.
[37] M. D. Cook, G. G. Parker, R. D. Robinett, and W. W. Weaver, ‘‘Decen-
tralized Mode-Adaptive Guidance and Control for DC Microgrid,’’ IEEE
Transactions on Power Delivery, vol. 32, no. 1, pp. 263–271, Feb 2017.
[38] J. He, Y. W. Li, J. M. Guerrero, F. Blaabjerg, and J. C. Vasquez, ‘‘An
Islanding Microgrid Power Sharing Approach Using Enhanced Virtual
Impedance Control Scheme,’’ IEEE Transactions on Power Electronics,
vol. 28, no. 11, pp. 5272–5282, Nov 2013.
[39] R. S. Balog, ‘‘Autonomous Local Control in Distributed DC Power
Systems,’’ PhD Dissertation, Department of Electrical and Computer Engi-
neering, University of Illinois at Urbana-Champaign, 2006.
[40] F. Gao, S. Bozhko, G. Asher, P. Wheeler, and C. Patel, ‘‘An Improved
Voltage Compensation Approach in a Droop-Controlled DC Power System
for the More Electric Aircraft,’’ IEEE Transactions on Power Electronics,
vol. 31, no. 10, pp. 7369–7383, Oct 2016.
[41] M. B. Shadmand, R. S. Balog, and H. Abu-Rub, ‘‘Model Predictive Control
of PV Sources in a Smart dc Distribution System: Maximum Power Point
Tracking and Droop Control,’’ IEEE Transactions on Energy Conversion,
vol. 29, no. 4, pp. 913–921, Dec 2014.
[42] X. Lu, J. M. Guerrero, K. Sun, and J. C. Vasquez, ‘‘An Improved Droop
Control Method for dc Microgrids Based on Low Bandwidth Communica-
tion with dc Bus Voltage Restoration and Enhanced Current Sharing
Accuracy,’’ IEEE Transactions on Power Electronics, vol. 29, no. 4,

pp. 1800–1812, Apr 2014.
[43] W. W. Weaver and P. T. Krein, ‘‘Optimal Geometric Control of Power Buf-
fers,’’ IEEE Transactions on Power Electronics, vol. 24, no. 5, pp. 1248–1258,
May 2009.
of DC Microgrids With Instantaneous Constant-Power Loads,’’ IEEE
Transactions on Power Electronics, vol. 26, no. 3, pp. 822–834, Mar 2011.
[45] N. Bottrell, M. Prodanovic, and T. C. Green, ‘‘Dynamic Stability of a
Microgrid With an Active Load,’’ IEEE Transactions on Power Electronics,
vol. 28, no. 11, pp. 5107–5119, Nov 2013.
Chapter 8
Electric vehicle charging infrastructure
and dc microgrids
Srdjan Srdic1 and Srdjan Lukic1
8.1 Overview of EV and EVSE markets and trends

With substantial growth in sales of electric vehicles (EVs) globally, there is a push
for expansion of the recharging infrastructure to service these vehicles. Over
2 million of electric cars (battery-electric and plug-in hybrid electric), 200 million
electric motorcycles and 345 thousand buses (primarily in China) were deployed
worldwide by the end of 2016 [1], and over 1.2 million of electric cars were sold
globally in 2017 alone [2]. However, the global electric car stock made only a 0.2%
of the total number of passenger cars globally in 2017. Assessments of country
targets, original equipment manufacturer announcements and deployment scenarios
for electric cars indicate that the number of EVs will range between 9 and
20 million by 2020 and between 40 and 70 million by 2025 [1]. Furthermore, a
number of countries have decided to end the sales of fossil-fuel-powered cars in the
near future (Norway by 2025, India and Netherlands by 2030, Scotland by 2032,
France and rest of the United Kingdom by 2040), further accelerating the shift to
electric transportation. The electric vehicle supply equipment (EVSE) is closely
following the EV stock growth, with 2.3 million EVSE outlets (including 110,000
publicly available fast-charging outlets) available globally in 2016, and predicted
six-fold increase in the available outlets by 2025 [1]. The fastest growing EVSE
market is the Chinese market, with over 88,000 publicly available fast-charging
outlets in 2016.
With the advancement and maturity of batteries used in EVs, the storage
capacity and, therefore, the range of EVs are increasing, while the vehicle cost is
dropping. Major vehicle manufacturers are already producing EVs with range
exceeding 200 mi (320 km) at prices competitive with conventional vehicles. With
these trends in place, the power of EV chargers needs to increase in order to reduce
the recharging time and provide the convenience inherent in refueling conventional
vehicles at the gas stations. According to the US Environmental Protection Agency,
the fuel economy of electric cars produced from 2016 to 2018 ranges from 72 to
1
Department of Electrical and Computer Engineering, North Carolina State University, USA
136 mi/gal equivalent [3]. Expressed in terms of energy consumption, modern

electric cars consume between 25 and 47 kWh of energy (stored in the battery) per
100 mi (15.6–29.4 kWh per 100 km). Therefore, a charger with a power between
300 and 564 kW would be needed to add 200 mi of range to a modern electric car in
10 min. However, number of barriers related to battery technology cost, vehicle
electrical architecture and power electronics subsystems, and infrastructure issues
related to high power demand, would still need to be overcome before the ultrafast
charging (350 kW or higher) can become a commercially viable EV recharging
option [4]. To support the need for shorter recharge times, dc fast chargers rated
150 kW have been recently deployed across the world and the new high power
units (350 kW or higher) have been announced for 2018 by major EV and EVSE
makers.
8.2 dc Fast charging systems and requirements

8.2.1 dc Fast charging systems and standards
To ensure charging compatibility, a number of governing bodies, including the
Society of Automotive Engineers (SAE) and International Electrotechnical Com-
mission (IEC), have developed standardized protocols and couplers that interface a
plug-in EV with the EVSE. The five standard dc fast charging systems that are in
use are listed in Table 8.1. The IEC 62196-3 Standard defines four different vehicle
coupler configurations for dc fast charging: Configuration AA [proposed and
implemented by CHAdeMO (CHArge de MOve) Association], Configuration BB
(Also known as GB/T and available only in China), Configuration EE [Type 1
Combined Charging System (CCS), adopted in North America] and Configuration
FF (Type 2 CCS, adopted in Europe and Australia). There is also a proprietary
system developed by Tesla Inc. and used exclusively for Tesla vehicles.
With over 18,000 charging outlets deployed worldwide throughout the end of
first quarter of 2018, CHAdeMO interface has the largest global coverage of all fast
charging systems [5]. The system was proposed by several Japanese companies
who formally established CHAdeMO Association in March 2010. The current
version (v1.2) of the CHAdeMO charging protocol, published in March 2017,
supports dc fast charging with power levels up to 200 kW (500 V/400 A), and the
400-kW (1,000 V/400 A) version of the protocol is announced for 2018 [6]. The
new CHAdeMO high-power charging system was demonstrated in March 2017,
with 150 kW units charging modified Nissan Leaf at 100 kW peak power.
The GB/T is a fast charging standard used predominantly in China, and it
originates from a Chinese national standard (GuoBiao) GB/T 20234.3 [7], issued
by the Standardization Administration of the People’s Republic of China. Since
2017, this standard is mandatory for all new EVs and fast charging infrastructure in
China [8]. At the beginning of 2018, the dc fast chargers for passenger cars with
GB/T coupler were available at power levels up to 120 kW in single-port config-
uration and 160 kW (2 80 kW) in dual-port configuration.
Table 8.1 Standard dc fast charging systems
dc Fast CHAdeMO GB/T CCS Type 1 (US) CCS Type 2 (EU) Tesla
charging IEEE 2030.1.1, (China) SAE J1772, IEC 62196-3 (US, EU)
system IEC 62196-3 GB/T 20234.3, IEC 62196-3 (Configuration FF)
(Configuration AA) IEC 62196-3 (Configuration EE)
(Configuration BB)
Charge coupler
inlet
Supercharger
Modified Mennekes
Type 2 (EU)
Standard max. 600 V 750 V 600 V 1,000 V 410 V

rated voltage (1,000 V per GB/
T 20234.3-2015)
Standard max. 200 A 250 A 200 A 200 A 330 A continuous shared,
rated current (400 A per or 2 200 A continuous
CHAdeMO v1.2) in Urban Superchargers
Charging power 50 kW 120 kW 150 kW 175 kW 135 kW shared (2 posts),
available in (150 kW per or 2 72 kW (2 posts)
2017 CHAdeMO v1.2) in Urban Superchargers
Communication Controller area Controller area Power line Power line Proprietary
protocol network (CAN) network (CAN) communication communication
(PLC) (PLC)
CharIN (Charging interface) initiative is established in 2015 with the aim to

globally promote the CCS for EVs, to define requirements for standards and
certification of the CCS and to prepare global introduction of the CCS [9]. The
CharIN initiative had over 100 members at the beginning of 2018, including some
of the major automobile manufacturers. Approximately 7,000 CCS charging outlets
(both Type 1 and Type 2) were available globally at the end of 2017 [10]. The
CharIN initiative promotes two systems that support dc fast charging: CCS Type 1
(also known as Combo 1) system, which is mainly in use in North America and
supports dc fast charging with power up to 120 kW (per CCS v1.0), and a CCS
Type 2 (also known as Combo 2) system, which supports dc fast charging with
power up to 200 kW (per CCS v1.0), and is predominantly used in Europe [11].
Both systems also support ac charging up to 43 kW, and both are derived by adding
two additional high-current-carrying pins to the existing Type 1 and Type 2 ac
charging couplers defined by IEC 62196-2 and SAE J1772. The announced future
version of the standard (CCS v2.0) will support dc fast charging with power up to
350 kW. Even though high-power charging is not supported by the CCS v1.0
specification, a few high-power charging stations advertised as 350 kW-capable
have already been opened for public: a 150-kW prototype station in the United
States [12] and a 4 175-kW station in Germany [13]. The future version of the
standard (CCS v2.0), which is announced for 2018, will support dc fast charging
with power up to 350 kW at voltages up to 1,000 V [11,14].
Tesla Inc. has developed a proprietary charging system for their Superchargers,
which differs from the systems described above. More than 1,000 Supercharger
stations with almost 7,500 outlets (stalls) have been available globally at the
beginning of 2018 [15]. Tesla started deploying their superchargers in October
2012, with 120 kW output power (up to 210 A continuous, and 50–410 V at
the output) shared between two stalls. The new Gen. II units with 135 kW of
power shared between two stalls were introduced at the beginning of 2014.
In September 2017, Tesla has introduced Urban Superchargers with two compact
stalls per charger, delivering up to 72 kW (200 A, 50–410 V) each [16].
8.2.2 State-of-the-art EV dc fast chargers

The majority of EV dc fast chargers available on the market at the beginning of
2018 are rated at 50 kW and supplied from a three-phase low-voltage distribution
grid (up to 480 V line-to-line, depending on the region). The technical specifica-
tions of these dc fast chargers with different output power levels are listed in
Table 8.2.
Most of the modern dc fast chargers support more than one charging standard,
and some of them support the ac charging as well. The peak efficiency of most of
the dc fast chargers is below 94%. The efficiency of Tesla’s 135 kW Supercharger
was assumed to be the same as for Tesla’s onboard charger [17], since Super-
chargers are made by combining 12 onboard chargers [18].
The EV charging process is a sequence of events defined by the charging
protocol. The charging sequence typically starts with signal handshaking and
Electric vehicle charging infrastructure and dc microgrids 193
Table 8.2 Technical specifications of state-of-the-art dc fast chargers
Manufacturer ABB Tritium PHIHONG Tesla EVTEC

and model Terra 53 Veefil-RT integrated type supercharger espresso and
120 kW charge Delta
Stromtankstelle
4.0
Maximum 50 50 120 135 150
power (kW)
Supported CCS Type 1 CCS Type 1 and 2 GB/T Supercharger SAE Combo-1
standards CHAdeMO 1.0 CHAdeMO 1.0 CHAdeMO 1.0
Input voltage 480 V ac 380–480 V ac 380 V ac 15% 200–480 V ac 400 V ac 10%
600–900 V dc 480 V ac 10%
Output dc 200–500 200–500 200–750 50–410 170–500
voltage (V) 50–500 50–500
Output dc 120 125 240 330 300
current (A)
Peak efficiency 94 >92 93.5 91 93
(charger only)
(%)
Volume (L) 758 495 591 1,047 1,581
Weight (lb) 880 (400 kg) 364 (165 kg) 529 (240 kg) 1,320 (600 kg) 880 (400 kg)
exchange of maximum charging parameters between a car and a charger. The car
would also perform the insulation test before the charging can begin. If all the
required criteria are met, the car closes its dc contactor and the charging may begin.
During the charging process, the charger follows the voltage and current commands
from the car’s Battery Management System (BMS). Each charging event typically
consists of two charging modes (regimes): a ‘‘Constant Current’’ (CC) mode fol-
lowed by a ‘‘Constant Voltage’’ (CV) mode. During the CC mode, the charger
follows a slow varying current reference set by the BMS. After a certain time spent
in the CC mode (i.e., when the battery voltage reaches a certain value), the BMS
switches to CV mode and sends a voltage reference command, which increases
slowly as the charging progresses. The charging voltage increases much faster in
the CC mode, than in the CV mode. During the CV mode, the charging current
decreases. The time spent in CC and CV modes depends mainly on the applied
charging current and the initial battery State of Charge (SOC). When the battery
reaches a certain preset SOC, the car signals the charger to terminate the charging
by reducing the charging current to zero. The car then disconnects itself from the
charger by opening its dc contactor. The example of a charging profile of a Tesla
model S85 and the corresponding charging power and the battery SOC are shown
in Figure 8.1 [19]. The battery was charged from 12% SOC to 94% SOC, with a
peak power of 117.2 kW occurring at 16% SOC (2 min after the charging was
started). In this case, the CC mode lasted for approximately 2 min. During the
charging process, the car and the charger typically exchange the information on
current and voltage reference, the battery SOC and some status information. In case
of Tesla Superchargers, the car will also periodically send its Vehicle Identification
Number (VIN) to identify itself as a Tesla vehicle.
Battery current Battery voltage Charging power Battery SOC
350 410 120 100
110 90
300 400
Charging power (kW)

100 80
Battery voltage (V)

Battery current (A)
Battery SOC (%)

250 390 90 70
200 380 80 60
70 50
150 370 60 40
100 360 50 30
40 20
50 350 10
30
0 340 20 0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 0 5 10 15 20 25 30 35 40 45 50 55 60 65
Time (min) Time (min)
Figure 8.1 The charging profile of a Tesla model S85 and the corresponding
charging power and the battery SOC
A B C
Isolated dc/dc stage
PFC
Input Rectifier Output
circuit (dc/ac) (ac/dc)
filter (ac/dc) filter
(dc/dc)
HF
transformer
Figure 8.2 The simplified block diagram of a conventional dc fast charger power
conversion system
8.2.3 dc Fast charger power converter topologies

The conventional dc fast chargers typically have two power electronics conversion
stages: ac/dc rectification stage with power factor correction (PFC), which converts
three-phase input ac voltage to an intermediate dc voltage, and an isolated dc/dc
stage, which converts the intermediate dc voltage into regulated dc voltage required
to charge the EV and provides galvanic isolation from the ac grid. The fast charger
power conversion system is typically based on well-known converter topologies
used in motor drives. The simplified block diagram of a conventional dc fast
charger power conversion system is shown in Figure 8.2 [20].
Even though the input ac/dc stage can be made by implementing separate
converters for the rectifier and the PFC, a single-stage converter is preferred at
higher power levels due to its simplicity and higher reliability. An LLC filter and a
three-phase active pulse width-modulated (PWM) rectifier (shown in Figure 8.3)
are often used as an input stage in conventional dc fast chargers. This input stage is
usually based on converter modules from commercial motor drives [20,21], which
are typically made using the reverse conducting IGBTs, and operate at relatively
low switching frequencies (typically up to 10 kHz). The more efficient and more
compact solution can be achieved using the silicon carbide (SiC) or gallium nitride
(GaN) power MOSFETs (packaged as discrete devices or more often as power
modules) which can operate at higher switching frequencies (typically several tens
of kHz) and have substantially lower total losses, while the amperage rating for SiC
devices is less than half of the required amperage for Si IGBTs [22]. Alternatively,
a cheaper and lighter LLCL filter (which has additional inductors connected in
Figure 8.3 The boost-type six-switch PFC converter using IGBTs with input
LLC filter
Figure 8.4 The buck-type six-switch PFC converter with input and output filters
series with the capacitors) can be used instead of the LLC filter to make the system
more efficient [23,24]. The six-switch PWM converter shown in Figure 8.3 has a
boost-type characteristic (the output dc voltage must be higher than the peak input
ac line-to-line voltage), enables sinusoidal input currents, bidirectional power flow
and can operate with high efficiency and arbitrary phase difference between the
input voltage and the input current fundamental [25,26].
The sinusoidal waveform of the input current is achieved by a closed-loop
current control, which requires sensing of the input ac currents. The high turn-on
switching loss in each IGBT as a result of a reverse recovery of the antiparallel
diode of the other IGBT in the same phase leg can be avoided by using SiC
MOSFETs with antiparallel Junction Barrier Schottky (JBS) diodes which practi-
cally do not exhibit any reverse recovery.
If only unidirectional power flow is required, a buck-type six-switch converter
shown in Figure 8.4 can be used. This converter has some advantages over the
Figure 8.5 The Vienna rectifier with input LLC filter
boost-type rectifier from Figure 8.3, such as inherent short-circuit protection,

simple inrush current control and lower output voltage [25]. An additional advan-
tage is that the input current can be controlled in an open loop. The conduction
losses are generally higher than in the case of the boost-type converter, because
more devices are connected in series, but the switching losses can be lower than in
the case of the boost-type converter. The buck-type converter can still operate at
very high efficiency, as reported in [26], in which an efficiency of 98.9% was
achieved. The power flow can be reversed only if the output voltage is reversed.
The achievable phase difference between the input voltage and the input current
fundamental depends on the required output voltage. In order to achieve the higher
phase difference, the converter needs to operate with the reduced output voltage
range (i.e., if a wide output voltage range is required, the phase shift between the
input voltage and input current fundamental needs to be kept small).
The Vienna rectifier [27] shown in Figure 8.5 is another boost-type topology
that can be used as an input stage of a dc fast charger if only unidirectional power
flow is required. Being a three-level converter, the Vienna rectifier offers advan-
tages of 50% lower voltage stress on the used semiconductors switches, lower input
current ripple and less-generated EMI. Its limitations are the unidirectional power
flow, need for a dc-link capacitor voltage balancing and narrow range of phase
difference between the input voltage and the input current which depends on the
output voltage (the range is 30 < f < 30 when the output voltage is higher than
twice the peak input ac line-to-line voltage, and it is reduced to f ¼ 0 if the output
voltage is equal to the peak input ac line-to-line voltage) [26,28,29].
When it comes to the isolated dc/dc stage, several topologies can be used [30]. If
only unidirectional power flow is required, a simple topology shown in Figure 8.6
with the full-bridge inverter and the diode bridge rectifier can be used. The full zero-
voltage switching (ZVS) operation of the active switches can be achieved when
phase-shift PWM control is applied [31]. The main disadvantages of this topology
are relatively high switching losses in the output diodes (even when the SiC JBS
diodes are used), the high voltage spike due to the current–source character of the
rectifying part of the converter (the diode bridge is connected between two current
Figure 8.6 Four-switch unidirectional dc/dc converter with output LC filter
(a) (b) (c)
Figure 8.7 (a) Unidirectional dc/dc converter with active snubber and LC output
filter, (b) passive RCD snubber and (c) passive CDD snubber
sources: the transformer leakage inductance and the output filter inductance) and
severe ringing at the diode bridge output due to the interaction of the transformer
leakage inductance and the parasitic capacitance of the reverse biased diodes. To
reduce the voltage overshoot and the ringing, the different types of snubber circuits
are proposed and applied. The ones that are typically used due to their simplicity and
effectiveness are shown in Figure 8.7. The active snubber circuit shows the best
performance since it can completely eliminate the voltage ringing and the overshoot
[31]. However, that comes with a price of the increased control complexity. The
passive RCD snubber is a simple but effective alternative, and the CDD snubber is a
cost-effective and low-loss version of the passive snubber with better clamping
ability than the RCD snubber [32].
If bidirectional power flow is required, an active rectifier can be used instead
of the diode bridge at the transformer’s secondary. The snubber would still be
required to reduce the ringing across the secondary-side switches.
A dual active bridge (DAB) converter (shown in Figure 8.8) is another bidir-
ectional converter that has been proposed for EV charging applications mainly due
to its high power density and high efficiency, buck-boost operation, low device
stress, small filter components and low sensitivity to system parasitics [33–40].
When introduced in 1991 [40], the DAB converter initially did not receive a
wider adoption due to the high power losses and relatively low switching frequency
Figure 8.8 Dual active bridge converter
(a) (b)
Figure 8.9 Dual half-bridge converter: (a) with voltage-fed output and (b) with
current-fed output
operation of the power semiconductor devices at that time. However, it started

regaining the researchers’ attention with the introduction of new SiC- and GaN-based
power semiconductor devices and the advances in nanocrystalline soft magnetic
materials, which enabled the converter efficiency and power density improvements.
All the switches operate at 50% duty cycle producing square voltage waveforms at
both primary and secondary of the transformer. Using transformer leakage induc-
tance as an energy transfer element, the power flow is controlled by adjusting the
time delay (also called ‘‘phase shift’’) between the primary and the secondary voltage
waveforms. The active power will be transferred from primary to secondary if
the primary waveform is leading the secondary waveform, and vice versa. The ZVS
operation of all switches can be achieved only for a narrow range of input-to-output
voltage ratio, and it cannot be maintained at light load operation. The DAB can be
controlled either by using the voltage mode, current mode or by ‘‘mode-hopping’’
control approach (at light load operation, the system is cycled between the forward
and reverse power flow to maintain the required average output voltage) [33].
Several other converter topologies were proposed for the isolated dc/dc stage,
including dual half-bridge (DHB) with voltage-fed output [41] and with current-fed
output [42], which use only four active switches (two half-bridges, as shown in
Figure 8.9), feature ZVS and bidirectional operation, but have the same high device-
stress problem as DAB when the input voltage range is wider than 2:1 (i.e., wider
than V–2V).
8.3 Microgrid topologies for EV charging

Larger scale penetration of EVs is expected to have significant impacts on power
grids. Since current EVs act as electric loads, the main expected impacts are
increased power demand and energy consumption, increased grid power losses,
more prominent load and voltage profile changes along the power grid and even-
tually requirements to reinforce the grid. With the wider adoption of vehicle-to-grid
(V2G) technology, EVs would be able to act as distributed energy storage and
provide power to the grid while parked. Based on considerations in [43], the EVs
with V2G functionality would not be suitable for baseload power generation
(minimum power that needs to be generated at any given time), since baseload
power can be generated more cheaply by large nuclear or coal-fired power plants.
However, EVs could provide peak power and ancillary services (to serve spinning
reserve and regulation markets) which typically account for 5%–10% of the total
electricity cost. The EVs with V2G technology could be very effective in isolated
power grids (e.g., on islands), where they could enable higher penetration of
renewables [44] and, if EV penetration is high enough, significantly help to reduce
the variations in daily power demand [45]. When it comes to the integration of EVs
into the electric power system, several approaches are proposed. An approach that
includes a local entities called Aggregators, which serve as an interface between the
EVs and the electricity markets, is proposed in [46]. The Aggregators would group
EVs based on their owners’ willingness to participate in the electricity markets and
perform load shifting to provide financial benefits to their clients. Another
approach, proposed in [47], features coordinated charging that can help lower
power losses and voltage deviations by flattening the grid output power peaks. The
approach proposed in [47] features a more integrated way of managing technical
and commercial aspects of EV integration, which requires implementation of smart
metering to coordinate charging and communication between the EVs, the dis-
tribution system operator and the transmission system operator. If no EV charging
coordination is applied, the typical semiurban 15 kV MV grid could accept up to
10% of EVs (out of total number of EVs calculated as 1.5 EV per household),
charged at home at the same time, without any significant changes to the grid
operation and planning procedures [46].
Since EV chargers present a large electric load, it is conceivable to form a
microgrid that would be able to supply multiple EVs simultaneously. Such a system
would try to replicate the design of a conventional gasoline refueling station,
where the same infrastructure is able to supply multiple vehicles by using the same
reservoir, and multiple pumps. By sharing a large common power supply, the
system would benefit from load diversification. The work in [48] explored
the design of the power supply for a system with large number of Level 2 chargers
and showed that for 50 Level 2 chargers, when accounting for diversity, the system
would require a power supply of about 60% of the worst-case (rated power) sce-
nario. Similarly, in case of a system with a single ac/dc conversion stage and a dc
distribution to dc/dc fast charging units, it was found that the average power load
would be substantially lower than the peak power load, and that the grid tie could
be sized for the average power demand, rather than the peak demand [49]. In this
configuration, an energy storage system is connected to the dc bus to supply power
when the demand exceeds the average that can be provided from the grid. It is
shown that over 98% of power demand could be satisfied even with a relatively
small storage system that is rated based on the average value of the needed storage
power. Even though the charging would be delayed for over 60% of the customers,
maximum delay time would be approximately 2 min, and the average delay time
would not be longer than 10 s.
8.3.1 State-of-the-art dc fast charger installation

A typical dc fast charger station consists of an MV/LV service transformer,
switchgear, low-voltage metering and dc fast charging units. State-of-the-art dc fast
chargers are typically rated at 50 kW, installed as single units, and can add up to
100 mi (161 km) of range in about 30 min. Higher power Tesla’s stations are
equipped with multiple 135 kW units (each 135-kW unit serves two charging
stalls), which can add up to 170 mi (273 km) in 30 min. As an example, the circuit
diagram and the rendering of an actual installation with four Tesla Supercharger
units (eight stalls) is illustrated in Figure 8.10 [50].
The dc fast charger and the MV/LV transformer have substantial size and
weight which add cost and complexity to the installation, as both the charger and
the transformer are commonly installed on a concrete foundation. Additionally,
installation in seismically active areas would require seismically restrained con-
crete foundations for both the transformer and the fast charger, further increasing
the installation costs. The single-port state-of-the-art dc fast charger unit cost ran-
ges from $10,000 to $40,000 [51], and the installation costs for a state-of-the-art dc
fast charger unit are in the range from $8,500 to $50,820 [based on the data from
111 units deployed until 2013 and including estimates for another 50þ fast charger
sites, obtained by the Idaho National Laboratory (INL)] [52]. According to the INL
white paper, the main cost drivers include requirements for new electric service,
type of surface material under which the electrical wiring/conduits were installed,
distance from the power source to the service transformer and from the service
transformer to the charger, materials, permits and engineering drawings. The
installation of a charging station with multiple charging stalls is even more costly,
since it involves the installation of the switchgear (shown in Figure 8.10), more
boring/trenching operations and more materials. The result can be a complex and
expensive system that requires significant infrastructure.
Considering that the next generation dc fast chargers will have significantly
higher power ratings to enable ultrafast charging, designing the dc fast charger
stations for future upgradeability may be more cost-effective in a long term, as
concluded from a case study performed in [53], where the cost of community and
corridor dc fast charging complexes was analyzed. If the system is designed with
upgradeability in mind, the initial system cost would be approximately $25,000
higher (for both the community and corridor dc fast charging complexes), but up to
$120,000 would be saved in the long run.
Incoming Distribution panelboard
MV grid
800 A
250 A 250 A 250 A 250 A 15 A
CT 5 5 5 5 3
MV/LV Main breaker

Super- Super- Super- Super-
transformer and metering charger 1 charger 2 charger 3 charger 4
Pedestrian
(a) 1A 1B 2A 2B 3A 3B 4A AB lighting
Distribution Main breaker Incoming

panelboard and metering
Tesla Superchargers 1–4
MV/LV transformer
(b) 24′-2″(7.36 m) × 10′-3″(3.12 m)

5′(1.52 m) × 5′-8″(1.72 m)
Figure 8.10 State-of-the-art dc fast charger installation with four Tesla

Supercharger units: (a) one-line circuit diagram, (b) rendering of the
system (charging stalls and seclusion composite fence surrounding
the equipment are not shown)
As already mentioned [49], adding the energy storage to the dc fast charging
system could minimize the peak demand charges and therefore reduce the operat-
ing costs. The added energy storage needs to be sized properly to be capable of
buffering the power and energy demands of the charging station during the peak
demand period. During the off-peak hours, the stored energy can be used to provide
ancillary services for improved grid stability and security. The onsite photovoltaic
energy generation system can also be added to the charging station to improve
decoupling of power provided to the vehicles from the power drawn from the grid.
State-of-the-art systems are typically coupled on the LV ac side through ac-to-dc or
dc-to-ac converters, as illustrated in Figure 8.11(a). Even though bidirectional
operation is enabled by charging standards, majority of modern dc fast chargers
operate in unidirectional mode.
One significant advantage of using the ac bus is the availability and maturity of
the rectifier and inverter technology, availability of ac switchgear and protection
devices, and well-established standards and practices for the ac power distribution
systems. On the other hand, having more conversion stages in the ac-coupled
Energy
PV storage
dc/dc dc/dc
MV
grid
MV/LV
dc/ac dc/ac
ac/dc ac/dc
dc/dc dc/dc
(a)
PV Energy
storage
MV
grid
MV/LV
dc/dc dc/dc
ac/dc
dc/dc dc/dc
(b)
Figure 8.11 Microgrid configurations for EV charging applications: (a) typically

used ac microgrid and (b) dc microgrid
system (to interface dc generation, energy storage and dc loads to the ac system)
increases the system complexity and cost and decreases the system efficiency.
Also, dealing with reactive power, inverter synchronization, and voltage and fre-
quency control during islanded operation of the ac-connected systems makes the
ac systems more complicated to control than the dc-connected systems. A dc-
connected system, illustrated in Figure 8.11(b), features one centralized ac/dc
converter and provides more convenient and more energy efficient way of con-
necting the energy storage and the renewable energy sources to the central dc bus.
The test results of a simple ac-coupled system with 20 kW (16 kWh) lithium–
polymer battery storage and an EV charger commercial prototype with maximum
50 kW dc and 22 kW ac output power are reported in [54]. The battery voltage was
first boosted to 600 V and then fed to an inverter which was connected to the 400-V
ac grid through the low-frequency isolation transformer. The peak power shaving
ability is demonstrated for two different cases: charging Nissan Leaf from 65%
SOC for approximately 10 min, and from 40% to 100% SOC. During the second
test, the energy delivered from the energy storage was 4.5 kWh, which was 36% of
the total energy absorbed by the charger (12.4 kWh). An implementation of
dc-coupled system is reported in [55], where 288 V/40 Ah (11.5 kWh) lead-acid
battery is used as a storage to reduce the power drawn from the ac grid during the
fast charging of the 50 V/40 Ah (2 kWh) LiFePO4 battery used in two wheelers.
8.3.2 Medium-voltage dc fast chargers: a new approach

to high-power EV fast charging
Another approach to EV fast charging is to connect the fast charger directly to MV
distribution line and eliminate the low-frequency MV/LV distribution transformer.
This approach is based on solid-state transformer (SST) technology and wide
bandgap (WBG) semiconductor power devices (typically 1.2 kV SiC MOSFETs)
and offers several benefits, such as higher efficiency, higher power density and
system-level cost savings, compared to the SOA transformer-and-charger
approach. Eliminating low-frequency distribution transformer can be particularly
beneficial if the SST-based dc fast charger is installed in densely populated areas.
The SST-based MVAC/LVDC (or MVAC/MVDC) converter provides either uni-
directional or bidirectional power flow, regulates active and reactive power, per-
forms PFC function and provides isolation and protection features. The typical
implementation of a system that uses this approach is illustrated in Figure 8.12
[56,57]. The system uses a common dc bus to interface renewable energy genera-
tion systems and energy storage to EVs via dc/dc converters only, ensuring reduced
number of conversion stages and high efficiency. Several studies and SST-based
EV fast charger prototypes demonstrating these benefits have been reported in the
literature.
The Utility Direct Fast Charger (UDFC) has been developed and demonstrated
by the Electric Power Research Institute in 2013 [58,59]. Two prototype dc fast
chargers were demonstrated: a 50-kW prototype that connects to the 2.4-kV ac
Energy
PV
storage
MV grid
SST
dc/dc dc/dc
HF TR
ac/dc dc/dc
dc/dc dc/dc
Figure 8.12 An SST-based dc microgrid that connects directly to MV distribution

grid
State-of-the-art system SST-based MV solution

Power: 50 kW Power: 50 kW
Volume: 4,300 L Volume: 140 L
Mass: 1,600 kg Mass: 100 kg
Efficiency: 92.6% EV charger Efficiency: 96.6%
Service transformer
Figure 8.13 Comparison of the SOA EV charger system and the SST-based
MV solution from [60]
single-phase line and a 25-kW prototype that connects to the 7.2-kV ac single-phase
line. In both the cases, the rated output voltage was 400 V and the grid-to-vehicle
efficiency was higher than 96%. Another, more recent effort, has demonstrated a
modular 50-kW dc fast charger that uses 1.2 kV SiC devices, connects to 2.4 kV ac
single-phase line, outputs 400 V and has the efficiency of over 96% [60]. The ben-
efits of the MV approach with SiC devices are illustrated in Figure 8.13.
As previously mentioned, a dedicated service transformer is typically required
to step down the distribution system medium voltage and provide three-phase supply
to the commercial dc fast charger. With the service transformer’s efficiency of
98.5% (at 50 kVA level), the grid-to-battery efficiency of the 50-kW transformer-
and-charger systems is less than 93%. The efficiency of higher power chargers
(135 kW Tesla Supercharger or 150 kW Delta Stromtankstelle 4.0) do not exceed
93% while still requiring bulky low-voltage service transformers to provide
208/480 V three-phase input, which itself has an efficiency below 99% (at 150 kVA
power level). The efficiency of the entire system with the 150-kV charger is therefore
approximately 92%. Compared to the SOA system, the MV dc fast charger prototype
reported in [60] reduces the size and the weight by ten-fold and improves the system
efficiency from 93% to over 96%—a reduction in losses by approximately 50%.
Additionally, by connecting directly to the MV line, the installation costs can be
reduced by at least 40%, compared to the conventional solution [61].
The MV EV fast chargers are designed to be modular, using identical modules
as building blocks to reach the desired power and voltage levels. The modules are
typically connected in series at the input and in parallel at the output, as illustrated
in Figure 8.14 [58,60,62,63]. The series connection at the input allows the rela-
tively low-voltage off-the-shelf silicon IGBTs or SiC MOSFETs to serve the
medium-voltage application. The use of SiC devices substantially increases the
Module 1 Module 1 200 – 450 V
800 V 200 – 500 V

625 V
800 V 625 V
2.4 kV 2.4 kV
Module 2 Module 2
Module 3 Module 3
(a) (b)
Module 1 Module 1
VOUT VOUT
VIN VIN
Module 2 Module 2
Module N Module N
(c) (d)
Figure 8.14 Input-series–output-parallel MV charger topologies: (a) multicell boost (MCB) topology [60], (b) UDFC topology [58],
(c) cascaded H-bridge topology with integrated split battery [62], (d) cascaded H-Bridge with current-fed DAB [63]
efficiency of the system compared to the use of the silicon equivalents, due to
the lower switching and conduction losses. Additionally, the high switching fre-
quency operation of SiC MOSFETs enables significant reduction in size of the
inductors and the transformers used to provide the required galvanic isolation, thus
reducing the overall system size and weight.
With the input-series connection, a special care must be taken to balance the
voltage of dc link capacitors in each module, while ensuring the low total harmonic
distortion (THD) of the input current [63,64]. The high isolation of gate drive
power supplies is required for safe operation at high common-mode voltage. Low
coupling capacitance of gate drive power supplies and power transformers is
essential for reduction of the conducted EMI, especially if high-speed WBG devi-
ces are used. Another significant technical challenge is related to the design and
implementation of the fast-acting system protections and safety features at medium
voltage level. The system must be able to withstand Basic Insulation Level tests
and low-frequency-voltage insulation tests, where the converter is exposed to vol-
tages significantly higher than its rated voltage. Therefore, additional circuitry is
required to clamp the voltage seen by the converter in order to meet these
requirements. Furthermore, the converter must have an internal protection
mechanism that clears any system fault in a few microseconds if the converter
devices are to be protected and the system is to continue to operate after the fault is
cleared. Such fast clearing of a fault cannot be achieved using conventional MV
mechanical switches, and solid-state solutions are required.
The design of a 16.7-kW isolated dc/dc MCB module used in a 50-kW MV fast
charger prototype from [60] is illustrated in Figure 8.15 [65]. The module topology
uses low number of active semiconductor switches, has high utilization of the
switches, and enables independent control of input and output stages and simple dc
bus capacitor voltage balancing control [66]. The switching frequency of the three-
level boost stage is set to 25 kHz and the minimum switching frequency is limited
by the required THD of the input current. The switching frequency of the NPC
converter is set to 50 kHz.
The SST-based MV EV charger system can service many other applications,
including industrial applications, data centers, and aerospace and military appli-
cations. As an example, the reduction of weight and volume is particularly sig-
nificant in naval shipboard applications, where the transformers for a redundant
propulsion system can weigh as much as 35 t. The power converters for shipboard
applications need to have high power density in order to save space and fuel, and dc
MV systems with 100 kHz switching frequencies are seen as an emerging bench-
mark for these systems in the 2025–30 time frame [67].
The proposed high-power medium-voltage modular rectifier can also serve as a
power supply for data centers. According to DOE’s Center of Expertise for Energy
Efficiency in Data Center’s recent report [68], data centers in the United States
consumed an estimated 70 billion kWh in 2014, representing about 1.8% of the
total US electricity consumption. Additionally, based on current trends, US data
centers are projected to consume approximately 73 billion kWh in 2020. Most of
these data centers (except high-end and hyperscale centers) operate with the Power
3-level boost + NPC converter Output inductor
DC link capacitors HF power transformer Output diode bridge + Snubber
3-level boost + NPC converter
Output diode bridge + snubber
DC link capacitors
HF power transformer
Output inductor
Figure 8.15 16.7 kW isolated dc/dc module as a building block for MV EV fast
charger [65]
Usage Effectiveness of approximately 2, which means that half of the used energy
is consumed by the data center’s supporting infrastructure (cooling equipment,
uninterrupted power supplies, lighting, etc.). As a result, any power savings that can
be made in the operation of the data center will have significant environmental
implications as well as impacts on the system cost of operation. Recently, data
centers with 380 V dc distribution systems have been proposed, with their inher-
ently higher efficiency due to the fewer conversion stages. The SST-based topology
used for MV EV chargers can be used to efficiently supply 380 V dc to the data
centers. The benefits of higher system efficiency, better power conditioning and
better quality power can all be achieved with the SST-based design [69].
8.4 Conclusions and future trends

A number of cost-competitive EVs with a driving range of over 200 mi have
recently entered the EV marketplace. Also, a number of countries have stated that
they plan to end the sales of fossil-fuel-powered cars in the near future, further
accelerating the shift to electric transportation. In parallel, advances in lithium ion
battery technology promise to deliver better charge acceptance, allowing even
faster charging rates. To serve this new vehicle fleet, there will be a need for fast
chargers with a very large charging capacity. As a result, a number of automakers
have reached the consensus that the long-term goal is to supply up to 350 kW to a
single vehicle in order to enable a gas station-like experience. One way of pro-
viding the fast EV charging services while maintaining the grid stability and
security would be through the utilization of the SST-based energy routers, which
would provide both ac and dc grid interconnectivity, grid intelligence and neces-
sary protection features, as illustrated in Figure 8.16. This is part of the Future
Renewable Electric Energy Delivery and Management visions to enable the
seamless interaction of distributed energy resources, distributed intelligence and
advanced power electronics [70]. The SST provides the potential for revolutionizing
MV 3Φ distribution
SST
ac
dc ac
dc
dc
Breaker dc
dc
dc
DESD
Net meter dc Battery
dc
Utility owned
Breaker
dc Battery
dc
Breaker Breaker
dc dc
800 V
dc dc
Breaker 400 V
Net meter dc
dc
dc bus ac bus
Figure 8.16 FREDM vision with SST as an intelligent energy router

the power distribution system by completely decoupling the medium voltage dis-
tribution grid from the low voltage feeder sections. The bidirectional power flow
capability of the SST provides possibilities to integrate the distributed energy
storage devices and to feed locally generated power back to the grid in a controlled
way. Recent technology advances in WBG semiconductor devices, such as high-
voltage SiC MOSFETs and diodes, serve as the enabling technology for the SST.
Despite its unparalleled capabilities, significant challenges stand in the way of
the complete adoption of the SST by the electric utilities today. The key barriers
include (1) inherent reliability concerns of replacing the passive transformer with a
power electronic device, (2) relatively low penetration of renewable resources seen
in the system today, which can mostly be handled by the existing infrastructure,
(3) the large inertia still present in the distribution system, supplied by the legacy
generators and (4) limited ability to monetize the perfect power quality supplied by
the SST. As a result, the other venues to demonstrate the SST technology can be
explored and showcase the benefits of direct connection of power electronics to the
MV grid. Medium-voltage SSTs using WBG devices can serve many other emer-
ging and established applications including EV fast chargers, data center power
supplies and military power conversion stages, to name a few. The EV fast chargers
present the largest opportunity, given the size and the rate of growth of the market.
Further, the use of a medium-voltage dc fast charger has clear monetary advantages
since this solution will substantially reduce the installation cost, which makes up to
half of the price of deploying a commercial dc fast charger, and it is particularly
suitable for deployment in densely populated urban areas. Overall, the SST-based
approach has potential to help reduce the EV charging time and bring a gas station
experience to the EV users, while providing the integration of renewable resources
and energy storage in a cost-competitive way.
References
[1] International Energy Agency (IEA) and the Electric Vehicles Initiative of the
Clean Energy Ministerial (EVI), ‘‘Global EV Outlook 2017,’’ OECD/IEA,
Paris, France, Jul. 2017.
[2] ‘‘Monthly Plug-In Sales Scorecard’’. [Online]. Available: https://insideevs.
com/monthly-plug-in-sales-scorecard/. Accessed on Jan. 5, 2018.
[3] ‘‘Compare Electric Cars Side-by-Side’’. [Online]. Available: https://www.
fueleconomy.gov/feg/evsbs.shtml. Accessed on Jan. 5, 2018.
[4] D. Howell, S. Ahmed, R. B. Carlson, et al., ‘‘Enabling Fast Charging: A
Technology Gap Assessment,’’ U.S. Department of Energy, Office of Energy
efficiency & Renewable Energy, Report INL/EXT-17-41638, Oct. 2017.
[Online]. Available: https://energy.gov/sites/prod/files/2017/10/f38/XFC%
20Technology%20Gap%20Assessment%20Report_FINAL_10202017.pdf.
[5] ‘‘Largest Global Charger Coverage’’. [Online]. Available: https://www.
chademo.com/. Accessed on Jan. 5, 2018.
[6] CHAdeMO, ‘‘2016 Activity Report (1 April 2016–31 March 2017)’’,

CHAdeMO Association. [Online]. Available: http://www.chademo.com/
wp2016/wp-content/japan-uploads/2017GA/Activityreport2016en.pdf.
[7] Chinese National Standard, ‘‘Connection Set for Conductive Charging of
Electric Vehicles – Part 3: DC Charging Coupler,’’ National Standard of the
People’s Republic of China, 2015, GB/T 20234.3-2015.
[8] D. Hall, N. Lutsey, ‘‘Emerging Best Practices for Electric Vehicle Charging
Infrastructure,’’ International Council on Clean Transportation, Washington,
DC, Oct. 2017. [Online]. Available: https://www.theicct.org/sites/default/
files/publications/EV-charging-best-practices_ICCT-white-paper_04102017_
vF.pdf.
[9] ‘‘Articles of Association of Charging Interface Initiative e.V.,’’ Berlin, May
19, 2015. [Online]. Available: http://charinev.org/fileadmin/Downloads/
Association_Documents/ArticlesOfAssociation_CharIN.pdf.
[10] ‘‘DC-CCS-Charging Stations in the Main Markets’’. [Online]. Available:
http://charinev.org/ccs-at-a-glance/infrastructure/. Accessed on Jan. 5, 2018.
[11] Combined Charging System Standard Specification 1.0, ‘‘Combined Char-
ging System 1.0 Specification – CCS 1.0,’’ Coordination Office Charging
Interface, c/o innos-Sperlich GmbH, Jan. 2017.
[12] ‘‘EVgo and ABB to Deploy Nation’s First High-Power Electric Vehicle Fast
Charging Station,’’ Press Release, Feb. 27, 2017. [Online]. Available:
https://www.evgo.com/about/news/evgo-abb-deploy-nations-first-high-power-
electric-vehicle-fast-charging-station/.
[13] ‘‘Europe’s First Public Ultra-Fast Charging Station Now Operational,’’ Press
Release, Dec. 20, 2017. [Online]. Available: https://www.allego.eu/wp-content/
uploads/2017/12/ENG-20171221_Ultra-E_PM.pdf.
[14] ‘‘Charging Interface Initiative e.V. – Industry Statement on Future Charging
Infrastructure,’’ Coordination Office Charging Interface, c/o innos-Sperlich
GmbH. Feb. 20, 2017. [Online]. Available: http://www.charinev.org/fileadmin/
Downloads/Papers_and_Regulations/CharIN_industry_statement.pdf.
[15] ‘‘Superchargers’’. [Online]. Available: https://www.tesla.com/supercharger.
Accessed on Jan. 5, 2018.
[16] The Tesla Team, ‘‘Supercharging Cities,’’ Sep. 11, 2017. [Online]. Available:
https://www.tesla.com/blog/supercharging-cities.
[17] ‘‘2014 Tesla Model S 85 kWh Advanced Vehicle Testing – Baseline Vehicle
Testing Results’’, Idaho National Laboratory, Factsheet INL/MIS-15-34211,
Mar. 31, 2016. [Online]. Available: https://avt.inl.gov/sites/default/files/pdf/
fsev/fact4500tesla2014.pdf.
[18] E. Loveday, ‘‘Rare Look Inside Tesla Supercharger’’. Jun. 24, 2014. [Online].
Available: https://insideevs.com/rare-look-inside-tesla-supercharger/.
[19] ‘‘Supercharge Data from Cross-Country Road Trip’’. Dec. 8, 2014.
[Online]. Available: https://teslamotorsclub.com/tmc/threads/supercharge-
data-from-cross-country-road-trip.39184/.
[20] T. Anegawa, ‘‘Safety Design of CHAdeMO Quick Charging System,’’ in
World Electric Vehicle Journal, vol. 4, no. 4, pp. 855–859, 2010.
[21] D. Aggeler, F. Canales, H. Zelaya, et al., ‘‘Ultra-fast DC-charge infrastructures

for EV-mobility and future smart grids,’’ 2010 IEEE PES Innovative Smart Grid
Technologies Conference Europe (ISGT Europe), Gothenburg, 2010, pp. 1–8.
[22] J. Casady, ‘‘Power products commercial roadmap for SiC from 2012–2020,’’
Cree Power – Sept 2014 HMW Direct-Drive Motor Workshop, 2014.
[Online]. Available: https://www.nist.gov/document-2374.
[23] W. Wu, Y. He and F. Blaabjerg, ‘‘An LLCL Power Filter for Single-Phase
Grid-Tied Inverter,’’ in IEEE Transactions on Power Electronics, vol. 27,
no. 2, pp. 782–789, Feb. 2012.
[24] W. Wu, M. Huang, F. Blaabjerg, ‘‘Efficiency Comparison between the
LLCL and LCL-Filters Based Single-Phase Grid-Tied Inverters,’’ in
Archives of Electrical Engineering, vol. 63, no. 1, pp. 63–79, 2014.
[25] H. Mao, C. Y. Lee, D. Boroyevich and S. Hiti, ‘‘Review of High-Performance
Three-Phase Power-Factor Correction Circuits,’’ in IEEE Transactions on
Industrial Electronics, vol. 44, no. 4, pp. 437–446, Aug. 1997.
[26] J. W. Kolar, T. Friedli, ‘‘The Essence of Three-Phase PFC Rectifier
Systems—Part I,’’ in IEEE Transactions on Power Electronics, vol. 28,
no. 1, pp. 176–198, Jan. 2013.
[27] J. W. Kolar and F. C. Zach, ‘‘A novel three-phase three-switch three-level
PWM rectifier,’’ Proc. 28th Power Conversion Conference, PCIM‘94, Jun.
28–30, 1994, pp. 125–138.
[28] R. Lai, F. Wang, R. Burgos, D. Boroyevich, D. Jiang and D. Zhang,
‘‘Average Modeling and Control Design for VIENNA-Type Rectifiers
Considering the DC-Link Voltage Balance,’’ in IEEE Transactions on Power
Electronics, vol. 24, no. 11, pp. 2509–2522, Nov. 2009.
[29] J. Liu, W. Ding, Han Qiu, C. Zhang and B. Duan, ‘‘Neutral-point voltage
balance control and oscillation suppression for VIENNA rectifier,’’ 2017
IEEE 3rd International Future Energy Electronics Conference and ECCE
Asia (IFEEC 2017 – ECCE Asia), Kaohsiung, 2017, pp. 1275–1279.
[30] Y. Du, S. Lukic, B. Jacobson and A. Huang, ‘‘Review of high power isolated
bi-directional DC–DC converters for PHEV/EV DC charging infra-
structure,’’ 2011 IEEE Energy Conversion Congress and Exposition,
Phoenix, AZ, 2011, pp. 553–560.
[31] J. A. Sabate, V. Vlatkovic, R. B. Ridley and F. C. Lee, ‘‘High-voltage, high-
power, ZVS, full-bridge PWM converter employing an active snubber,’’
Conference Proceedings of the Sixth Annual Applied Power Electronics
Conference and Exposition, APEC‘91, Dallas, TX, 1991, pp. 158–163.
[32] J.-G. Cho, J.-W. Baek, C.-Y. Jeong and G.-H. Rim, ‘‘Novel Zero-Voltage
and Zero-Current-Switching Full-Bridge PWM Converter Using a Simple
Auxiliary Circuit,’’ in IEEE Transactions on Industry Applications, vol. 35,
no. 1, pp. 15–20, Jan./Feb. 1999.
[33] M. N. Kheraluwala, R. W. Gascoigne, D. M. Divan and E. D. Baumann,
‘‘Performance Characterization of a High-Power Dual Active Bridge DC-to-
DC Converter,’’ in IEEE Transactions on Industry Applications, vol. 28,
no. 6, pp. 1294–1301, Nov./Dec. 1992.
[34] S. Han and D. Divan, ‘‘Bi-directional DC/DC converters for plug-in hybrid
electric vehicle (PHEV) applications,’’ 2008 Twenty-Third Annual IEEE
Applied Power Electronics Conference and Exposition, Austin, TX, 2008,
pp. 784–789.
[35] C. Mi, H. Bai, C. Wang and S. Gargies, ‘‘Operation, Design and Control of
Dual H-Bridge-based Isolated Bidirectional DC–DC Converter,’’ in IET
Power Electronics, vol. 1, no. 4, pp. 507–517, Dec. 2008.
[36] R. J. Ferreira, L. M. Miranda, R. E. Araújo and J. P. Lopes, ‘‘A new
bi-directional charger for vehicle-to-grid integration,’’ 2011 2nd IEEE PES
International Conference and Exhibition on Innovative Smart Grid Tech-
nologies, Manchester, 2011, pp. 1–5.
[37] H. van Hoek, M. Neubert and R. W. De Doncker, ‘‘Enhanced Modulation
Strategy for a Three-Phase Dual Active Bridge—Boosting Efficiency of an
Electric Vehicle Converter,’’ in IEEE Transactions on Power Electronics,
vol. 28, no. 12, pp. 5499–5507, Dec. 2013.
[38] L. Xue, Z. Shen, D. Boroyevich, P. Mattavelli and D. Diaz, ‘‘Dual Active
Bridge-Based Battery Charger for Plug-in Hybrid Electric Vehicle with
Charging Current Containing Low Frequency Ripple,’’ in IEEE Transactions
on Power Electronics, vol. 30, no. 12, pp. 7299–7307, Dec. 2015.
[39] L. Xue, M. Mu, D. Boroyevich and P. Mattavelli, ‘‘The optimal design of
GaN-based Dual Active Bridge for bi-directional Plug-IN Hybrid Electric
Vehicle (PHEV) charger,’’ 2015 IEEE Applied Power Electronics
Conference and Exposition (APEC), Charlotte, NC, 2015, pp. 602–608.
[40] R. W. A. A. De Doncker, D. M. Divan and M. H. Kheraluwala, ‘‘A Three-
Phase Soft-Switched High-Power-Density DC/DC Converter for High-
Power Applications,’’ in IEEE Transactions on Industry Applications,
vol. 27, no. 1, pp. 63–73, Jan./Feb. 1991.
[41] H. Li, D. Liu, F. Z. Peng and G.-J. Su, ‘‘Small signal analysis of a dual half
bridge isolated ZVS Bi-directional DC–DC converter for electrical vehicle
applications,’’ 2005 IEEE 36th Power Electronics Specialists Conference,
Recife, 2005, pp. 2777–2782.
[42] M. Vasiladiotis, B. Bahrani, N. Burger and A. Rufer, ‘‘Modular converter
architecture for medium voltage ultra fast EV charging stations: Dual half-
bridge-based isolation stage,’’ 2014 International Power Electronics Conference
(IPEC-Hiroshima 2014 – ECCE ASIA), Hiroshima, 2014, pp. 1386–1393.
[43] W. Kempton, J. Tomić, ‘‘Vehicle-to-Grid Power Fundamentals: Calculating
Capacity and Net Revenue,’’ in Journal of Power Sources, vol. 144, no. 1,
pp. 268–279, 2005.
[44] J. A. P. Lopes, P. M. R. Almeida and F. J. Soares, ‘‘Using vehicle-to-grid to
maximize the integration of intermittent renewable energy resources in
islanded electric grids,’’ 2009 International Conference on Clean Electrical
Power, Capri, 2009, pp. 290–295.
[45] P. Kadurek, C. Ioakimidis and P. Ferrao, ‘‘Electric vehicles and their impact
to the electric grid in isolated systems,’’ 2009 International Conference on
Power Engineering, Energy and Electrical Drives, Lisbon, 2009, pp. 49–54.
[46] J. A. P. Lopes, F. J. Soares and P. M. R. Almeida, ‘‘Integration of Electric

Vehicles in the Electric Power System,’’ in Proceedings of the IEEE, vol. 99,
no. 1, pp. 168–183, Jan. 2011.
[47] K. Clement-Nyns, E. Haesen and J. Driesen, ‘‘The Impact of Charging Plu-
g-In Hybrid Electric Vehicles on a Residential Distribution Grid,’’ in IEEE
Transactions on Power Systems, vol. 25, no. 1, pp. 371–380, Feb. 2010.
[48] S. Hutchinson, M. Baran and S. Lukic, ‘‘Power supply for an electric vehicle
charging system for a large parking deck,’’ 2009 IEEE Industry Applications
Society Annual Meeting, Houston, TX, 2009, pp. 1–4.
[49] S. Bai, D. Yu and S. Lukic, ‘‘Optimum design of an EV/PHEV charging
station with DC bus and storage system,’’ 2010 IEEE Energy Conversion
Congress and Exposition, Atlanta, GA, 2010, pp. 1178–1184.
[50] ‘‘Tesla Supercharging station Madison, WI009 Madison, 3801 E Washington
Ave, Madison, WI 53704’’, Tesla, Inc. Jun. 2017. [Online]. Available: https://
www.cityofmadison.com/dpced/planning/documents/3801ewa_site.pdf.
[51] M. Smith and J. Castellano, ‘‘Costs Associated with Non-Residential Electric
Vehicle Supply Equipment – Factors to Consider in the Implementation of
Electric Vehicle Charging Stations,’’ U.S. Department of Energy Vehicle
Technologies Office, no. DOE/EE-1289, Nov. 2015. [Online]. Available:
https://www.afdc.energy.gov/uploads/publication/evse_cost_report_2015.pdf.
[52] Idaho National Laboratory (INL) 2015. ‘‘What were the Cost Drivers for the
Direct Current Fast Charging Installations?’’ INL/MIS-15-35060, Mar. 2015.
[Online]. Available: http://avt.inel.gov/pdf/EVProj/WhatWereTheCostDrivers
ForDCFCinstallations.pdf.
[53] J. Francfort, S. Salisbury, J. Smart, T. Garetson and D. Karner, ‘‘Considerations
for Corridor and Community DC Fast Charging Complex System Design,’’
INL/EXT-17-40829, Idaho National Laboratory, Mar. 2017. [Online]. Avail-
able: https://avt.inl.gov/sites/default/files/pdf/reports/DCFCChargingComplex
SystemDesign.pdf.
[54] D. Sbordone, I. Bertini, B. Di Pietra, M. C. Falvo, A. Genovese and
L. Martirano, ‘‘EV Fast Charging Stations and Energy Storage Technolo-
gies: A Real Implementation in the Smart Micro Grid Paradigm,’’ in Electric
Power Systems Research, vol. 120, pp. 96–108, 2015.
[55] O. Veneri, C. Capasso and D. Iannuzzi, ‘‘Experimental Evaluation of DC
Charging Architecture for Fully-Electrified Low-Power Two-Wheeler,’’ in
Applied Energy, vol. 162, pp. 1428–1438, 2016.
[56] X. Yu, X. She, X. Ni and A. Q. Huang, ‘‘System Integration and Hierarchical
Power Management Strategy for a Solid-State Transformer Interfaced
Microgrid System,’’ in IEEE Transactions on Power Electronics, vol. 29,
no. 8, pp. 4414–4425, Aug. 2014.
[57] R. Crosier, S. Wang and M. Jamshidi, ‘‘A 4800-V grid-connected electric
vehicle charging station that provides STACOM-APF functions with a bi-
directional, multi-level, cascaded converter,’’ 2012 Twenty-Seventh Annual
IEEE Applied Power Electronics Conference and Exposition (APEC),
Orlando, FL, 2012, pp. 1508–1515.
[58] EPRI, Utility Direct Medium Voltage DC Fast Charger Update: DC Fast
Charger Characterization. EPRI, Palo Alto, CA: 2012. 1024106.
[59] S. Rajagopalan, A. Maitra, J. Halliwell, M. Davis and M. Duvall, ‘‘Fast
charging: An in-depth look at market penetration, charging characteristics,
and advanced technologies,’’ 2013 World Electric Vehicle Symposium and
Exhibition (EVS27), Barcelona, 2013, pp. 1–11.
[60] S. Srdic, X. Liang, C. Zhang, W. Yu and S. Lukic, ‘‘A SiC-based high-
performance medium-voltage fast charger for plug-in electric vehicles,’’
2016 IEEE Energy Conversion Congress and Exposition (ECCE), Milwaukee,
WI, 2016, pp. 1–6.
[61] A. Maitra, ‘‘EPRI’s utility direct DC fast charger–development, testing,
demonstration,’’ EPRI IWC Meeting, Atlanta, GA, Mar. 28, 2012.
[62] M. Vasiladiotis and A. Rufer, ‘‘A Modular Multiport Power Electronic
Transformer with Integrated Split Battery Energy Storage for Versatile
Ultrafast EV Charging Stations,’’ in IEEE Transactions on Industrial Elec-
tronics, vol. 62, no. 5, pp. 3213–3222, May 2015.
[63] D. Sha, G. Xu and Y. Xu, ‘‘Utility Direct Interfaced Charger/Discharger
Employing Unified Voltage Balance Control for Cascaded H-Bridge Units
and Decentralized Control for CF-DAB Modules,’’ in IEEE Transactions on
Industrial Electronics, vol. 64, no. 10, pp. 7831–7841, Oct. 2017.
[64] X. Liang, C. Zhang, S. Srdic and S. Lukic, ‘‘Predictive Control of a
Series-Interleaved Multi-Cell Three-Level Boost Power Factor Correction
Converter,’’ in IEEE Transactions on Power Electronics, vol. 33, no. 10,
pp. 8948–8960, Oct. 2018.
[65] S. Srdic, C. Zhang, X. Liang, W. Yu and S. Lukic, ‘‘A SiC-based power
converter module for medium-voltage fast charger for plug-in electric
vehicles,’’ 2016 IEEE Applied Power Electronics Conference and Exposi-
tion (APEC), Long Beach, CA, 2016, pp. 2714–2719.
[66] D. Rothmund, G. Ortiz and J. W. Kolar, ‘‘SiC-based unidirectional solid-
state transformer concepts for directly interfacing 400 V DC to medium-
voltage AC distribution systems,’’ 2014 IEEE 36th International
Telecommunications Energy Conference (INTELEC), Vancouver, BC,
2014, pp. 1–9.
[67] F. Wang, Z. Zhang, T. Ericsen, R. Raju, R. Burgos and D. Boroyevich,
‘‘Advances in Power Conversion and Drives for Shipboard Systems,’’ in
Proceedings of the IEEE, vol. 103, no. 12, pp. 2285–2311, Dec. 2015.
[68] A. Shehabi, S. Smith, D. Sartor, et al., ‘‘United States Data Center Energy
Usage Report,’’ Lawrence Berkeley National Laboratory, Jun. 2016. [Online].
Available: https://eta.lbl.gov/sites/all/files/publications/lbnl-1005775_v2.pdf.
[69] G. Zhabelova, A. Yavarian, V. Vyatkin and A. Q. Huang, ‘‘Data center
energy efficiency and power quality: An alternative approach with solid state
transformer,’’ IECON 2015 – 41st Annual Conference of the IEEE Industrial
Electronics Society, Yokohama, 2015, pp. 1294–1300.
[70] A. Huang, ‘‘FREEDM system – a vision for the future grid,’’ IEEE PES
General Meeting, Minneapolis, MN, 2010, pp. 1–4.
Chapter 9
Overview and design of solid-state transformers
Levy Costa1, Marco Liserre1, and Giampaolo Buticchi1
9.1 Solid-state transformer: concept
The concept of electronic transformer was first introduced by [1] in the end of
1960s, where the main concept would be to adjust the output voltage by using the
power electronics, while contributing to size and volume reduction. The apparatus
in question isolated the primary and secondary side in frequency higher than the
grid frequency and for that reason the size, weight and volume could be con-
siderably reduced. At that time, there was no real applicability for such apparatus,
mainly because of the performance limitation of the available semiconductors on
the market.
However, with the advancement of the power electronics and semiconductors
technology, new faster devices with lower switching energy were developed,
enabling faster switching frequency operation. As a result, the volume and size of
the power converters were considerably reduced, bringing advantages for those
applications in which the power density is very important. At the same time, the
classic solution used in traction for the electric locomotive was very heavy, bulky
and inefficient, because of the low frequency transformer (LFT) employed [2]
associated to the power electronics converters required for providing the suitable
dc voltage to the variable speed drive on the locomotive. In this sense, the power
electronics transformer or solid-state transformer (SST), as usually called, could
provide the suitable controlled and isolated dc voltage in a unique solution, offering
either volume and weight reduction (around 20%–50%) and efficiency improve-
ment (from 93% to 96%) [2,3]. Thus, the SST became a reality in traction
applications.
Figure 9.1 illustrates the conventional solution (state of the art) used on the
locomotives, as well as the modern solution based on SST. The benefits offered by
the SST in traction application are mainly on volume and weight reduction. Thus,
the advantages of the system are only related to its hardware. The SST has been
frequently conceptualized as presented below.
1
Chair of Power Electronics Faculty of Engineering, Christian-Albrechts-University of Kiel, Germany.
MVAC MVAC
Conventional solution SST-based solution
LFT Modernization
Traction
ac dc Solid-
M state M
trans
dc ac
Weight and volume reduction
Figure 9.1 Solid-state transformer in traction application. MVAC, medium

voltage ac; LFT, low frequency transformer
Solid-state transformer (SST):

Power electronics-based system with galvanic isolation in medium or high
frequency between the input and output having waveform conditioning
capability. It is normally employed to replace the traditional transformer,
offering advantages in volume and weight reduction. Thus, the advantages
offered by the SST are related to its hardware stage.
9.2 SST in electric distribution grid application

Apart from the advantages achieved in terms of hardware, the power electronics
involved in the SST enables the implementation of an advanced control strategy with
multiple control loops (e.g., power, voltage and current). This extra ability enlarges
the SST application to other fields, such as electric distribution system, in which
power and voltage control is the central requirement. Thus, the SST is very suitable
for electric distribution system application, because it can perform as an LFT
(replacing it), while providing ancillary services to the grid through its high-advanced
control platform, improving the overall power quality of the distribution grid.
Therefore, this technology has received a lot of attention in the research community
(even academia and industry), and it was cited as one of the most emerging tech-
nologies by Massachusetts Institute of Technology Review in 2010 [4–6].
The new functionalities in the form of ancillary services to the electric grid are
discussed in [3,7]. Among them, the reverse power flow condition, storage inte-
gration, management of hybrid grids (dc and ac) and power quality improvement
are highlighted. Then, in this application, the SST is not only supposed to replace
the conventional LFT, connecting the medium-voltage (MV) grid to the low-
voltage (LV) grid, but also to offer dc connectivity and services to both LV and MV
grids, as highlighted in Figure 9.2. In this case, weight and volume advantages
have a limited impact, while the efficiency, reliability and functionalities are the
primordial requirements. The aforementioned functionalities, enabled by the
advanced software communication and control system, make the SST a smart
Overview and design of solid-state transformers 217
Conventional solution SST-based solution
MVAC MVAC
Services MVDC
MV side
3 3
Distribution system (smart grid)

Smart
LFT Modernization transformer
EV
(ST) station
Services
LV side 4
OLTC LV DC grid
LVAC
4 LVAC dc
dc dc
ac ac
load
ac ac ac ac ac
load
dc dc ac PV
G
PV dc G
load Functionalities
(voltage and current controllability)
Figure 9.2 Solid-state transformer in electric distribution grid application,

including the connection to dc grids
Efficiency Volume
Distribution (ST) Traction (SST)
Reliability Weight
Functionality Cost
Figure 9.3 Requirements and performance of the SST and ST for traction and
distribution grid applications
device, leading to the concept of smart transformer (ST). Then the ST can be
conceptualized as described below.
Smart transformer (ST):

Power electronics based system with a galvanic isolation in medium or high
frequency between the input and output with extra control functionalities,
obtained through advanced communication and control system. In addition
to the hardware advantage, its enhanced control makes the ST a powerful
technology able to provide ancillary services to the distribution grid, solving
most of the problems posed by the grid modernization. This system is a
remarkable technology able to fulfill most of the requirements for the
implementation of a smart grid, either ac or dc.
Figure 9.3 shows the main requirements and performance of the SST for both
described applications. While the volume and weight reduction are the most
important requirement for traction application, the functionalities, reliability and

efficiency play the most important roles in the electric distribution grid application.
9.3 ST architecture classification

The need to guarantee high efficiency and high reliability, offering new function-
alities, has a direct impact on the ST architecture selection. To assist a proper
structure selection, Figure 9.4 provides an overview of the possible architectures.
9.3.1 Power conversion stages

The most discussed classification is regarded the number of power conversion
stages, that can be single-stage, two-stages or three-stages. The single-stage is
normally implemented by a direct isolated matrix converter, and it is characterized
to have high power density, because of the absence of the dc-link capacitors.
However, the LV and MV are not decoupled, limiting the control capability.
The two-stage topology is usually based on the indirect matrix converter and an
inverter. This structure might present up to one constant dc-link, improving the
controllability in respect to the single-stage structure. However, the two-stage
structure offers limited functionalities. The three-stage is implemented by an MV
ac–dc converter, isolated dc–dc converter and an LV inverter. This topology has
normally two dc-links, in which at least one is available for connectivity. The
decoupling between MV and LV sides provides more degree of freedom for the
SST architecture
Power stages
1: MV ac-dc
1 Stage 2 Stages 3 Stages 2: Isolated dc-dc
3: LV dc-ac
Direct matrix converter Indirect matrix converter MV and LV decoupling
High power density Partial functionalities More functionalities
No extra functionalities
Scalability in
Modularity
voltage and power

Modular Semimodular Nonmodular
Fault tolerance
capability Cell level modularity
WBG semiconductors
LV semiconductors One nonreplaceable part
Simple control/communication
Availability Modularity
level
Cell level Converter level System level

of dc links
No dc link Single dc link Double dc link

LVDC available or LVDC and MVDC
MVDC available available
Figure 9.4 Classification of the ST architectures according to the number

of power conversions, modularity and modularity level
system control, allowing the implementation of functionalities. Hence, this is the

preferred architecture for the ST implementation.
9.3.2 Modularity
The next classification for the ST architecture concerns the degree of modularity.
It can be classified as modular, nonmodular or semimodular. The nonmodular system
is based on a single power converter, as exemplified in Figure 9.5(a), and usually
takes the advantage of high voltage wide-bandgap (WBG) semiconductors [4,5,8].
Special devices with high blocking voltage capabilities, such as 10 kV Silicon Car-
bide (SiC) MOSFETs or 15 kV SiC IGBTs [9–11] are needed to handle the MV level
in the power converter. Because these devices are not commercially available, but
only for research purposes, no currently available products are using this technology.
A standard approach to handle the MV in a nonmodular solution used in industrial
applications is to employ semiconductor connected in series; however, this demands
an additional control effort to balance the voltage over the semiconductors.
Conversely, the modular approach consists of several basic converters rated for
LV or low current, which are used as building blocks for the entire system, as
depicted in Figure 9.5(b). In this solution, the basic building block shares the voltage
dυ
dt
Nonmodular
MVDC LVDC
link link
Esw
(a)
Module
dυ
dt
LVDC
Modular
link
Esw
(b)
Figure 9.5 Possible implementation approaches of the ST architecture to handle

the MV level involved in the power conversion: (a) nonmodular and
(b) modular
on the MV side and the current in the LV among them, enabling the use of LV rated
devices. In spite of the fact that more components are used compared to the non-
modular solution, the modular approach is more economically advantageous, as
demonstrated in [12]. Additionally, modular architectures bring several advantages to
the power and voltage scalability, maintenance and the implementation of fault-
tolerant strategies. In comparison to the nonmodular, the modular architectures have
reduced electromagnetic interference (EMI) emissions [due to the low dv=dt and
di=dt, as illustrated in Figure 9.5(a) and (b)] and the possibility to use standard
LV rated devices that perform well, benefiting both efficiency and cost.
Some features from the modular and nonmodular concepts can be combined to
form the semimodular concept, as an alternative for both previous approaches. In
this case, the architecture is implemented using basic modules sharing the power,
similar to the modular approach, but a nonreplaceable component processing the
total system power is part of the architecture. It can also be called as a partially
modular approach, in which the modules on the primary side divide the power
among them, while the high frequency transformer (HFT) and the secondary side
bridge processes all the power.
9.3.3 Modularity level

The modular approach can be implemented in different level, as shown in Fig-
ure 9.6, and they are classified as cell level, converter level and system level.
Basic unit
Basic unit
Cell level
(a)
Basic unit Basic System level

unit System 2
System 1
Converter level
System level
(b) (c)
Figure 9.6 Modularization level: (a) cell level, (b) converter level and (c) system level
Cell level:
This is the most basic level, in which the basic unit is the cell used inside the
converter, as exemplified in Figure 9.6(a). Another example of a cell-level modular
system is the modular multilevel converter that is a single converter using several
cells as basic units.
Converter level:
The next level is the converter level. In this case, the power converter is the basic
unit, and they are combined in series and/or parallel to implement the entire system,
as shown in Figure 9.6(b). This is the most employed modularization level and the
input-series output-parallel connection is very popular. Interleaved converter is
another example of the converter level modular system, in which several converters
are connected in parallel.
System level:
Finally, the system level is presented in Figure 9.6(c). In this case, the full system
is used as the basic unit. It is often used in application where fault-tolerance or
scalability is highly desired.
All these levels can be freely combined, resulting in a modular system in
multiple levels. In modular ST architectures, the converter level is the most com-
mon one, although the cell level can also be used.
9.3.4 dc Link voltage availability

The conceptualization of the ST is an important task during its conceivement,
because it affects and constraints the further design phases. Besides the number
of stages and the modularity level, another very important decision must be
taken during the conceiving of the ST architecture: the availability of the dc
link. Recently, the dc microgrid technology has been developed significantly,
and many solutions for issues related to the grid stability and protection have
been proposed in literature. This sort of grid is usually integrated to the ac grid
through an LFT associated to a front-end rectifier for forming the required dc
voltage. Alternatively, the dc microgrid can be connected directly to the dc link
of the ST, so that its integration into the conventional ac grid happens through
the ST. Hence, the ST technology enables efficiently the integration of the dc
microgrid, and this feature play an important role during the conceptualization
of the ST architecture.
The modular three-stage architecture has enough degree of freedom to provide
one, two or even no connectivity to the dc links. It depends on the configuration
of the power converters, as depicted in Figure 9.7 for a simplified single-
phase structure. The availability of one dc link on the LV side (i.e., LVDC link) is
the state-of-the-art and the most adopted option, enabling the dc microgrid con-
nection [3–5].
The configurations without the LVDC link connectivity, as seen in Figure 9.7(a)
and (c), have distributed dc links on the LV side, instead of having them
MV MV MV MV
AC AC AC AC
LVDC
(a) (b)
MV MV MV MV
AC AC AC AC
LVDC
MVDC MVDC
(c) (d)
Figure 9.7 Different ST architecture configuration regarding the dc link

connectivity for a simplified single-phase structure: (a) without
dc link availability, (b) availability of the LVDC link,
(c) availability of the MVDC link, (d) availability of both
LVDC and MVDC links
connected. This strategy does not bring any advantages, and for this reason it has
not being adopted [3]. On the other hand, the distributed dc links on the MV side
provides advantages to the system, because the MV level involved in this side can
be distributed among the modules. For this reason, this concept has been inten-
sively used [3,4,12–16] and the most adopted structure is the one presented in
Figure 9.7(b).
Recently, the availability of the MVDC link has been discussed by the research
community, but there is no solution in this regard. The few amount of dc loads
applicable for MV level compared to the LV one (which is already well estab-
lished) makes the MVDC availability still controversial. Nevertheless, new dc
loads in LV level are emerging (e.g., electric fast charging station), supporting
research on the architectures with MVDC link. Consequently, the architecture with
MVDC link as shown in Figure 9.7(c) and (d) is not well adopted yet, but it is a
high potential structure and deserves attention.
In summary, the availability of the LVDC link is of paramount importance,
and then two configurations with respect to the dc link connectivity are highlighted:
the available LVDC link [Figure 9.7(b)] and the available LVDC and MVDC links
[Figure 9.7(d)].
9.4 Solid-state transformer and smart transformer

architectures overview
In the last two decades, many different SST and ST architectures have been pro-
posed and discussed in literature for traction and distribution grid applications. The
most relevant architectures used in both application are discussed.
In traction applications, the architectures are characterized to use commercia-
lized HV Si-IGBT, two power processing stages, as well as modular approach. The
efficiencies for each architecture for traction system are presented in Table 9.1. All
of them provide higher efficiency and power density, when compared to the con-
ventional solution using the LFT.
The three-stage architecture offers decoupling between the input and output,
besides the availability of at least one dc link. Consequently, it is the preferable
structure for grid applications, as seen in Table 9.1. To handle the MV level, two
approaches are commonly used: modular with commercial LV devices and non-
modular with customized HV devices. Regardless the adopted concept, most of
these architecture takes advantages of the high performance WBG SiC devices.
They offer lower conduction and switching losses compared to the standard devi-
ces, and they are very promising in ST application.
Figure 9.8(a) shows the efficiency values available in the literature according
to power of the system, as denoted in Table 9.1. For distribution system, very few
ST demonstrators with solid results have been presented in the literature so far. The
Table 9.1 Comparison of the ST and SST architectures from different research
groups/companies
Distribution application
Name Project/group Stages Concept DC link References

A FREEDM 3 Modular LV [13]
B FREEDM 3 Nonmodular MV/LV [11,17]
C FREEDM 3 Nonmodular MV/LV [9,10]
D ETH 3 Modular LV [18]
E Uniflex 3 Modular No [19–21]
F GE 2 Modular No [22]
Traction application
Name Company Stages Concept Efficiency (%) References
G ABB 1 Modular 94.5 [23,24]
H ABB 2 Modular 96.2 [14,25]
I Alstom 2 Partially modular 94 [26,27]
J Bombardier 2 Modular 96.2 [28]
K Siemens 2 Partially modular 97 [29,30]
98 25
Distribution F K Distribution
H D F
96 20
Isolation frequency (kHz)

E
Efficiency (%)
C B G
94
C
I 15
B Traction
92 Traction 10 J
90 5 A I
Experimental E
Theoretical
H K
88 A 0 G
101 102 103 101 102 103 104
(a) Power rating (kVA) (b) Power rating (kVA)
Figure 9.8 Performance comparison of the most relevant ST and SST

architectures developed by industries and universities (see Table 9.1):
(a) achieved efficiency compared to the power rating and (b) isolation
frequency as a function of the power rating
available data shows that the efficiency achieved by this kind of system is around
94.5%–97%. For traction systems, the ST efficiency is in the range of 94.5%–96%,
according to the available data provided experimentally (excluding the theoretical
one). Although the ST architectures in the distribution use more power stages
compared to those used in Traction, their performance is potentially higher,
because the utilized WBG semiconductors that plays a very important role in this
regards. Figure 9.8(b) shows the isolation frequency for each ST architecture
shown in Table 9.1, according to the power level. In traction, the isolation fre-
quency is normally below 10 kHz, whereas in distribution grid is mostly around
10–20 kHz. The limitation of switching frequency is ST for traction is explained by
the fact of HV Si-IGBT are normally used to save cost. It results in more switching
losses or switching frequency limitations.
9.5 dc–dc Stage: power converter topologies

The dc–dc stage is considered as the core element of the ST architecture, because it
is responsible to connect the MV side to the LV side providing galvanic isolation in
medium or high frequency among them. In consequence of its strict requirements
discussed next, this stage is the most challenging for the implementation [3,16,18,
31,32]. The high voltage and current levels involved on the power conversion
makes the dc–dc stage responsible for the majority of the system losses; hence, it
deserves more attention during the ST design. Moreover, this stage controls the
LVDC link, in which the dc microgrid is connected to. The requirements and a
review of the possible solutions for the dc–dc stage is discussed next.
9.5.1 Requirements
The requirements for the dc–dc stage of the ST, independently of the adopted
approach (modular or nonmodular), are summarized as follows:
● High voltage capability in the MV side: Handle voltage level of around 6–25 kV.
● High current in the LV side: Current level in the range of 100–2,000 A is
expected [3].
● High voltage isolation: The MF/HF transformer employed has an isolation
requirement defined by the own MVAC level, i.e., between 10 and 15 kV,
regardless of the solution adopted for the dc–dc stage.
● Decoupling degree between the MV dc side and the LV dc side: Ability to
provide total decoupling between the MVAC grid and the LVAC grid, so that
the MV stage and LV stage converters can operate independently from each
other. It means that the MV ac–dc stage can operate and provide the required
ancillary services to the MV grid (e.g., provide reactive power), without dis-
turbing or being disturbed by the LV dc–ac stage operation. The same remark
is also valid for the LV dc–ac stage operation. To achieve such decoupling
degree, both dc links (MV and LV) must be maintained constant with minimal
oscillation voltage (i.e., 5% of the LVDC [3]), and the dc–dc converter should
guarantee the voltage regulation.
● Control of the power flow: Ability to control the power flow between the
LVDC link and MVDC link and manage the connection of loads of both sides
[4,5,8].
● Bidirectionality: Many publications agree that the bidirectional power flow
capability is required for the dc–dc stage [3–5,8,33]. On the other hand, this
requirement is questionable, because the reverse power is an exceptional case
in the distribution system. If economically viable, unidirectional solution could
also be accepted. Despite that, bidirectionality is considered as a requirement
in modern ST.
● dc Breaker function: Overload and short-circuit protection (working as a
dc breaker) of the possible load/source/microgrid connected to the LVDC link)
[3–5,8,33].
● High efficiency: The ST is not only intended to replace the LFT of the dis-
tribution system but also solve problems coming from the grid modernization
[3]. However, high efficiency is expected for the ST, in order to compete with
the LFT. The highest efficiencies obtained from an SST in traction applications
are around 95%–97%, for two processing stages (i.e., ac–dc and dc–dc) [14].
Taking this fact into account, an expected efficiency of 96% is defined for
the whole ST system, as shown in Figure 9.9(b). The MV stage can achieve
an efficiency around 98.6%–99.3% [14,34], when multilevel converters are
used, while the LV stage can offer an efficiency in the range of 98.8%–99.2%
[35–37]. The dc–dc stage should provide an efficiency between 97.5% and
98.6%, so that the entire system achieves the desired efficiency, as shown in
Figure 9.9. Thus, an efficiency goal above 97.5% can be defined.
9.5.2 Basic module topologies

From the dc–dc converter requirements, the high efficiency is the most severe one.
The modular approach contributes a lot in terms of performance improvement of
the ST, because this approach allows the use of lower voltage rated devices, which
performs significantly better than those rated for higher voltage [3]. As described
MV stage dc–dc stage ST design challenges

Reliability Efficiency Cost
dc–dc module
High voltage HFT High current
3 ηMV η(dc–dc) ηLV 4
dc–dc module 98.6%–99.3% 98.8%–99.2%
[Ref MV] 97.5%–98.6% [Ref LV]
MVAC LVAC
dc–dc module ηSTdesired = 96%

(b)
dc–dc module
dc–dc module dc
(a) CHB (c)
Figure 9.9 (a) Equivalent single phase modular ST architecture adopting the
defined CHB, (b) expected efficiency for the whole system, as well as
for each power conversion stage ([Ref MV] ¼ [14,34], [Ref LV] ¼
[35–37]), (c) implementation of the basic module of the modular
dc–dc stage
Table 9.2 Specification range of the basic module

of the dc–dc stage
Power level 5–100 kW

Input voltage 0.6–1.7 kV
Output voltage 600–800 V
Isolation frequency 1–40 kHz
Desired efficiency >97.5%
before, this approach consists of several basic modules combined to build the entire
architecture, as shown in Figure 9.9(a). The basic modules are isolated dc–dc
converters, as depicted in Figure 9.9(c), and many topology options are available.
The topology choice of the basic module plays an important role in the ST design,
once the performance of the entire system depends on the efficiency of the indi-
vidual module. The specification of the dc–dc basic module is usually in the range
shown in Table 9.2. Considering this and the requirements described earlier, an
overview of the most suitable topologies for the basic modules is presented as
follows.
Among several options of the dc–dc converters, two families of converters are
pointed out as the most promising: the SRCs and the active bridges converters.
9.6 Series resonant converter

The series resonant converter (SRC) is the most adopted converter in ST,
considering the family of resonant converters, and its topology is presented in
s1 s3 s1,s s3,s
1:n
iL Cr Rp
leak
Lp
leak
Lsleak Rsleak
Ci Co
Vi Vo
υp Lm Rm υs
s2 s4 s2,s s4,s
Figure 9.10 Topology of the series-resonant dc–dc converter
Table 9.3 Main parameters used on the analysis of the

series resonant converter in Figure 9.10
Parameter Definition
Angular switching frequency ws ¼ 2pfs
Angular resonant frequency wo ¼ 2pfo
Relative angular frequency w ¼ ws =wps ffiffiffiffiffiffiffiffiffiffi
Resonant frequency fo ¼ 1=2p Lr Cr
Quality factor Q ¼ 2pfo Lr =Ra ¼ 1=2pfo Cr Ra
Figure 9.10. The SRC converter is composed by a full-bridge converter on the

primary and secondary sides, connected through a resonant tank circuit and a
transformer. The tank circuit is composed of a resonant capacitor (Cr ) connected in
series to the resonant inductor (Lr ). The SRC has different operation modes,
according to the load and to the relationship between the switching frequency (fs )
and resonance frequency (fo ). The resonance frequency depends only on the pair Lr
and Cr , and it is defined in Table 9.3. With respect to the load, the SRC can operate
in continuous conduction mode (CCM) or discontinuous conduction mode (DCM).
Regarding the switching frequency, it can operate below the resonant frequency,
equal to the resonant frequency and above the resonant frequency [38].
Independent of the operation mode, the SRC converter can generally be ana-
lyzed according to [38–40]. In those works, the input–output voltage relation is
derived, and it described in (9.1). Note that the variables presented in this equation
are described in Table 9.3.
Vo 1
ðjwÞ ¼ (9.1)
Vi 1 þ jQððws =wo Þ ðwo =ws ÞÞ
Figure 9.11(b) shows the normalized voltage gain (Vo =Vi ) of the SRC as a
function of the normalized operation frequency (fo =fs ), according to the quality
factor Q. According to this figure, the gain of the SRC varies with the frequency,
Vi
υp 1
Q=1
2/f s
Vi 0.8
fo<fs
Gain (Vo/Vi)
iLr fo=fs
0.6
γ Q=2
0.4
Vcpk
ΔυCr = 2Vcpk 0.2
Q=3
υCr Q=4
fs = fo
Q=5
0
Vcpk 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Normalized frequency ( f = fo/ fs)
(a) (b)
Figure 9.11 Series resonant converter: (a) main waveforms of the SRC and
(b) normalized voltage gain (Vo =Vi ) of the SRC as a function of the
normalized operation frequency (fo =fs ), according to the quality
factor Q
and usually the switching frequency is used to control the output voltage of
the SRC. However, this strategy implies a complex control system and poor
utilization of the HFT, once it must be designed to operate for the minimum
frequency. Thus, it is not suitable for ST applications, in which the power level
is relatively high.
The most efficient operating point of the SRC is when it operates in the
DCM with the switching frequency (fs ) equal or slightly below the resonance
frequency (fo ), i.e., with unity gain. In this operation mode, the primary side
switches achieve ZVS and the output diodes achieve zero-current-switching.
Moreover, the converter has good transformer utilization and also low EMI
emission due to the smooth current shape (low di=dt). In this operation point, the
power flow is not directly controlled, and the power conversion between input
and output is defined by (9.2), where n is the transformer turns ratio. On the
other hand, it is not a disadvantage in ST application, once the system has
enough degrees of freedom to control the input voltage by using the front-end
rectifier (first stage).
V o ¼ n Vi (9.2)
9.6.1 Current stress

To design the SRC converter, mapping the stresses on the semiconductors and
transformers are required. They are normally calculated using waveforms illu-
strated in Figure 9.11(a) and using the concept of average and rms value. Using this
approach, the mean value and rms value of the current in the semiconductors can be
calculated using the equations shown in Table 9.4. Likewise, the current stresses on
the transformer are calculated by the equations given in Table 9.5.
Table 9.4 Current stresses on the components S1b (MV side)

of the SRC [16]
Semiconductor AVG RMS

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffi ffi
MV side nPo =Vo nPo =8Vo 2p=ðfs Lr Cr Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffi ffi
LV side Po =Vo Po =8Vo 2p=ðfs Lr Cr Þ
Table 9.5 Current stresses on the transformer

of the SRC [16]
MV side LV side
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffi
nPo =4Vo p=ðfs Lr Cr Þ Po =4Vo p=ðfs Lr Cr Þ
9.6.2 Efficiency expectation

A high efficiency SRC is reported in [41] for high power and LV level applications
(5 kW, 600 V). In that paper, SiC MOSFETs are used, presenting a peak efficiency
of 97.6%, which is one of the highest efficiencies reported in the literature, con-
sidering the power/voltage level. Similarly, the SRC has been adopted in SST for
traction applications in [14,25], where an efficiency of 98% was reported for a
converter of 160 kW, employing special IGBTs devices.
Similarly, an optimum design of an SRC in ST application has been reported
in [16]. In this paper, the authors proposed a design methodology to improve the
efficiency and lifetime of the SRC, and the performance of the converter was
evaluated considering different semiconductors. The converter provides an effi-
ciency of 98.61%, when SiC-MOSFETs are used. Consequently, a very high effi-
ciency can be expected for this kind of converter using WBG devices.
The main drawback of the SRC for ST application is the lack of controllability.
The converter operates in open-loop, and it is not able to control the power flow.
Consequently, the control of the LVDC link for dc microgrid connectivity must be
shifted to other converter, when the SRC is used in the ST.
9.7 Dual active bridge (DAB) converter
The most popular converter of the active bridge family is the dual active bridge
(DAB) and its topology is presented in Figure 9.12. The DAB converter is com-
posed of two active bridges, connected through the inductance LDAB and a trans-
former. Even though the DAB topology is similar to the SRC, its operation is
totally different. Other topologies of the active bridge converter family, such as the
three-phase DAB and the multiple-active-bridge (MAB), can also be adopted as
a basic module of the ST. In fact, the MAB converter has been intensively
S1 S3 S1,s S3,s
Primary Secondary
bridge
LDAB bridge
iL
iL LDAB 1:n
Ci Co
Vi Vo
υp υs υp υs
S2 S4 S2,s S4,s
(a) (b)
Figure 9.12 Topology of the dual active bridge (DAB) dc–dc converter (a) and its
equivalent circuit (b)
s1 s1 s1
s3 s3 s3
S1s S1s S1s
S3s S3s S3s
υs υp
υp υp υp
υs υs υs j
iL iL iL
j D1 D2
D2
is1 is1 is1
υs1 υs1 υs1 D1
(a) (b) (c)
Figure 9.13 Main waveforms of the DAB converter considering different

modulation strategies: (a) phase-shift modulation (PSM),
(b) triangular current modulation (TCM), (c) trapezoidal
modulation (TPM)
investigated for ST applications like in [42], and good results have been presented.
However, the DAB converter is analyzed next for the sake of simplicity.
9.7.1 Modulation strategy

To modulate the DAB, many strategies have been presented [43–48], but the phase
shift modulation (PSM) is the most adopted one. The main waveforms of the DAB
using the PSM is depicted in Figure 9.13(a). In this scheme, rectangular voltages
vp and vs with 50% duty-cycles and phase shift of j are applied to the transformer
with a constant switching frequency fs . The power is controlled by the phase dif-
ference between the bridges. The PSM is characterized by ZVS turn-on of the
semiconductors, low rms current, symmetrical share of the losses in all switches,
high power transfer capability besides its simplicity. On the other hand, these
advantages are obtained depending on the input–output voltage ratio and also the
chosen operation point (nominal phase shift j).
When the input voltage and/or output voltage of the converter deviate sig-
nificantly from their nominal values (around 20%, according to [46,49]), the soft-
switching operation is normally lost. In this condition, the PSM is not advantageous
anymore [46]. To overcome this problem, alternative modulation strategies have
been proposed, like the triangular current mode (TCM) and the trapezoidal mod-
ulation (TPM), as shown in Figure 9.13(b) and (c), respectively. The TCM imposes
a triangular shape current on the transformer, while the TPM generates a trape-
zoidal current waveform on the transformer, as shown in (b) and (c), respectively.
In contrary to the PSM, the TCM and TPM control the duty-cycle for transferring
power between the input and output of the converter. These strategies can operate
with soft-switching independently from the input and/or output voltage deviation,
but with penalty on the peak value of the currents. To transfer the same amount of
power, the DAB converter presents higher peak values of currents, when TCM and
TPM are adopted in comparison with the PSM, because of the triangular or rec-
tangular shape of the current. It means that the rms values on the semiconductors
and transformers are also expected to be higher, implying on possible increase of
the conduction losses.
In ST application, the input and output voltage are typically constant without
large deviation of the nominal value. In this sense, the PSM is very advantageous,
and therefore the DAB converter is analyzed next using this strategy.
9.7.2 Analysis of the DAB using the PSM

For the theoretical analysis of the DAB converter, the equivalent circuit shown
Figure 9.12(b) can be used, where the active bridges are replaced by rectangular
voltage sources vp and vs . The waveforms of vp and vs are shown in Figure 9.14.
Analyzing the equivalent circuit, the inductor current is described by (9.3).
ð t0 þTs ð t0 þTs
1 1
iL ðtÞ ¼ iL ðt0 Þ þ vL ðtÞdt ¼ iL ðt0 Þ þ ðvp nvs Þdt (9.3)
L t0 L t0
The average power transferred can be defined by the following equation:

ð t0 þTs ð t0 þTs =2 ð t0 þTs =2
1 2 2Vi
P¼ pðtÞdt ¼ vp ðtÞiL ðtÞdt ¼ iL ðtÞdt (9.4)
Ts t0 Ts t0 Ts t0
Considering the current waveforms shown in Figure 9.14, the inductor current
can be described by (9.5). In this equation, the initial values are defined by (9.6)
and (9.7), respectively.
8 vp þ nvs
>
< iL ðto Þ þ L
> t; to < t < t1
DAB
iL ¼ (9.5)
>
> v þ nvs Ts
: iL ðt1 Þ þ p ðt t1 Þ t1 < t < t0 þ
LDAB 2
S1
S3
S1s
S3s
φ υs
υp
υp
υs
iL(to) iL(π)
iL
iL(0)
is1
υs1
Ts / 2
t0 t1 t2 t3 Ts
(π) (2π)
Figure 9.14 Voltage and current waveforms of the DAB converter operating with
PSM
ðVi nVo Þp 2nVo j

iL ðto Þ ¼ (9.6)
4pfs LDAB
ðnVo Vi Þp 2Vi j
iL ðt1 Þ ¼ (9.7)
4pfs LDAB
Note that iL ðto Þ ¼ iL ðpÞ in steady-state condition. Replacing (9.5) in (9.4)
and rearranging it, the equation of the power transference of the DAB is obtained,
as shown in (9.8) [49]. The input–output voltage relation of the DAB can be
defined as d ¼ nVo =Vi . The output power can be rewritten in terms of d, as shown
in the following equation:

Vi Vo jj j Vi2 jjj
P¼ j 1 ¼ dj 1 (9.8)
2pfs LDAB n p 2pfs LDAB p
The soft-switching is achieved when the devices are conducting at turn-off.
This means that the condition iL ð0Þ 0 and iL ðjÞ 0 must be satisfied. The
requirement of iL ðjÞ 0 ensures the turn-off under ZVS of the switches s1 and s4
(see Figure 9.14), while iL ðjÞ 0 guarantees the soft-switching of the devices s2
and s3 . Applying these limits in (9.6) and (9.7), the ZVS boundaries conduction
is obtained and described in (9.9). Accordingly, the area constrained by (9.9)
1.2
dup d = 1.5
Hard switching
1
d = 1.25
Power (P/Pmax) 0.8

ZVS d=1
0.6
d = 0.75
0.4 d = 0.5
0.2
ddown
0
0 π/6 π/4 π/3 π/2
Phase – shift (ϕ)
Figure 9.15 Output power of the DAB converter in function of phase-shift angle
j, considering several input–output voltage relation d
indicates the soft-switching region of the DAB converter, which is the desired
operation region to reduce losses.
8
> 3j
< dZVSðupÞ ¼ 1
2p (9.9)
> 1
: dZVSðdownÞ ¼
1 ð3j=2pÞ
Figure 9.15 shows the variation of the normalized power as a function of the
phase-shift angle for different values of d, in which the soft-switching boundaries
are highlighted. Note that the DAB converter operates in soft-switching for the
entire range of load if d ¼ 1, i.e., Vi ¼ nVo . Consequently, the DAB converter can
be properly designed, so that it operates with soft-switching for the whole load
range.
9.7.3 Current stresses on the DAB

To calculate the current stresses on the semiconductors and transformer, the peak
current on the LV side inductor (ILaðpkÞ ) and MV side inductor (ILbðpkÞ ) are required,
and those can be calculated by (9.10) and (9.11), respectively.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
VL VL2 8Leq fs nVL Io
ILPK ðLV Þ ¼ (9.10)
4Leq fs
ILaðpk Þ
ILPK ðMV Þ ¼ (9.11)
n
Considering the current waveforms presented in Figure 9.16 for positive power
flow, i.e., from MV to LV side, the current stresses on the semiconductors and
transformer are calculated and presented in Tables 9.6–9.8. Table 9.6 shows the
S1a S3a ILa(pk) ILb(pk)

ia ib
iLa iS1a uS1a S1b

S1b
Ci
Vi
ua
ZVS uS2a iS2a uS2b ZVS i
S2b
S2a S4a
Diode Channel Diode Channel

cond cond cond cond
(a) (b) (c)
Figure 9.16 Dual active bridge: (a) topology of the active bridge and the current
and voltage waveforms on the semiconductors of the (b) MV bridge
and (c) LV bridge
Table 9.6 Current stresses on the semiconductor S1b (MV side) of the DAB
converter
Component AVG RMS

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Channel (S1b ) ILPK ðLV Þ =6nð1 ð3j=4pÞÞ ILPK ðLV Þ =3n 1=2ð1 ð5j=12pÞÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Body diode (iD1b ) ðILPK ðLV Þ =3nÞðj=8pÞ iD1bðrmsÞ ¼ ðILPK ðLV Þ =3nÞ ðj=12pÞ
Table 9.7 Current stresses on the semiconductor S1a (LV side) of the DAB
converter
Component AVG RMS

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Channel (S1a ) ðILPK ðLV Þ =6nÞð1 ð3j=4pÞÞ ðILPK ðLV Þ =3nÞ 1=2ð1 ð5j=12pÞÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Body diode (iD1a ) ðILPK ðLV Þ =3nÞðj=8pÞ iD1bðrmsÞ ¼ ðILPK ðLV Þ =3nÞ ðj=12pÞ
Table 9.8 Current stresses on the transformer of the

DAB converter
MV side i Lb MV side iLa

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðILPK ðLV Þ =3nÞ 1 ð2j=3pÞ ILPK ðLV Þ 1 ð2j=3pÞ
average and rms currents on the channel and body diode of the semiconductor S1 a,
while Table 9.6 shows the current effort on the semiconductor of the LV side S1 b of
the DAB converter. The current stresses on the inductor regardless it is connected
to the LV or MV side are presented in Table 9.8.
As observed in these equations, the current stresses of all components are
somehow proportional to defining j. Furthermore, when the power flows from
the MV to the LV, most of the current flows through the channel of the MV side
semiconductors, whereas it flows mostly through the body diode on the LV side.
Therefore, it is very important to select a power semiconductor for the LV side with
low forward drop voltage for the body diode.
9.7.4 Efficiency calculation

High performance is expected for the DAB converter, because of its soft-switching
operation, but it depends on the design and the employed semiconductors. For
instance, if a large nominal phase-shift is selected, i.e., larger than 60 , the con-
verter operates with high reactive power, deteriorating the efficiency. In this case,
even operating under soft-switching, the converter might have low performance.
Thus, the design and selected components are of paramount importance.
In [48], a 4 kW DAB prototype based on 1.2 kV Silicon (Si) IGBT has
achieved an efficiency of 97.1%, while an efficiency of 97.5% was reported in [50],
for a 6.25 kW prototype based on silicon carbide (SiC) MOSFETs. Similarly, an
optimized design of the DAB is presented in [51], where 1.2 kV SiC MOSFETs
were employed, and an efficiency of 98.2% was obtained.
The quadruple active bridge (QAB) [52], another topology from the active
bridge converters family, has been used as modules for the ST architecture in [42].
Differently from the DAB, in which only two bridges are connected to the HFT, the
QAB converter has four bridges connected to the HFT, as it is illustrated in Fig-
ure 9.17. This converter operates similarly to the QAB when employed to the ST
application. Hence, a similar efficiency is expected. In fact, the QAB is more
advantageous than the DAB in ST application according to [12], because the QAB
can perform similarly to the DAB, but offering a significant cost reduction. The
cost reduction is obtained due to the fact that less bridges are used on the LV side
(see Figure 9.17), requiring less auxiliary power supplies and gate drivers, which
are costly.
In [12,42], an optimum design of the QAB converter is reported for optimizing
the efficiency and cost. The converter has presented an efficiency of 97.5% when
SiC-MOSFETs are used, which is considerably high. Although the SRC tends to
present higher efficiency than the QAB converter, this last converter presents two
advantages over the SRC: lower cost and high-control capability. This last feature
is of paramount importance, because the QAB converter is able to control the
LVDC link and enabling the connection of the dc microgrid directly to the LVDC
without an extra converter. It is highly desired in ST applications, and then the
QAB converter is highlighted to be a very advantageous choice for ST
applications.
s1b s3b
iLb Lb 1 : n
Ci
Vb
υb
s2b s4b
s1a s3a
s1c s3c
iLc Lc La
Ci Co
Vc Va
υc υa
s2c s4c
s2a s4a
s1d s3d
iLd Ld
Ci
Vd
υd
s2d s4d
Figure 9.17 The topology of the quadruple active bridge (QAB)
DC link control
d Phase
Vdc Σ PI shift
modulator
V*dc
Figure 9.18 General form of a dc voltage control for a DAB
9.8 Active-bridge converter: control description and tuning
The objective of the control is to ensure a stable bus regulation, with a good
immunity from the load variations. Figure 9.18 shows a simplified diagram of the
control. The normalized phase shift d ¼ ðj=2pÞ is the output of a PI regulator that
controls the dc voltage.
Considering the equation that regulates the power transfer (9.12), the current
that flows into the output port can be obtained by Idc ¼ Pdc =Vdc .
nVin Vdc
Pdc ¼ d ð1 2jdjÞ (9.12)
fsw Llk
KDAB Cdc Iload

d
1 + 1.5s/fsw
Figure 9.19 Simplified model for the voltage control of the dc bus
By linearizing the relationship, (13) can be written.

nVin
IDC ¼ ð1 4d Þjd¼d0 ;Vin ¼Vin0 ðd d0 Þ
fsw Llk
nVin
þ d ð1 2jdjÞjd¼d0 ;Vin ¼Vin0 ðVin Vin0 Þ (9.13)
fsw Llk
As can be seen, the small-signal gain depends on the linearization point and is
also dependent on the input voltage. The common simplification is to consider the
input voltage constant in a switching cycle, neglecting the second part of (9.13).
Moreover, since the design of the DAB is normally realized for small phase shift to
limit the reactive power, the linearization around the point d0 ¼ 0 is performed. It
is important to note that this also constitutes the worst case scenario for the stability
point of view (maximum gain).
In order to study the control, a simplified model is shown in Figure 9.19, where
KDAB ¼ 4ðVin =fsw Llk Þ. In order to take into account the switching frequency and
the discretized control, a first-order transfer function with a time constant of
Tc ¼ 1:5=fsw is added.
Neglecting the load current for the controller tuning, the system presents a
pole in the origin due to the capacitor. The tuning of the PI can be performed with
the symmetrical optimum criterion that aims at maximizing the phase margin. The
target is the crossover frequency of the open-loop transfer function fcut . The PI
regulator is chosen so that the ratio between the target crossover frequency and its
corner frequency (1ti ) is the same as the ratio between the high-frequency pole of
the plant and the target crossover frequency R ¼ fcut ti ¼ 1=fcut Tc .
In this way,
Cdc
Kp ¼ 2pfcut pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (9.14)
KDAB 1 þ ð1=R2 Þ
Kp
Ki ¼ (9.15)
ti
The control realized with a simple PI regulator has a main drawback: the
higher the phase margin of the design, the higher the output impedance becomes.
The smaller the output impedance ZðsÞ ¼ ðVdc ðsÞ=Iload ðsÞÞ, the more immune the
control system is to load variation. As a consequence, fast-changing load will
deteriorate the voltage regulation unless a very fast controller is realized.
Open-loop transfer function

200
Magnitude (dB) 300 Hz
600 Hz
100
1,800 Hz
–100
–90
Phase (°)
–135
–180
10–1 10 0 101 10 2 10 3 10 4 10 5
Frequency (Hz)
Output impedance
50
Magnitude (dB)
–50
–100
–150
90
45
Phase (°)
–45
–90
10–1 100 101 10 2 10 3 10 4 10 5
Frequency (Hz)
Figure 9.20 Frequency response of the voltage control and of the output
impedance
Figure 9.20 shows the frequency responses of the voltage control and of the
output impedance with the voltage control. As can be seen, the crossover of the
open-loop transfer function always happens at the maximum of the phase. Of
course, choosing a higher crossover frequency reduces the phase margin.
9.9 Summary
The SST has several advantages when applied in the modern electric distribution
grid. Among them, the availability of the dc link for connection of dc microgrid is
mentioned in this chapter. There are many different possibilities to implement the
architecture of the SST, and an overview and classification have been presented.
For electric distribution, the three power processing stages architecture is cited to
be the best choice, because it provides the decoupling between the MV side and LV
side and also providing dc link connectivity. Furthermore, the dc–dc stage plays an
important role on the SST architecture design, because it is responsible for the
major losses of the system, besides to be in charge for controlling the LVDC link.
Therefore, the SRC and the DAB have been pointed out as the most promising
choice. The SRC can provide a very high efficiency, but it is not able to control
the power flow and the regulate the LVDC link properly. The DAB converter, on
the other hand, can provide a high efficiency solution and provide full control. The
design aspects of the power stage and control stage of the DAB converter focusing
on the SST application has also been discussed in this chapter.
References
[1] E. McMurray, ‘‘Power converter circuits having a high frequency link,’’

1968, U.S. Patent US3517300.
[2] D. Dujic, F. Kieferndorf, F. Canales, and U. Drofenik, ‘‘Power electronic
traction transformer technology,’’ in Proceedings of The 7th International
Power Electronics and Motion Control Conference, vol. 1, Jun 2012,
pp. 636–642.
[3] M. Liserre, G. Buticchi, M. Andresen, G. D. Carne, L. F. Costa, and Z. X.
Zou, ‘‘The smart transformer: impact on the electric grid and technology
challenges,’’ IEEE Industrial Electronics Magazine, vol. 10, no. 2,
pp. 46–58, Summer 2016.
[4] X. She, A. Q. Huang, and R. Burgos, ‘‘Review of solid-state transformer
technologies and their application in power distribution systems,’’ IEEE
Journal of Emerging and Selected Topics in Power Electronics, vol. 1, no. 3,
pp. 186–198, Sep 2013.
[5] A. Q. Huang, M. L. Crow, G. T. Heydt, J. P. Zheng, and S. J. Dale, ‘‘The
future renewable electric energy delivery and management (FREEDM)
system: the energy internet,’’ Proceedings of the IEEE, vol. 99, no. 1,
pp. 133–148, Jan 2011.
[6] A. Q. Huang, ‘‘Medium-voltage solid-state transformer: technology for a
smarter and resilient grid,’’ IEEE Industrial Electronics Magazine, vol. 10,
no. 3, pp. 29–42, Sep 2016.
[7] B. Olek and M. Wierzbowski, ‘‘Local energy balancing and ancillary ser-
vices in low-voltage networks with distributed generation, energy storage,
and active loads,’’ IEEE Transactions on Industrial Electronics, vol. 62,
no. 4, pp. 2499–2508, Apr 2015.
[8] J. Wang, A. Q. Huang, W. Sung, Y. Liu, and B. J. Baliga, ‘‘Smart
grid technologies,’’ IEEE Industrial Electronics Magazine, vol. 3, no. 2,
pp. 16–23, Jun 2009.
[9] K. Mainali, A. Tripathi, S. Madhusoodhanan, et al., ‘‘A transformerless

intelligent power substation: a three-phase SST enabled by a 15-kV SiC
IGBT,’’ IEEE Power Electron Magazine, vol. 2, no. 3, pp. 31–43, Sep 2015.
[10] S. Madhusoodhanan, A. Tripathi, D. Patel, et al., ‘‘Solid-state transformer
and MV grid tie applications enabled by 15 kV SiC IGBTs and 10 kV SiC
MOSFETs based multilevel converters,’’ IEEE Transactions on Industry
Applications, vol. 51, no. 4, pp. 3343–3360, Jul 2015.
[11] F. Wang, G. Wang, A. Huang, W. Yu, and X. Ni, ‘‘Design and operation
of a 3.6 kV high performance solid state transformer based on 13 kV SiC
MOSFET and JBS diode,’’ in 2014 IEEE Energy Conversion Congress and
Exposition (ECCE), Sep 2014, pp. 4553–4560.
[12] L. F. Costa, F. Hoffmann, G. Buticchi, and M. Liserre, ‘‘Comparative
analysis of MAB dc–dc converters configurations in modular smart trans-
former,’’ in 2017 IEEE 8th International Symposium on Power Electronics
for Distributed Generation Systems (PEDG), Apr 2017, pp. 1–8.
[13] X. She, X. Yu, F. Wang, and A. Q. Huang, ‘‘Design and demonstration of a
3.6-kV 120-V/10-kVA solid-state transformer for smart grid application,’’
IEEE Transactions on Power Electronics, vol. 29, no. 8, pp. 3982–3996,
Aug 2014.
[14] C. Zhao, D. Dujic, A. Mester, et al., ‘‘Power electronic traction transformer:
medium voltage prototype,’’ IEEE Transactions on Industrial Electronics,
vol. 61, no. 7, pp. 3257–3268, Jul 2014.
[15] M. Liserre, M. Andresen, L. Costa, and G. Buticchi, ‘‘Power routing in mod-
ular smart transformers: active thermal control through uneven loading of
cells,’’ IEEE Power Electronics Magazine, vol. 10, no. 3, pp. 43–53, Sept 2016.
[16] L. F. Costa, G. Buticchi, and M. Liserre, ‘‘Highly efficient and reliable
SiC-based dc–dc converter for smart transformer,’’ IEEE Transactions on
Industrial Electronics, vol. 64, no. 10, pp. 8383–8392, Oct, 2017.
[17] F. Wang, G. Yao, A. Huang, W. Song, and X. Ni, ‘‘A 3.6 kV high perfor-
mance solid state transformer based on 13 kV SiC MOSFET,’’ in 2014 IEEE
5th International Symposium on Power Electronics for Distributed
Generation Systems (PEDG), Galway, Jun 2014, pp. 1–8.
[18] J. W. Kolar, and J. Huber, ‘‘Solid-state transformers (SST) concepts, chal-
lenges and opportunities,’’ in ECPE Workshop on Smart Transformers for
Traction and Future Grid Application, Zurich, Switzerland, Feb 2016.
[19] S. Bifaretti, P. Zanchetta, A. Watson, L. Tarisciotti, and J. C. Clare,
‘‘Advanced power electronic conversion and control system for universal
and flexible power management,’’ IEEE Transactions on Smart Grid, vol. 2,
no. 2, pp. 231–243, Jun 2011.
[20] J. Clare, ‘‘Advanced power converters for universal and flexible power
management in future electricity networks,’’ in 2009 13th European Con-
ference on Power Electronics and Applications, Sep 2009, pp. 1–29.
[21] H. Q. S. Dang, A. Watson, J. Clare, et al., ‘‘Advanced integration of multi-
level converters into power system,’’ in 2008 34th Annual Conference of
IEEE Industrial Electronics, Nov 2008, pp. 3188–3194.
[22] M. K. Das, C. Capell, D. E. Grider, et al., ‘‘10 kV, 120 a SiC half H-bridge
power MOSFET modules suitable for high frequency, medium voltage
applications,’’ in 2011 IEEE Energy Conversion Congress and Exposition,
Sep 2011, pp. 2689–2692.
[23] L. Meysenc, P. Stefanutti, P. Noisette, N. Hugo, and A. Akdag, ‘‘Multilevel
ac/dc converter for traction application,’’ 2009, U.S. Patent US7558087.
[24] N. Hugo, P. Stefanutti, M. Pellerin, and A. Akdag, ‘‘Power electronics
traction transformer,’’ in 2007 European Conference on Power Electronics
and Applications, Sep 2007, pp. 1–10.
[25] D. Dujic, C. Zhao, A. Mester, et al., ‘‘Power electronic traction transformer-
low voltage prototype,’’ IEEE Transactions on Power Electronics, vol. 28,
no. 12, pp. 5522–5534, Dec 2013.
[26] J. Taufiq, ‘‘Power electronics technologies for railway vehicles,’’ in 2007
Power Conversion Conference – Nagoya, Apr 2007, pp. 1388–1393.
[27] J. Martin, P. Ladoux, B. Chauchat, J. Casarin, and S. Nicolau, ‘‘Medium fre-
quency transformer for railway traction: soft switching converter with high
voltage semi-conductors,’’ in 2008 International Symposium on Power Elec-
tronics, Electrical Drives, Automation and Motion, Jun 2008, pp. 1180–1185.
[28] M. Steiner and H. Reinold, ‘‘Medium frequency topology in railway
applications,’’ in 2007 European Conference on Power Electronics and
Applications, Sep 2007, pp. 1–10.
[29] M. Glinka and R. Marquardt, ‘‘A new ac/ac multilevel converter family,’’
IEEE Transactions on Industrial Electronics, vol. 52, no. 3, pp. 662–669,
Jun 2005.
[30] M. Glinka, ‘‘Prototype of multiphase modular-multilevel-converter with
2 MW power rating and 17-level-output-voltage,’’ in 2004 IEEE 35th
Annual Power Electronics Specialists Conference (IEEE Cat. No.
04CH37551), vol. 4, 2004, pp. 2572–2576.
[31] M. Liserre, M. Andresen, L. Costa, and G. Buticchi, ‘‘Power routing in
modular smart transformers: active thermal control through uneven loading
of cells,’’ IEEE Industrial Electronics Magazine, vol. 10, no. 3, pp. 43–53,
Fall 2016.
[32] L. F. Costa, G. Buticchi, and M. Liserre, ‘‘Quad-active-bridge as cross-link
for medium voltage modular inverters,’’ in IEEE Energy Conversion Con-
gress and Exposition (ECCE), Sep 2015, pp. 645–652.
[33] J. E. Huber and J. W. Kolar, ‘‘Solid-state transformers: on the origins and
evolution of key concepts,’’ IEEE Industrial Electronics Magazine, vol. 10,
no. 3, pp. 19–28, Sep 2016.
[34] L. Schrittwieser, M. Leibl, M. Haider, F. Thöny, J. W. Kolar, and T. B.
Soeiro, ‘‘99.3% Efficient three-phase buck-type all-SiC SWISS rectifier
for dc distribution systems,’’ in 2017 IEEE Applied Power Electronics
Conference and Exposition (APEC), March 2017, pp. 2173–2178.
[35] B. Benkendorff, F. W. Fuchs, and M. Liserre, ‘‘Simulated and measured
efficiency verification power circulation method of a high power low voltage
NPC converter for wind turbines,’’ in 2016 18th European Conference

on Power Electronics and Applications (EPE’16 ECCE Europe), Sep 2016,
pp. 1–10.
[36] Y. Shi, R. Xie, L. Wang, Y. Shi, and H. Li, ‘‘Switching characterization and
short-circuit protection of 1200 V SiC MOSFET T-type module in PV
inverter application,’’ IEEE Transactions on Industrial Electronics, vol. 64,
no. 11, pp. 9135–9143, Nov 2017.
[37] J. Colmenares, D. Peftitsis, J. Rabkowski, D. P. Sadik, G. Tolstoy, and H. P.
Nee, ‘‘High-efficiency 312-kVA three-phase inverter using parallel con-
nection of silicon carbide MOSFET power modules,’’ IEEE Transactions on
Industry Applications, vol. 51, no. 6, pp. 4664–4676, Nov 2015.
[38] K. Afridi, ‘‘Resonant and soft-switching techniques in power electronics,’’
Department of Electrical, Computer and Energy, Colorado University,
Colorado, USA, Lectures Note, 2014.
[39] R. L. Steigerwald, ‘‘A comparison of half-bridge resonant converter topol-
ogies,’’ IEEE Transactions on Power Electronics, vol. 3, no. 2, pp. 174–182,
Apr 1988.
[40] P. K. Jain, ‘‘Power electronic course: resonant dc/dc converters,’’ Depart-
ment of Electrical and Computer Engineering, Queen’s University, Canada,
Ontario, Lectures Note, 2016.
[41] S. Inoue and H. Akagi, ‘‘A bidirectional isolated dc/dc converter as a core
circuit of the next-generation medium-voltage power conversion system,’’
Mar 2007.
[42] L. F. Costa, G. Buticchi, and M. Liserre, ‘‘Quadruple active bridge dc–dc
converter as the basic cell of a modular smart transformer,’’ in 2016 IEEE
Applied Power Electronics Conference and Exposition (APEC), Mar 2016,
pp. 2449–2456.
[43] D. J. Costinett, ‘‘Analysis and design of high efficiency, high conversion
ratio, dc–dc power converters,’’ Ph.D. dissertation, University of Colorado at
Boulder, 2010.
[44] S. Han, ‘‘High-power bi-directional dc/dc converters with controlled device
stresses,’’ Ph.D. dissertation, Georgia Institute of Technology, 2012.
[45] H. van Hoek, ‘‘Design and operation considerations of three-phase dual
active bridge converters for low-power applications with wide voltage
ranges,’’ Ph.D. dissertation, RWTH Aachen University, 2017.
[46] N. Schibli, ‘‘Symmetrical multilevel converters with two quadrant dc–dc
feeding,’’ EPFL, Lausanne, 2000.
[47] F. Krismer and J. W. Kolar, ‘‘Closed form solution for minimum conduction
loss modulation of DAB converters,’’ IEEE Transactions on Power
Electronics, vol. 27, no. 1, pp. 174–188, Jan 2012.
[48] T. Todorčević, R. van Kessel, P. Bauer, and J. A. Ferreira, ‘‘A modulation
strategy for wide voltage output in DAB-based dc–dc modular multilevel
converter for DEAP wave energy conversion,’’ IEEE Journal of Emerging
and Selected Topics in Power Electronics, vol. 3, no. 4, pp. 1171–1181,

Dec 2015.
[49] R. W. A. A. D. Doncker, D. M. Divan, and M. H. Kheraluwala, ‘‘A three-
phase soft-switched high-power-density dc/dc converter for high-power
applications,’’ IEEE Transactions on Industry Applications, vol. 27, no. 1,
pp. 63–73, Jan 1991.
[50] C. Gammeter, F. Krismer, and J. W. Kolar, ‘‘Comprehensive con-
ceptualization, design, and experimental verification of a weight-optimized
all-SiC 2 kV/700 V DAB for an airborne wind turbine,’’ IEEE Journal
of Emerging and Selected Topics in Power Electronics, vol. 4, no. 2,
pp. 638–656, Jun 2016.
[51] R. M. Burkart and J. W. Kolar, ‘‘Comparative eta–rho–sigma pareto opti-
mization of Si and SiC multilevel dual-active-bridge topologies with wide
input voltage range,’’ IEEE Transactions on Power Electronics, vol. 32,
no. 7, pp. 5258–5270, Jul 2017.
[52] S. Falcones, R. Ayyanar, and X. Mao, ‘‘A dc–dc multiport-converter-
based solid-state transformer integrating distributed generation and storage,’’
May 2013.
Chapter 10
Bipolar-type DC microgrids for high-quality
power distribution
Sebastian Rivera1,2, Ricardo Lizana3,
Samir Kouro1 and Bin Wu4
10.1 Introduction
The implementation of active distribution networks aims to modernize and improve
the sustainability of power systems, in order to include distributed energy genera-
tion resources and also to address the growing presence of power electronics in
different applications. This will be done through the integration of communication
and control technologies for effective control and monitoring of the system, in the
form of active and flexible DC distributed networks. The deployment of these
active distribution networks improves the performance of the traditional electric
system in terms of cost, efficiency, space and reliability. Different factors moti-
vated studies and development of a wide variety of DC distribution systems during
the last years [1], and led to a regained popularity in DC energy systems. Some of
these reasons are the growing presence of DC-based consumer equipment, the
significant increase in the installation of renewable energy conversion systems and
their detrimental effects on the power quality in conventional AC grids.
Moreover, the successful introduction of transmission of electrical power using
high-voltage DC (HVDC) systems has further validated their enhanced efficiency,
higher power capacity and controllability when compared to their AC counterparts,
especially in long distance and underground or submarine transmission systems.
These features allowed important developments in HVDC transmission config-
urations over the last years [2–6]. Consequently, these improvements later extended
to distribution level, enabled by advances in power electronics, the relative
maturity of DC-based consumer appliances and their subsequent widespread
growth [7,8]. Currently, DC-based equipment is increasing its penetration in low-
voltage kilowatt-scale applications in many sectors: aerospace, automotive, data
1
Department of Electronic Engineering, Universidad Tecnica Federico Santa Maria, Chile
2
Faculty of Engineering and Applied Sciences, Universidad de los Andes, Chile
3
Faculty of Engineering, Universidad Catolica de la Santisima Concepcion, Chile
4
Department of Electrical and Computer Engineering, Ryerson University, Canada
centers, high-efficiency households or marine, to name a few. It is expected that

DC grids will become key players in the large-scale deployment of distributed
generators (DGs), allowing increased power capacity, enhanced power quality and
resilience against power surges, disturbances and asymmetrical loads [7]. This is
also supported by the increasing number of both loads and generation units inter-
faced by power electronics [8].
In general terms, DC grids represent a much more compact, reliable and sim-
pler approach for power distribution. By minimizing the use of transformers, the
size and volume of components of an equivalent DC system are reduced, while they
are capable of transferring higher power levels using the same infrastructure.
Moreover, when power is delivered at DC it does not exhibit issues with synchro-
nization nor reactive power, and only the regulation of the voltage is required.
Additionally, efficiency is enhanced given the DC nature of the majority of appli-
ances, storage elements and DGs, through the elimination of additional power
converter stages, required to integrate them to AC grids. In other words, their
integration to a DC network is done with a single power conversion stage. More
importantly, their capacity of delivering a continuous supply of high-quality power
regardless the conditions in the utility system [9,10] facilitates the integration of
growing technologies such as renewable generation, electric vehicle charging
infrastructure and energy storage solutions.
Nevertheless, several challenges need to be addressed before shifting to DC
systems. The main barriers are the considerably challenging design of the protective
devices and the lack of standardization, despite several organizations are making
efforts promoting and developing standardized practices. The existence of regulation
will allow the proper preparation of personnel and equipment, and eventually will
create a market for DC networks supporting their proliferation [7,11].
Given the growing adoption of DC distribution architectures, important efforts
are being made toward standardization and masiffication of this emerging tech-
nology, proposing guidelines regarding DC voltage levels, installation design or
maintenance practices to name a few [11]. For example, the EU Low Voltage
Directive (LVD 2014/35/EU) defines the boundaries for voltages used in LVDC
systems [12]. This directive defines guidelines and requirements for the electrical
equipment with voltage ratings within 75 V and 1,500 V to be commercialized and
also guarantees the free commercialization of conforming equipment. Additionally,
EMerge Alliance, the International Electrotechnical Commission and the Institute
of Electrical and Electronics Engineers are other important organizations engaged
in developing voltage and performance standards for the wide-scale adoption of DC
distribution systems.
As stated earlier in this book, there are many ways for implementing a DC
distribution system [11]. However, two of them are the dominant methods: using a
unipolar or monopolar architecture, where two conductors are employed for gen-
erating a single voltage level; and a bipolar one, that resembles the three-phase AC
system and uses three wires to generate two different voltage levels [8,13]. The first
option features a simpler structure, with a reduced supply reliability, while the
latter one represents a more reliable and flexible alternative, given the possibility of
Bipolar-type DC microgrids for high-quality power distribution 247
power transfer when a line fails or that allows the connection of the loads to two
regulated voltages [13]. In addition, the converters in bipolar systems can either
operate with relatively low-rated voltage, thereby securing converter cost compe-
titiveness, or be connected to higher AC voltages while maintaining the step-down
effort of the DC–DC stages. Efficiency is also enhanced in three-wire systems, as
the currents are reduced for a system with the same power rating [14].
On the other hand, grounding of DC systems is a complex issue and its design
involves several factors [11]. One of the features of a bipolar network is that it has
higher safety levels since the grounding connection is easier and simpler using the
neutral conductor [1]. Unlike unipolar systems, the lack of uncertainty in the pole-
to-ground voltages allows to clear faults easily and quickly [1,8,15]. However, if
bipolar systems present differences in the generation or power consumption of the
loads connected to each pole, voltage imbalance can occur, which can adversely
affect the receiving voltage end or the loads connected to it, if not addressed
properly [8,16,17]. The current circulating through the neutral conductor affects the
voltage balance of the system and also its efficiency, as during such scenarios the
losses in the feeder are increased [7].
For these reasons, balance control becomes essential in bipolar systems [10], as
there is no way to guarantee the loads will be identical in both of the buses of the
distributed system. Aiming in this direction, different balancing methods have been
proposed in the literature, which will be discussed in detail in the upcoming sec-
tions. Considering the tremendous potential that bipolar DC microgrids hold for the
development of DC distribution systems, the remainder of this chapter will provide
an overall view in the efforts being made in this architecture and the current
technologies existing in the literature.
10.2 Bipolar-type DC distribution systems
The bipolar architecture for DC distribution networks offers interesting features over
the conventional unipolar counterpart. This structure is exhibited in Figure 10.1, where
it can be seen that the utility AC voltage is converted to DC with the use of a dis-
tribution transformer and an active rectifier, also called distribution converter. Then, at
DC level the system adopts a three-wire system that consists of the positive conductor
(p), the negative conductor (n) and the neutral conductor (z). As it has been demon-
strated at transmission level for HVDC, and despite being a more complex technical
solution, the aforementioned configuration presents clear advantages in terms of effi-
ciency, reliability, safety and transmission capacity when compared to conventional
two-wire systems. Advances in power electronics and their decreasing costs have
allowed an increased penetration of power converters in different applications, thus
enabling the consequent expansion of DC systems at distribution level [1].
Overall, the system exhibited in Figure 10.1 features several benefits. First, the
three-wire configuration resembles the traditional AC system in the sense that it
provides the connection to two different regulated voltages. The voltage between
the positive and negative poles is analogous to the line-to-line voltage, while the
Wind
PV
LED lighting
Heating
DC AC DC
DC DC DC
p
MVAC Distribution n
transformer
DC DC DC DC
Distribution DC AC DC DC
converter
EV charger AC drive Fuel cell Battery
Figure 10.1 Concept of low-voltage bipolar-type DC network
voltage of a pole with respect to the neutral connector is analogous to the phase-to-
neutral one [8]. This allows to accommodate a wide set of DGs and loads with
different voltage and power ratings combinations in a single DC network [18–20].
From an economical point of view, bipolar systems secure the cost competi-
tiveness of the converters interfacing the different stages of the network. This is
related with the reduction of the voltage ratings of the power electronic compo-
nents, the efficiency improvement due the reduction of the rated current, besides
the elimination of unnecessary DC–AC stages [14]. The latter is because most of
the stages involved in DGs are either DC-based, e.g., photovoltaic (PV) panels, fuel
cells and batteries or generate outputs with variable voltage/frequency, e.g., wind,
small hydro, wave/tidal energy conversion systems, hence require power electro-
nics devices to accommodate their output to network conditions [14]. In addition,
the presence of batteries and storage stages further enhances the benefits of DC
distribution, as it yields to greater efficiency improvements [21,22]. The result is a
substantial enhancement in the quality of the electric system and at the same time,
reduction of the costs when compared to conventional AC solutions [13].
The aforementioned resemblance with three-phase AC systems is also bene-
ficial for the migration process to bipolar DC networks. Considering a conventional
grounded three-phase system, which requires five conductors for its realization
(three-phase conductors, one neutral and the protective earth or grounding con-
ductor). These installed cables can serve in a retrofitted DC system that will be
superior in terms of power ratings capability and efficiency than the original one
[14]. Typically, low-voltage power cables can be found in different sizes and
ratings. As an example, multi-wire 230 V AC installations use 300 V/500 V 10 A
cables, following the structure in Figure 10.2(a). The system is then rated for
6.21 kW (assuming a power factor of 0.9). If such system would be replaced by
(a) a 325 V DC network or (b) a 325 V DC network, using the configuration
S p p
R T PE n PE p z PE z
N n n
(a) (b) (c)
Figure 10.2 Use of cables in retrofitted systems or new DC systems: (a) AC cable;
(b) unipolar DC cable; and (c) bipolar DC cable
proposed in Figure 10.2(b) and (c), the power transfer capability will be enhanced
to 6.5 kW in both cases, with no considerations for reactive power and using the
same cables. However, bipolar DC system cables only have to withstand 10 A while
the unipolar one should be rated for 20 A. This illustrates some of the benefits of
implementing a bipolar DC microgrid from an existing AC installation.
10.3 Topologies and operational aspects of bipolar LVDC grids

Power electronics are an essential part in this modernization of the electrical
system. The large presence of DC-based equipment, along with variable voltage/
frequency DGs require power converters for the connection to DC grids. For these
reasons, the following sections will provide a topological analysis of the different
stages involved in this promising DC network architecture. Both, the distribution
converters, that enable the split-DC bus network and balancing converters, which
perform the power consumption relocation to keep the system balanced will be
covered and briefly discussed.
10.3.1 Distribution converter topologies

The grid-tied or distribution converter stage plays a fundamental role in the DC
network of Figure 10.1, as it acts as the power hub on the LVDC grid [8,23,24]. For
this reason, the bidirectional power flow capability is mandatory, besides a wide set
of additional features given the challenges associated to distributed generation.
High power quality, adjustable power factor, low filtering requirements, reliability,
while simultaneously featuring a reduced number in both active and passive com-
ponents are desirable features in this stage [25,26].
A brief review of the most common converters used as the grid interface of DC
microgrids can be found below. Despite several topologies are able to act as the
rectifier stage of the system, only two alternatives will be analyzed in detail: the two-
level voltage source converter (2L-VSC) and the three-level neutral point clamped
(3L-NPC) converter. This lies in their proven cost effectiveness, efficiency and
reliability, besides being the two most widespread three-phase converters employed
in the industry [23].
Ip
Sa Sb Sc
Lg iga Vd1 C1
Iz
MVAC Vd2 C2
Sa Sb Sc
In
Figure 10.3 2L-VSC with neutral line connected to DC midpoint
10.3.1.1 Two-level voltage source converter

The 2L-VSC displayed in Figure 10.3 has a strong presence in the industry, acting
as the grid interface for a wide series of applications. As exhibited in the figure, the
circuit consists of an array of six switching devices, typically insulated-gate bipolar
transistors (IGBTs) and a DC-link capacitor. The presence of these active switches,
together with a proper control scheme, generates sinusoidal currents at the input
side, along with adjustable power factor and bidirectional power flow [27]. This
converter also steps up the voltage to higher values than the input grid voltages.
The generation of the gating signals can be done using carrier-based pulse-
width modulation (PWM), space vector modulation (SVM), besides variable
switching frequency methods, such as table or prediction-based methods [28–30].
The resulting binary pulse train in the phase voltages, which alternate between 0
and Vd originates the name of the converter. The topology presents a line-to-line
voltage with a three-level waveform. This impacts the total harmonic distortion
(THD) of the input current, as larger active/passive filters or higher switching
frequencies are needed in order to meet the requirements imposed by the grid code.
This topology has been widely applied to different types of bipolar DC net-
works [7,8,16,23,31,32]. Among the options explored, the simplest approach for
implementing a bipolar system is through the use of two cascaded 2L-VSC and a
multi-winding transformer, as exhibited in Figure 10.4. In this configuration, each
2L-VSC is employed to enable independent DC voltage sources. Consequently, this
leads to independent operation of the positive and negative poles, hence there are
no issues during asymmetrical operation. However, this alternative is not the pre-
ferred method for operating as the distribution converter. Despite the simplicity of
the operation offered by the cascaded connection, this implementation requires
additional hardware as two separate converters along with the distribution trans-
former based in two isolated windings. This increases the cost and the size of the
total system. Additionally, the topology in Figure 10.4 requires its secondary
windings to be designed to withstand the DC offset caused by the series connection
of the converters and the asymmetries that could arise in loads being fed [32].
A single 2L-VSC can also generate a bipolar DC output with additional cir-
cuitry. These configurations are all based in the generation of alternative paths for
the returning currents, which otherwise flow through the DC bus midpoint during
asymmetrical operation and therefore unbalance the DC voltages. The system
Ip
S Sb Sc
Lg iga a
Vd 1 C1
- - -
Sa Sb Sc
Iz
MVAC S S b̃ S c̃
L g igã ã
Vd2 C2
- - -
S ã S b̃ S c̃
In
Figure 10.4 Cascaded 2L-VSC with bipolar DC output
exhibited in Figure 10.3 does it by connecting to the neutral point of the transfor-
mer with the DC bus midpoint. This additional current path allows the regulation of
the midpoint current, in order to prevent DC bus voltages from drifting and main-
tain their proper regulation. However, the generation of non-zero DC currents at the
AC side can lead to transformer saturation and thereby it should be strictly limited.
To avoid this issue, different approaches have been proposed in the literature
[7,8,16,23,31,32], both based in passive or active methods, the latter ones enabled
by power electronics. The main solutions will be covered in the upcoming sections.
10.3.1.2 Three-level neutral point clamped converter

Another topology with a significant acceptance in industry applications is the
3L-NPC converter. Despite being originally proposed for MV drives, the 3L-NPC
has found his way to lower power applications during the last years, mainly in PV
applications. This inherent versatility in a wide voltage-power combination makes
it a natural candidate to act as the distribution converter. At present, commercial
NPCs reach voltages in the range of 2.2–6.6 kV, within a wide power rating range
(3 kW–50 MW). This converter is used either as an inverter that feeds AC loads or
as an active front-end converter that interfaces with the utility grid [33].
From the power circuit displayed in Figure 10.5(a), each phase of the 3L-NPC
converter is composed by four switching devices, which can either be IGBTs or
integrated gate-commutated thyristors, and two diodes connected to the neutral
point. The presence of these diodes allows generating a third voltage level in the
synthesized voltage, through connection of the phase with this neutral point when
the inner devices are turned on. Thereby, the converter is able to generate an output
voltage equal to Vd2 , 0 and Vd1 . The proper regulation of this floating voltage
represents a critical aspect for the 3L-NPC. Otherwise, the generated voltage
becomes distorted, worsening the power quality and also potentially damaging
the switching devices. These reasons have motivated important efforts in the
literature to solve this matter [34–36].
The selection of this topology as the distribution converter of the DC network
in Figure 10.1 is because of its higher grid power quality (three levels in the voltage
Ip
Sa1 Sb1 Sc1
Vd1 C1
Sa2 Sb2 Sc2
Lg iga Iz
z
MVAC Sa1 Sb1 Sc1
Vd2 C2
Sa2 Sb2 Sc2
In
(a)
1
0.8
Load ratio ε
0.6
0.4
0.2 Balanced operation

Balancing circuit
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
(b) Modulation index ma
Figure 10.5 Grid-connected three-level neutral point clamped converter:

(a) power circuit and (b) unbalanced load limitation
waveform), superior harmonic performance (reduced THD), lower switching fre-

quency, lower transformer ratio and the possibility of providing certain room for
unbalanced bipolar dc bus operation [17,23,34,37]. When feeding a bipolar DC
network, this topology enables a limited asymmetrical load operation.
By injecting a DC bias in the phase-to-neutral voltage, the 3L-NPC is able to
absorb current through its neutral point, hence extending the operation of the
active distribution network to a certain range of asymmetries. However, this
midpoint current capability is limited to a fraction of the DC current being fed,
i.e., the asymmetrical scenarios that the system can handle while keeping the
midpoint voltage controlled are limited and dependent of the amplitude modula-
tion index [17,23].
The unbalanced operation limitation is well defined for this converter as pre-
sented in Figure 10.5(b), and it can be used as a foundation for the balancing
scheme [17]. Unlike the previous converter, this topology can contribute to the
balancing of the buses, leading to two different kinds of balancing strategies that
extend the operation of the network. Despite both approaches require additional
circuitry they differ in the balancing efforts made by the stages. Following the
balancing schemes for the 2L-VSC, one possibility is to make all the balancing
efforts with the added circuits, through a virtual disconnection of the neutral point
from the unbalanced loads. In this way, the 3L-NPC focuses on regulating the input
currents and power quality while the DC voltages remain unaltered. This leads to
higher current stresses on the balancing stages, so interleaving channels can be
promoted in order to reduce their ratings [37–39].
A different concept is to implement a coordinated balancing scheme, where
both the central converter and the balancing stages share the current redistribution
efforts [17,20]. The balancing principle is the same as with the 2L-VSC, and the
exceeding current should be redistributed by the means of a power converter stage.
The DC-link structure opens the possibility to three-level DC–DC stages as it will
be covered in the upcoming section.
10.4 Balancing topologies
As depicted in the previous sections, the realization of a bipolar system using a

single power converter is not sufficient to handle asymmetrical generation or
unbalanced loads [7], hence additional circuitry is mandatory. The presence of
these circuits that balance the currents at DC side is beneficial for several reasons.
Depending on the way these converters interact with the DC systems can be
denoted as voltage balancers or current redistributors. Voltage balancers are DC–
DC converters that generate a split-DC bus based on the total DC voltage supplied
by the rectifier Vd , and are also able to perform the regulation of the center line.
Thus the voltages of the bipolar system (Vd1 and Vd2 ) are balanced to Vd =2.
On the other hand, current redistributors are used to balance the poles in
bipolar systems. This redistribution reduces the current flowing through the neutral
conductor to virtually zero, thereby the efficiency and stability of the feeder are
enhanced [7]. Moreover, balanced operation also eases the implementation and
operational requirements of the rectifier stage. This is related with the possibility of
using off-the-shelf products for the rectifier stage, as there are no special require-
ments on its operation. By adding a balancing stage, the rectifier can be used
exclusively for the grid-connection related tasks adjusting the power factor, reg-
ulating Vd and shaping the input currents [7,19].
The presence of these circuits also allows to isolate the neutral line in the
system and facilitate its connection to ground, thus improving the safety levels in
case of remote energy supply.
Taking into consideration the structure of the DC bus enabled by these power
converter stages, the balancing circuitry will be denoted as voltage balancer or
current redistributor. The basic operational principle of these complementary cir-
cuits is the possibility of relocating the energy consumption or generation, in such a
way that voltage balancing is achieved regardless the power flow between the DC
buses. To perform such corrections, these stages should have access to both of the
DC buses [37].
This section will provide an overview of the main power converter topologies,
as well as other non-power-electronics-based approaches for guaranteeing the
independent operation of the DC buses, regardless the load/generation condition.
Ip
Sa Sb Sc Sd
Lg iga Vd1 C1
Iz
MVAC
S¯a S¯b S¯c S¯d Vd2 C2
In
Figure 10.6 2L-VSC with voltage balancer enabling a bipolar DC output
+ + + +
S1 Lb1 S1 S1
Vd1+ C1 Vd1+ C1 Lb1 Vd1+ C1 Vd1+ C1
Lb Cb Lb2
Vd Vd Cb + Vd +
Vd +
Cb S2 S2
S2 +
Vd2 + S1 Lb2 Vd2+ C2 Lb1 S2 Vd2+ C2
C2
Lb2 S2 Vd2 C2
– – – –
(a) (b) (c) (d)
Figure 10.7 Bipolar balancer topologies derived from the bidirectional

buck/boost concept: (a) buck/boost; (b) Cuk; (c) super-Sepic;
and (d) super-Zeta
10.4.1 Bidirectional buck-boost topologies

The most elemental topology to generate a regulated bipolar DC bus with a single
2L-VSC is shown in Figure 10.6, where a buck-type voltage balancer has been
added. The presence of this additional half-bridge leg permits splitting the DC bus
voltage Vd into Vd1 and Vd2 . Additionally, these balancers allow the active redis-
tribution of the current during unbalanced scenarios [7,8]. This bidirectional buck/
boost structure is beneficial for the distribution transformer as well, because there
are no DC currents flowing through the AC neutral in contrast to the circuit of
Figure 10.3. The idea of this configuration is to reduce the dependency between the
operation of the poles, by properly modeling the inter-pole dynamics at the DC side
in order to adjust the currents accordingly. The result of this relocation of the power
consumption allows to keep the floating midpoint voltage tightly regulated.
Following the buck/boost concept, many balancing converters can be derived
using conventional DC–DC stages as building blocks, however it is important to
note that the bidirectional power flow is mandatory for modern distribution net-
works [7]. Figure 10.7 presents different balancing stages based on this structure,
which is displayed in Figure 10.7(a).
The circuit displayed in Figure 10.7(b) is a Cuk-type voltage balancer, which
besides providing the regulation of the bipolar bus eliminates the shoot-through pos-
sibility of the switching devices. The reason for this feature is the presence of an
impedance instead of a purely capacitive element as the DC-link, which is typical in
voltage source converters. Following the same concept, super-Sepic- and super-Zeta-
type voltage balancers have also been developed in the literature, and are displayed in
Figure 10.7(c) and 10.7(d), respectively [20]. Another way of enhancing the benefits of
I pc
+ Sa Sb Sc
Vd1 C1 Lb
Izc +
Vo Co
+
Vd2 C2 S¯a S¯b S¯c
Inc
Figure 10.8 2L-VSC-based current redistributor
Ip
Sa Sb Sc +
Lg iga Vd1 C1
Iz
MVAC +
Lc S¯a S¯b S¯c Vd2 C2
In
Figure 10.9 2L-VSC with a grounding inductor balancer
the voltage balancing stage is to use additional channels and operate them interleaved.
This connection will reduce the high current ripple exhibited in the DC inductor Lb of
the buck/boost stage shown in Figure 10.6, as the current will be equally shared by the
additional channels. By means of this configuration it is possible to reduce the inductor
size, as the interleaved operation increases the output equivalent frequency. This leads
to a smooth output current, which allows the reduction in the output inductor by a factor
of 1/n, for an n-channel converter [24].
The circuit in Figure 10.8 can also serve towards the power balance of the
buses. The topology proposed in [7] regulates the positive and negative bar currents
in such a way that the neutral current Izc is almost zero. However, this is not
sufficient to guarantee a stable operation, as the exceeding power must be relocated
to the opposite pole and returned to the grid. The result is a control scheme based
on the sum–difference domain, that ensures balance for asymmetrical operation in
split-DC networks. As it will be shown in the next section, the resemblance with a
three-phase system motivated a control scheme based on the symmetrical compo-
nent method applied for grid-connected converters.
10.4.2 Coupled inductor current redistributor

Following the same concept of Figure 10.3, the topology proposed in Figure 10.9
uses a grounding reactor to provide the current injection path between the AC side
terminals and the DC neutral conductor [32]. The main idea is to regulate the zero-
sequence DC current during unbalanced scenarios while limiting the AC currents,
in order to maintain the system’s efficiency. This is achieved through the proper
design of the grounding inductor, which uses a coupled magnetic core that reduces
the chances of saturation due the presence of DC flux.
Ip I pc
S1 S1
Lb
Vd1 C1 Vd1 C1
S2 S2
Iz ib Izc
Vo Co
Lb S S3
3
Vd2 C2 Vd2 C2
S4 S4
In Inc
(a) (b)
Figure 10.10 Three-level current redistributing topologies: (a) 3L-NPC

balancer and (b) three-level buck/boost
The design of the core leads to zero-sequence inductances that are substantially
lower than the stationary frame ones, which are desired to be large in order to
reduce sustained AC currents given the connection. The structure of the inductor
enables a viable path for the returning DC currents, enhancing the controllability of
the converter and regulating the DC bus voltages regardless the load condition.
This approach offers a trade-off between having less power electronics and
increasing the losses in the coupled inductor showing promising results.
10.4.3 Three-level DC–DC current redistributors

Given the 3L-NPC structure, which already includes a split DC bus, makes it a
natural candidate for bipolar DC networks. Nevertheless, despite this converter is
able to balance the midpoint voltage within a certain imbalance range, it cannot
guarantee the stability for any operation mode. This limitation motivated the pro-
posal in [17], where an additional 3L-NPC leg has been added and connected as
exhibited in Figure 10.10(a). The balancing leg can be seen as a bidirectional buck,
whose input is the full DC voltage and the output is the midpoint voltage, i.e., the
duty cycle will tend to 0:5 when the balance is reached. This is a crucial feature, as
it allows to determine a priori the current flowing through the DC inductor needed
to achieve balance. Consequently, this fourth leg performs the complementary
balancing functions, extending the range achieved by the 3L-NPC and allowing
balanced DC voltages even under severe unbalanced load conditions (rated load/
generation in one pole while the other one has no load/generation).
Please note that, if the DC inductor is properly sized, the aforementioned approach
is also able to exclusively handling the asymmetries at the DC side, performing
the virtual disconnection of the midpoint while the distribution converter focuses on the
regulation of the grid-side quantities. The main drawback will be that Lb should handle
the rated DC current, instead of a fraction of it as the approach in [17].
A different approach is proposed in [19], where this 3L-NPC leg is modulated
differently, in order to include free-wheeling times in the switching sequence in
such a way that the exceeding current is rearranged and keeps the DC voltages
controlled. The idea is the following, when asymmetrical power consumption
exists, the modulator takes the exceeding current from the bus lightly loaded and
supplies it to the other bus through the DC inductor.
The works in [38,39] use a different three-level DC–DC converter for feeding
high-power loads and also contribute to minimize power fluctuations in bipolar DC
systems. The power circuit is presented in Figure 10.10(b), where it can be seen
that the basic unit is composed by four switching devices and their corresponding
free-wheeling diodes, along with an output inductor and the common output filter
capacitor. Depending on the power level, more units can be connected in inter-
leaved configuration to minimize current stress in the switches and inductors [38].
One bipolar application of the aforementioned topology is interfacing an ESS
[39]. By replacing the additional balancing leg, this converter allows to accomplish
two simultaneous objectives: the main one is to perform the charge and discharge
of the storage element according to the energy management supervisor controller
of the DC network, but at the same time uses this process to relocate the power
consumption at DC level and complement the balancing task performed by the
distribution converter. Given the nature of the topology, the system is able to
guarantee stable operation with an ESS with reduced power ratings.
Alternatively, in [38], two interleaved channels of the converter in Figure 10.10(b)
are used to feed high-power loads. Besides enhancing the output power quality, reduce
the current stress and the filter size and its modularity, this three-level topology can also
assist the distribution converter to perform DC power balancing, allowing to improve
the overall system efficiency and the quality of the AC input currents. From the con-
trollability point of view, this stage can be either complementing the balancing efforts
carried by the central converter, or exclusively perform the balance in a distributed
manner between the remaining stages as suggested by the authors.
10.5 Control schemes

Following the topological analysis of bipolar DC distribution networks, a brief sur-
vey on control techniques used for enabling these configurations is presented. First,
the main schemes used for the balancing the currents at the DC side, followed by
grid-tied AC–DC stages, both single and three-phase approaches will be reviewed.
Given the complexity in the dynamics involved in the aforementioned topic,
leads to the necessity of implementing sophisticated control schemes that share the
same structure. This architecture corresponds to a two degree-of-freedom config-
uration for improving the dynamics while dealing with disturbances. Such structure
is known as Cascade Control. The main idea is based on the feedback of intermediate
variables, which have a direct relation with the main output of the system, in order to
improve the dynamic response of the loop and the robustness of the closed loop.
10.5.1 Cascade control

Figure 10.11 shows the diagram of the cascade control structure. In the figure, two
control loops are shown: the primary (or outer) loop with its controller C1 ðsÞ and a
secondary loop with its controller C2 ðsÞ. The design of the secondary loop is made
Go(S)
r1 + u1 r2 + u2 y2 y1
C1(S) C2(S) Go2(S) Go1(S)
– –
Inner loop
Outer loop
Figure 10.11 Cascade control structure
in such a way that it mitigates the effects of disturbances before they significantly
affect the system output y1 [40].
The main benefits of the cascaded control are obtained when when Go2 ðsÞ
presents significant non-linearities that limit loop performance or when Go1 ðsÞ
limits the bandwidth in a basic control loop (i.e., the presence of non-minimum
phase zeros and/or pure time delays). The first benefit, is explained because the
inclusion of C2 ðsÞ allows a high gain control, which tends to reduce the effect of
nonlinearities. The latter one, is because this inner controller pre-compensates the
effect of the disturbances that C1 ðsÞ has to deal with [40].
10.5.2 AC–DC converter control

The main goal of a grid-tied converter is enabling a regulated DC voltage, while
controlling the input currents at the grid side. However, these are not the only
features desired for a grid-tied converter, as requirements for these converters have
increased through time. The significant presence of power converters connected to
the utility grid, has imposed harder requirements through time. Thereby, in addition
to their main functions they are also required to reduce their electromagnetic
interference, harmonic content, reactive power generation, etc. The control
schemes have updated accordingly in order to maintain their performance and cope
with the additional objectives imposed by more stringent grid codes. Besides
controlling the aforementioned quantities, ideally an active rectifier stage should
provide additional functionalities such as:
● The reduction of the harmonic content because of its negative effect on the
electric system (e.g., voltage distortion, electromagnetic interference, increased
power ratings of power system equipment and so on).
● Adjustable power factor
● Bidirectional power flow
The appearance of these requirements has resulted in new challenges for the
control schemes and has prompted researchers to improve them to meet the control
objectives and at the same time maintain high performance.
10.5.2.1 Single-phase AC–DC converter control

For the regulation of single-phase rectifiers the control scheme presented in
Figure 10.12. From the figure, it can be seen that the scheme uses a cascade control
structure, where the grid-current is regulated through a proportional resonant
( )2
Vd
Notch +
filter
(Vd *)2+ – Pg* ig* + vL*– vc* S

PWM 1f -VSC
– + Gating
PI PR signals
Lg
1/Vg
ig
sin(wgt) vg
PLL
Grid
Figure 10.12 Single-phase AC–DC converter stage control scheme
vgd
igd*+ Vd
Vd*+ – + + +
– +
PI PI vcd*
wLg dq
igd Vref S
igq vcq* PWM 3f -VSC
wLg abc Gating
signals
igq*+ – + – Lg
Vg
+ θg
PI vgq PLL
vgd
dq axis vgq dq
decoupling igd ig
igq abc
Grid
Figure 10.13 Three-phase AC–DC converter stage VOC control scheme
controller in order to enhance the DC-link voltage dynamics, which are regulated
by an outer proportional integral (PI) controller. Note that the feedback of the DC
voltage is filtered with a notch in 2 wg to minimize the impact of the rectification
ripple in the grid current. For the synchronization of the scheme, a phase-locked
loop (PLL) is employed to guarantee the unity power factor operation.
10.5.2.2 Voltage-oriented control

Figure 10.13 displays the three-phase counter part of the previous control method.
This control scheme is called voltage-oriented control (VOC). Derived from the flux-
oriented control (FOC) of an induction motor [41], VOC is based on a cascade control
structure and a coordinate transformation between the ab stationary frame and the
Vd*+ – Pg*+ hP
– Switching S
PI Pg Qg*+ hQ 3f -VSC
table Gating
–
qk signal
Qg
Lg
qg vg
PLL
Power ig
estimator
Grid
Figure 10.14 Three-phase AC–DC converter stage DPC control scheme
dq synchronous frame. The result is fast transient response and high static perfor-
mance, with no steady-state error, as a result of the DC nature of the controlled
quantities [29].
From Figure 10.13, it can be seen that the measured signals are converted to
the dq frame, which is aligned with the grid voltage using a PLL, hence the name of
the scheme. This allows to decouple the regulation of the active and reactive power
through the orthogonal components of the current, igd and igq , respectively.
For the regulation of the current components and the DC-link voltage, PI
controllers are used, leading to zero steady-state tracking errors and good dynamic
performance. Finally, their actuations are transformed back to a three-phase refer-
ence frame and passed to the modulator. This modulation stage can be either PWM
or SVM, for the generation of the gating signals. This control scheme can also be
applied to the NPC converter, by modifying the modulation stage accordingly and
also considering an midpoint balancing mechanism. Conventionally, a PI controller
is implemented in order to perform the partial balancing strategy, which regulates
the midpoint deviations depending on the selected modulation strategy [17].
10.5.2.3 Direct power control

Another mainstream method for regulating three-phase converters is the direct
power control (DPC), presented in Figure 10.14. The scheme is also based on the
cascade control structure, except that the inner loop controllers are nonlinear.
The DC-link voltage Vd is controlled by an outer linear PI controller, which
provides the reference for the active power Pg , whereas the reactive power refer-
ence Qg can be set arbitrarily depending on the selected Vd reference. Both powers
are estimated from the grid voltage and current measurements feedback and con-
trolled with nonlinear hysteresis comparators. Their outputs hP and hQ along with
the voltage angle qs , are used to access a voltage vector look-up table. Finally, the
table delivers the gating signals to generate the selected voltage vector. Although
this method is designed for the conventional 2L-VSC, various adaptations for the
2ε
Ip
i*b + db Sd
In Max PWM
–
Iz ib PI
sgn
Figure 10.15 Voltage balancing circuit controller for the additional NPC leg
3L-NPC converter exist, which add an additional hysteresis comparator to regulate

the midpoint corrections [42–44].
10.5.2.4 Balancing leg control

The idea behind this approach is to extend the bipolar NPC operation through the
entire load range, either using the maximum balancing capabilities offered by the
central converter modulator or to completely perform the balancing of the system.
To do so for the first case, the balancing leg must be able to act as a virtual
impedance, such that the minimal load condition is met, and the modulator is able
to keep the voltages balanced. For the latter case, the balancing leg should absorb
the full difference between the currents flowing through the poles, which in other
words can be seen as the virtual disconnection of the midpoint.
Given the connection of the balancing leg, the duty cycle required to balance the
system is known a priori as the midpoint voltage should be equal to Vd =2, which allows
to exactly determine the current required to be handled by the balancing converter in
order to balance the bipolar system, and a simple PI controller will suffice to perform
the corrective actions and provide a good dynamic performance. The control scheme
for this converter is exhibited in Figure 10.15, considering the complementary balance
case and operates whenever the load ratio of the dc buses is lower than the critical ratio
e allowed by the modulation scheme [17]. If the current distributor is performing
exclusively the balancing tasks, then the balancing current reference ib is given by the
difference Ip In . This will result in the virtual disconnection of the midpoint, hence
keeping the DC voltages balanced in the presence of any imbalance.
10.5.2.5 Symmetrical component control-based methods

The similarity of the analyzed DC network with three-phase AC systems is also
beneficial from the control scheme perspective. The symmetrical component method
is a powerful and useful tool for analyzing and controlling unbalanced AC systems
[45–47]. Accordingly, this technique can also be applied in bipolar DC networks, so a
reference frame transformation T can be derived from AC theory that leads to:

xS 1 1 xp
¼ 0:5 : (10.1)
xD 1 1 xn
|fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}
T
Dp
Iz* + + I∑
* Ipc* + + +
– + –
Iz PID Ipc PI S
damp. T–1 + –
1+Do PWM VSC
–
Vo* + + I∆* *
Inc + +
– + – +
Vo PI Inc PI Dn
damp.
Figure 10.16 DC symmetrical component control scheme
The importance of this transformation is that decouples the collective power

transfer of the two poles and the inter-pole interaction to different components
[7,8], namely the differential mode xD and the common mode xS . This is relevant
for the analysis and control of the system, as it allows to develop decoupled circuits
and therefore allows to precisely model the inter-pole dynamics. Based on the
control scheme in Figure 10.13, the symmetrical component control scheme is
presented in Figure 10.16. From the figure, it becomes clear how the balancing
current Iz holds a relation with the common mode current IS while the output
voltage Vo is related with the differential current ID . Both, the output voltage and
the neutral current are regulated with linear controllers and set the references for
the sequence components. These references are transformed back to the DC side
and the references for the positive and negative bar currents in the redistributor, for
example the circuit in Figure 10.8. Finally, these currents Ipc and Inc are regulated
using PI controllers to generate the duty cycles for each leg in the converter.
10.6 Summary
During the last decades, AC systems dominated the power transmission and dis-
tribution applications almost exclusively. However, a recent convergence of needs
originated in different sectors (renewable energy conversion, information tech-
nology and transportation) have accelerated the development of DC systems. Nowa-
days, DC systems are present at both transmission and distribution levels, offering
high-performance solutions with enhanced efficiency and reliability, besides reducing
the number of power conversion stages involved and uninterrupted power delivery.
For LVDC active networks, two kinds of architectures are possible: unipolar and
bipolar. Despite being a more sophisticated and technically complex solution, bipolar
structure provide several advantages over conventional unipolar ones. Higher avail-
ability, efficiency and flexibility are just a few advantages featured by bipolar systems.
A typical bipolar grid has certain similarities with a three-phase AC system, where
the phase voltage in an AC system is the analog to the voltage between one pole and the
neutral point in the bipolar LVDC grid, and the line-to-line AC voltage can be con-
sidered as the voltage between the positive and negative poles of the bipolar LVDC
grid. This resemblance with AC systems is beneficial for the operation, retrofitting and
also for developing the control schemes.
However, if the loads connected to each pole of this LVDC grid are unba-
lanced, the system operates asymmetrically. This uneven power distribution along
with the dependency between the poles can be problematic if not taken care
properly. The main issues are the imbalances in the DC voltages, higher losses and
a decreased power quality of the system.
In order to solve these imbalance problems additional circuitry is required.
Voltage balancers and/or current redistributors are the key players that allow to deal
with asymmetrical operation and keep the system in a balanced and stable operation.
As it was discussed in the chapter, the converter topologies employed in bipolar
LVDC systems need to be bidirectional, in order to allow the reversible power flow
given the flexible structure of distributed networks.
This chapter presented a brief overview covering the different aspects of
bipolar LVDC networks. Distribution converter topologies, balancing stages and
also their control schemes are discussed in order to highlight the efforts being made
in this growing architecture.
Acknowledgement
The authors acknowledge the support provided by the CONICYT projects FON-
DECYT 11170774, AC3E (Basal/FB0008) and SERC Chile (FONDAP/15110019).
References
[1] D. Salomonsson. Modeling, Control and Protection of Low-Voltage DC
Microgrids. PhD thesis, Royal Institute of Technology, Sweden, Apr. 2009.
[2] C. W. Taylor and S. Lefebvre. HVDC Controls for System Dynamic
Performance. IEEE Trans. Power Syst., 6(2):743–752, May 1991.
[3] B. H. Bakken and H. H. Faanes. Technical and Economic Aspects of Using a
Long Submarine HVDC Connection for Frequency Control. IEEE Trans.
Power Syst., 12(3):1252–1258, Aug. 1997.
[4] L. Zhang, L. Harnefors, and H. P. Nee. Interconnection of Two Very Weak
AC Systems by VSC-HVDC Links Using Power-Synchronization Control.
IEEE Trans. Power Syst., 26(1):344–355, Feb. 2011.
[5] L. Zhang, L. Harnefors, and H. P. Nee. Modeling and Control of VSC-
HVDC Links Connected to Island Systems. IEEE Trans. Power Syst., 26(2):
783–793, May 2011.
[6] G. P. Adam, K. H. Ahmed, S. J. Finney, K. Bell, and B. W. Williams.
New Breed of Network Fault-Tolerant Voltage-Source-Converter
HVDC Transmission System. IEEE Trans. Power Syst., 28(1):335–346,
Feb. 2013.
[7] J. Lago and M. L. Heldwein. Operation and Control-Oriented Modeling of a
Power Converter for Current Balancing and Stability Improvement of DC
Active Distribution Networks. IEEE Trans. Power Electron., 26(3):877–885,
Mar. 2011.
[8] Y. Gu, W. Li, and X. He. Analysis and Control of Bipolar LVDC Grid With
DC Symmetrical Component Method. IEEE Trans. Power Syst., 31(1):
685–694, Jan. 2016.
[9] H. Kakigano, Y. Miura, and T. Ise. Distribution Voltage Control for DC
Microgrids Using Fuzzy Control and Gain-Scheduling Technique. IEEE
Trans. Power Electron., 28(5):2246–2258, May 2013.
[10] H. Kakigano, Y. Miura, and T. Ise. Low-Voltage Bipolar-Type DC Micro-
grid for Super High Quality Distribution. IEEE Trans. Power Electron.,
25(12):3066–3075, Dec. 2010.
[11] T. Dragicevic, X. Lu, J. C. Vasquez, and J. M. Guerrero. DC Microgrids –
Part II: A Review of Power Architectures, Applications, and Standardization
Issues. IEEE Trans. Power Electron., 31(5):3528–3549, May 2016.
[12] Directive 2014/35/EU of the European Parliament and of the Council of
26 February 2014 on the Harmonisation of the Laws of the Member States
Relating to the Making Available on the Market of Electrical Equipment
Designed for Use within Certain Voltage Limits (recast) (Text with
EEA Relevance). Official Journal of the European Union, L 96:357–374,
Mar. 2014.
[13] T. Kaipia, P. Salonen, J. Lassila, and J. Partanen. Possibilities of the Low
Voltage DC Distribution Systems. In Nordic Distribu. and Asset Manag.
Conference (NORDAC), Stockholm, Sweden, Aug. 2006.
[14] D. Salomonsson and A. Sannino. Low-Voltage DC Distribution System for
Commercial Power Systems With Sensitive Electronic Loads. IEEE Trans.
Power Del., 22(3):1620–1627, Jul. 2007.
[15] H. Kakigano, Y. Miura, T. Ise, and R. Uchida. DC Voltage Control of the
DC Micro-grid for Super High Quality Distribution. In Power Conversion
Conference – Nagoya, 2007. PCC ’07, pp. 518–525, Apr. 2007.
[16] Byung-Moon Han. A Half-Bridge Voltage Balancer with New Controller for
Bipolar DC Distribution Systems. Energies, 9(3): 182, Mar. 2016.
[17] S. Rivera, B. Wu, S. Kouro, V. Yaramasu, and J. Wang. Electric Vehicle
Charging Station Using a Neutral Point Clamped Converter with Bipolar DC
Bus. IEEE Trans. Ind. Electron., 62(4):1999–2009, Apr. 2015.
[18] J. Lago, J. Moia, and M. L. Heldwein. Evaluation of Power Converters to
Implement Bipolar DC Active Distribution Networks – DC-DC converters.
In 2011 IEEE Energy Conversion Congr. and Expo. (ECCE), pp. 985–990,
Sep. 2011.
[19] X. Zhang, C. Gong, and Z. Yao. Three-Level DC Converter for Balancing DC
800-V Voltage. IEEE Trans. Power Electron., 30(7):3499–3507, Jul. 2015.
[20] F. Wang, Z. Lei, X. Xu, and X. Shu. Topology Deduction and Analysis of
Voltage Balancers for DC Microgrid. IEEE Trans. Emerg. Sel. Topics Power
Electron., 5(2):672–680, Jun. 2017.
[21] E. Rodriguez-Diaz, M. Savaghebi, J. C. Vasquez, and J. M. Guerrero.
An Overview of Low Voltage DC Distribution Systems for Residential
Applications. In 2015 IEEE 5th International Conference on Consumer
Electronics – Berlin (ICCE-Berlin), pp. 318–322, Sep. 2015.
[22] E. Vossos. Optimizing Energy Savings from Direct-DC in U.S. Residential

Buildings. MSc thesis, San Jose State University, San Jose, California, USA,
Aug. 2011.
[23] J. Moia, J. Lago, A. J. Perin, and M. L. Heldwein. Comparison of Three-
Phase PWM Rectifiers to Interface AC Grids and Bipolar DC Active
Distribution Networks. In Power Electron. for Distributed Generation Syst.
(PEDG), 2012 3rd IEEE Int. Symp. on, pp. 221–228, Jun. 2012.
[24] S. Rivera, S. Kouro, and B. Wu. Charging Architectures for Electric and
Plug-In Hybrid Electric Vehicles, pp. 111–149. Springer International
Publishing, Basel, Switzerland, 2017.
[25] S. Dusmez, A. Cook, and A. Khaligh. Comprehensive Analysis of High
Quality Power Converters for Level 3 Off-Board Chargers. In 2011 IEEE
Vehicle Power and Propulsion Conference, pp. 1–10, Sep. 2011.
[26] A. Khaligh and S. Dusmez. Comprehensive Topological Analysis of
Conductive and Inductive Charging Solutions for Plug-In Electric Vehicles.
IEEE Trans. Veh. Technol., 61(8):3475–3489, Oct. 2012.
[27] James W. A. Wilson. The Forced-Commutated Inverter as a Regenerative
Rectifier. IEEE Trans. Ind. Appl., IA-14(4):335–340, Jul. 1978.
[28] B. Wu. High-Power Converters and AC Drives. Wiley-IEEE Press, Hoboken,
New Jersey, USA, Apr. 2006.
[29] M. P. Kazmierkowski, R. Krishnan, and F. Blaabjerg (editors). Control in
Power Electronics: Selected Problems. Academic Press Series in Engineering,
Cambridge, Massachusetts, USA, Sep. 2002.
[30] J. Rodriguez and P. Cortes. Predictive Control of Power Converters and
Electrical Drives. Wiley-IEEE Press, Chichester West Sussex, UK, Apr.
2012.
[31] T. H. Jung, G. H. Gwon, C. H. Kim, J. Han, Y. S. Oh, and C. H. Noh.
Voltage Regulation Method for Voltage Drop Compensation and Unbalance
Reduction in Bipolar Low-Voltage DC Distribution System. IEEE Trans.
Power Del., 33(1):141–149, Feb. 2018.
[32] Y. Li, A. Junyent-Ferré, and J. M. Rodriguez-Bernuz. A Three-Phase Active
Rectifier Topology for Bipolar DC Distribution. IEEE Trans. Power Electron.,
33(2):1063–1074, Feb. 2018.
[33] S. Kouro, M. Malinowski, K. Gopakumar, et al. Recent Advances and
Industrial Applications of Multilevel Converters. IEEE Trans. Ind. Electron.,
57(8):2553–2580, Aug. 2010.
[34] N. Celanovic and D. Boroyevich. A Comprehensive Study of Neutral-Point
Voltage Balancing Problem in Three-Level Neutral-Point-Clamped Voltage
Source PWM Inverters. IEEE Trans. Power Electron., 15(2):242–249,
Mar. 2000.
[35] A. Yazdani and R. Iravani. A Generalized State-Space Averaged Model of the
Three-Level NPC Converter for Systematic DC-Voltage-Balancer and Current-
Controller Design. IEEE Trans. Power Del., 20(2):1105–1114, Apr. 2005.
[36] Z. Mohzani, B. P. McGrath, and D. G. Holmes. The Balancing Properties
of DC Link Compensation for 3-Phase Neutral Point Clamped (NPC)
Converter. In Proceedings of The 7th International Power Electronics and

Motion Control Conference, volume 1, pp. 574–579, Jun. 2012.
[37] S. Rivera and B. Wu. Electric Vehicle Charging Station With an Energy
Storage Stage for Split-DC Bus Voltage Balancing. IEEE Trans. Power
Electron., 32(3):2376–2386, Mar. 2017.
[38] L. Tan, B. Wu, S. Rivera, and V. Yaramasu. Comprehensive DC Power
Balance Management in High-Power Three-Level DC-DC Converter for
Electric Vehicle Fast Charging. IEEE Trans. Power Electron., 31(1):89–100,
Jan. 2016.
[39] L. Tan, B. Wu, V. Yaramasu, S. Rivera, and X. Guo. Effective Voltage
Balance Control for Bipolar-DC-Bus-Fed EV Charging Station With Three-
Level DC-DC Fast Charger. IEEE Trans. Ind. Electron., 63(7):4031–4041,
Jul. 2016.
[40] G. C. Goodwin, S. F. Graebe, and M. E. Salgado. Control System Design.
Prentice Hall, Upper Saddle River, New Jersey, USA, Oct. 2000.
[41] F. Blaschke. The Principle of Field Orientation as Applied to the New
Transvector Closed-Loop System for Rotating-Field Machines. Siemens
Review, 34:217–220, May 1972.
[42] L. A. Serpa and J. W. Kolar. Virtual-Flux Direct Power Control for Mains
Connected Three-Level NPC Inverter Systems. In Power Conversion Con-
ference - Nagoya, 2007. PCC ’07, pp. 130–136, Apr. 2007.
[43] J. Eloy-Garcia, S. Arnaltes, and J. L. Rodriguez-Amenedo. Extended Direct
Power Control for Multilevel Inverters Including DC Link Middle Point
Voltage Control. IET Electronic Power Applications, 1(4):571–580, Jul.
2007.
[44] S. Rivera, S. Kouro, B. Wu, et al. Multilevel Direct Power Control –
A Generalized Approach for Grid-Tied Multilevel Converter Applications.
IEEE Trans. Power Electron., 29(10):5592–5604, Oct. 2014.
[45] G. C. Paap. Symmetrical Components in the Time Domain and Their
Application to Power Network Calculations. IEEE Trans. Power Syst.,
15(2):522–528, May 2000.
[46] F. J. Alcantara and P. Salmeron. A New Technique for Unbalance Current
and Voltage Estimation with Neural Networks. IEEE Trans. Power Syst.,
20(2):852–858, May 2005.
[47] M. Karimi-Ghartemani and H. Karimi. Processing of Symmetrical Compo-
nents in Time-Domain. IEEE Trans. Power Syst., 22(2):572–579, May 2007.
Chapter 11
Aircraft DC microgrids
Fei Gao1, Tao Yang2, Serhiy Bozhko2, and Pat Wheeler2
11.1 Introduction
Similar to utility grid and other terrestrial microgrids, aircraft electrical power system
(EPS) has its own power generation, distribution, utilization and energy storage. The
rapid development of power electronics technology has allowed the converters to
operate at DC voltage levels required for transmission, distribution and consumption.
Today there is a clear tendency that high voltage (HV) and low voltage have seen the
proliferation of DC systems and its implementation for transmission and distribution
of electricity in aircraft electrical systems.
This chapter will provide an overview of aircraft DC microgrids. Section 11.2
introduces the aircraft EPS, covering the topics from power generation, distribution
and utilization. Section 11.3 reviews aircraft electrical system standards and high-
lights the power quality and power factor requirement for aircraft applications.
Section 11.4 presents the concept of aircraft electric starter/generator system.
Control and stability analysis of aircraft DC microgrids are performed in Sections
11.5 and 11.6, respectively.
11.2 Aircraft electrical power system

Traditionally, there are two energy sources on conventional aircrafts. The primary
source is the engines on the aircraft. The secondary source includes electric,
hydraulic, pneumatic and mechanical systems powered by the primary source. The
power source distribution in conventional aircrafts is illustrated in Figure 11.1.
The more electric aircraft (MEA) concept is one of the major trends in modern
aerospace engineering aiming for reduction of the overall aircraft weight, operation
cost and environmental impact. As shown in Figure 11.2, electrical systems are
employed to replace existing hydraulic, pneumatic and mechanical actuators. As a
consequence, the onboard-installed electrical power increases significantly, and this
results in challenges in the design of the aircraft EPSs. In alignment with the tendency
1
Department of Engineering Science, University of Oxford, UK
2
Department of Electrical and Electronic Engineering, The University of Nottingham, UK
Jet fuel
Propulsion thrust
Gearbox-driven High-pressure air Gearbox-driven Fuel pumps, oil

generators “bled” from engine hydraulic pump pumps on engine
Electrical Pneumatic Hydraulic Mechanic
Figure 11.1 Power sources distribution in conventional aircraft [1]
Jet fuel
Propulsion thrust
Engine-driven
generators
Existing electrical Cabin pressurization Flight control actuation Fuel pumping

loads Air conditioning Landing gear/braking Engine ancillary
Icing protection doors
New electrical loads
Figure 11.2 Power sources distribution in more electric aircraft [1]
in the utility grid, the EPS architecture with DC distribution is considered as one of the
most promising candidates for MEA due to the potential advantages such as higher
efficiency compared to AC, less weight of harness, absence of reactive power related
issues and also reduced costs. Compared with existing DC distribution systems in
aircraft, the future architectures will move towards HV distribution.
11.2.1 Power generation

In large commercial aircrafts, one generator per engine typically performs electrical
generation. Depending on the aircraft type, there could be more than one generator
connected to each engine, such as back-up generators, in order to meet redundancy
Aircraft DC microgrids 269
and extended range twin-engine operations performance standard [2]. An addi-

tional source of electric power on an aircraft is an auxiliary power unit (APU),
which is essentially a small gas turbine-driven generator that can be used as an
additional power source. APU typically provides power when the aircraft is on the
ground and can also provide power, while in the air under emergency conditions.
Various batteries exist in the aircraft to start the APU and to provide back-up power
for critical equipment in the cockpit as well as other important functions such as the
emergency lighting for the aisles [3].
The 115 VAC at constant frequency (CF) aircraft system started in the 1960s.
This generation of electrical networks was equipped with fixed frequency (115/200
VAC) integrated drive generators as on the Airbus A320, A330 and A340. With
this architecture, the generator is connected to the main engine via a mechanical
drive, which keeps the mechanical speed, and hence the electrical frequency,
constant on the aircraft’s electric bus. The variable-frequency (VF) generations
have appeared since the 1990s. In order to simplify the constant-speed mechanical
gearbox, the VF generators are preferred, as on the Airbus A380 and on the next
A350, for which the voltage standard is doubled (230/400 VAC) with respect to the
increase in electrical power needs. Variable-speed CF (VSCF) aircraft appeared in
the 2000s. Due to the introduction of VF generation, several power loads are
supplied at fixed frequency, such as induction motor driven pumps, have been
replaced with power electronics controlled loads compatible with VF standards. For
example, this was the case for flight control actuation systems in which electro-
hydrostatic actuators were embedded on the A380 for hydraulic network backup.
As discussed above, the power generation system gradually evolves from 115
VAC CF, VSCF, to variable speed variable-frequency (VSVF) Table 11.1 lists the
power generation types of recent civil and military aircrafts.
11.2.2 Power distribution

EPS in conventional aircrafts often consists of two or more engine-driven gen-
erators to power AC loads. The engine-driven generators are connected to the
distribution buses in some civil aircraft configurations (i.e. each generator is
responsible for a specific numbers of buses). In the event that one generator fails, it
is automatically isolated from its respective busbar, and all busbar loads are then
taken over by the alternative generator.
If the electrical system and associated loads can be adapted to operate with a
VF, then it is possible to link a generator directly to the engine shaft, i.e., the
electrical output of the generator provides a VF supply with the frequency related to
the speed of the gas turbine, typically in the range of 360–800 Hz. EPS archi-
tectures with such VF distributions are known as ‘‘frequency-wild’’ systems. The
key advantage of these is the direct connection between the generator output and
the EPS, giving a simple and potentially very reliable configuration. The drawback
is that nearly all aircraft loads will require power converters for control, as the VF
supply cannot be used directly by some onboard loads. However, many loads, for
example, actuators, require this power conversion stage for control even when
Table 11.1 Power generation type for recent civil and military aircrafts [4]
Generation type Civil aircrafts Military aircrafts

Constant frequency B777 2 120 kV A Eurofighter
Typhoon
A340 4 90 kV A
MD-12 4 120 kV A
B747-X 4 120 kV A
B717 2 40 kV A
B737NG 2 90 kV A
B767-400 2 120 kV A
Variable speed con- F-18C/D 2 40/45 kV A
stant frequency F-18E/F 2 60/65 kV A
(cycloconverter)
115 VAC
Variable speed con- B777 (Backup) 2 20 kV A
stant frequency MD-90 2 75 kV A
(DC link)
115 VAC
Variable frequency Horizon 2 20/25 kV A Boeing JSF 2 50 kV A
(115 VAC/380- A380 4 150 kV A
760 Hz)
Variable frequency B787 4 250 kV A
230 VAC A350 4 100 kV A
operated from a fixed frequency supply. Having many distributed power converters
gives more options for a safe aircraft system design as redundancy can be built-in at
the systems level, avoiding any single points of failure within the design [1]. In new
aircrafts such as the Boeing B787, Airbus A380 and A350, this VSVF architecture
has been employed [5]. In these aircrafts, while the voltage is regulated at either
115 V (A380) or 230 VAC (B787), the frequency changes with the engine speed
and typically varies between 360 and 800 Hz [6].
Figure 11.3 shows the distribution system in B787. The 230 VAC bus,
generated by the variable speed synchronous generator, is then transformed and
rectified by an auto-transformer rectifier unit (ATRU) to supply 270 VDC bus.
The distribution buses also supply 115 VAC bus bars via auto transformer units,
and 28 VDC buses via transformer rectifier units (TRUs). This permits more con-
ventional consumers such as window heat, APU fuel pump and avionics compo-
nents, to be powered from less conventional 230 VAC supply. The availability of
the 115 VAC and 28 VDC buses therefore avoids the need to develop specialized
power conversion for each of the client loads. The distribution architecture also
furnishes 230 VAC direct to certain users for which no transformation is necessary.
These loads include the wing anti-ice heater mats, environment control system
(ECS) recirculation fans and cargo heaters. To provide a temporary source of
emergency power for essential loads, and to provide a source of energy for starting
the APU, the B787 has two 28 V batteries.
28 V
BAT
28 VDC bus
28 V BAT
EMERG LV
Emergency
loads
28 V
TRU loads
115 VAC
bus AC
ATU
loads
230 VAC bus
± 270 V DC
bus
ATRU
G1 ECS1
WIPS1
WIPS2 ± 270 V DC
G2 bus
ECS2
ATRU
115 VAC bus

AC
230 VAC bus ATU
loads
28 VDC bus
28 V 28 V
TRU loads ESS
Figure 11.3 Power generation and distribution system for Boeing 787 having twin
generators
With the development of HV technologies, there is a trend for using HVDC

power distribution system in MEA [7]. The benefits of VF generation with HVDC
distribution can be summarized as follows:
● Lower power losses in the power transmission cables. This is due to the fact
that only two conductors are required in HVDC distribution, whereas three
conductors are required in AC distributions. The HVDC system shows about
67% reduction compared with the conventional AC system (115 V CF) and
about 33% compared with the HVAC system.
● The reduction of skin effect in HVDC can also reduce the power losses and
also significantly reduce the dielectric losses in the power cables.
● Less corona effect in DC compared with AC conductors.
● No need for any reactive power compensation equipment: the capacity of wires
and devices can be reduced.
Figure 11.4 shows a HVDC power distribution system for future modern aircrafts.
270 V DC bus is powered by two generators via power electronic converters. ECSs
that maintain the pressure/temperature of the cabin and wing ice protection system
(WIPS) are also connected to the DC bus. The essential DC bus is used for the
flight critical actuation systems, represented by several permanent magnet machine
driven electromechanical actuators (EMA1 to EMA6).
Further analysis leads to the next identified radical change: one can consider
parallel operation of sources that was not allowed in previous EPS design
115 VAC bus
AC
Power electronic loads 28 V BAT
converter 1
G1
WIPS1 28 V BAT
EMERG LV
ECS1
Emergency
loads
270 VDC
PM EMA6 28 V
ESS
270 VDC loads
bus
28 V ESS
Power electronic
PM EMA2
converter 2
G2 ECS2
PM EMA1
WIPS2
115 VAC
bus
AC
loads
Figure 11.4 A single line diagram of an aircraft DC microgrid

approaches mainly for safety reasons. However, if the approach is properly estab-
lished, one can assume that the appropriate safety requirements can be met in
future. In return, sources paralleling idea will lead to (1) possibility of power
sharing among different sources with minimization of total weight of generation
systems, (2) ease of energy management and power flow control, (3) convenient
integration with power sources of different nature (fuel cells, energy storage devi-
ces, batteries, etc.), (4) improved availability of electrical energy onboard.
11.2.3 Power utilization

After the electrical power has been generated and distributed throughout the air-
craft, it is wide to the aircraft services, i.e., power utilization. These electrical
services/loads cover a range of functions: lighting services, heating devices, motors
and actuation, subsystem controllers and avionics systems. Figure 11.5 shows the
power consumption of onboard loads in an aircraft.
As it is known to all, the availability of adequate lighting is of great importance
to the safe operation of the aircraft. Lighting systems consist of internal lighting
and external lighting. Internal lighting includes cockpit/flight deck, passenger ser-
vice and emergency/evacuation lighting. On the other hand, external lighting
includes navigation, strobe, landing/taxi, formation, inspection, emergency eva-
cuation and logo lights. The heating load represents the power needed for WIPS.
Motors used onboard provide the linear actuation (engine control); rotary actuation
(position actuations for flap/slat operation); valve operation (electrical operation of
fuel, hydraulic, air and ancillary systems control valves); engine starter: starting of
the engine, APU and others to reach self-sustaining operation; gyroscope devices:
provision of power to run gyroscopes for flight instruments and auto-pilots.
Fuel Power
pumps 10 kW+ consumption
Flight in an aircraft
controls 3–40 kW
Landing
gear 25–70 kW
Air conditioning
4 × 60 kW+
(ECS)
250 kW+
WIPS
Loads
Figure 11.5 Power consumption of loads onboard in an aircraft

11.2.4 Energy storage system

With the improvement of the system reliability and the potential to reduce the size
of the main generators, the installation of energy storage system (ESS) is essential
in the MEA [8]. ESS can balance the power between sources and loads [9], improve
the transient performance and stability [10] and provide the backup electrical
power in the case of a main power fault [11]. A corresponding coordinated control
strategy needs to be developed dealing with the power balance between main
generators and ESS. When the load is within the capacity of the main generators,
they will supply power to feed the load, while the ESS can work in charging mode
or ‘‘offline’’ mode. When the load exceeds, the generator capacity under certain
scenarios, together with main generators ESS, works in discharging mode and
provides power to the load. If more than one set of ESS is installed onboard, the
ESS with higher state of charge should provide more power to the load. In this
sense, optimal-coordinated control methods using dynamic consensus algorithm,
fuzzy logic or droop principle can be developed.
11.3 Power quality requirements in aircrafts

There are several stands to specify the electrical system in aircrafts, such as ISO-
1540 Aerospace—Characteristics of aircraft electrical systems [12], DO-160F
Environmental conditions and test procedures for airborne equipment [13], MIL-
STD-704 Aircraft Electrical Power Characteristics [14], etc.
According to DO-160F, current harmonic limits for single phase and balanced
three phase systems are shown in Table 11.2. I1 is the maximum fundamental
current.
To deal with the AC/DC power conversion, conventional TRU [15] is suffering
from poor power factor and power quality. The 12- or 18-pulse ATRU is used in
Table 11.2 Current harmonic limits for single phase and balanced three phase
system in aircrafts [13]
Harmonic order Limits

Single phase Three phase
h¼3 Ih ¼ 0.15I1/h Ih ¼ 0.02I1
h ¼ 5, 7 Ih ¼ 0.02I1
h ¼ 11 Ih ¼ 0.1I1
h ¼ 13 Ih ¼ 0.3I1/h Ih ¼ 0.08I1
h ¼ 17, 19 Ih ¼ 0.04I1
h ¼ 23, 25 Ih ¼ 0.03I1
h ¼ 29, 31, 35, 37 Ih ¼ 0.3I1/h
Odd triplen harmonics (h ¼ 9, 15, 21, . . . , 39) Ih ¼ 0.15I1/h Ih ¼ 0.1I1/h
h ¼ 2, 4 Ih ¼ 0.01I1/h Ih ¼ 0.01I1/h
Even harmonics h > 4 (h ¼ 6, 8, 10, . . . , 40) Ih ¼ 0.0025I1 Ih ¼ 0.0025I1
order to meet the power quality requirement. Recent research showed that, to meet
the low input harmonic current requirements, the converters with active front-end
are attractive solutions for interfacing with the AC bus [16]. However, to effec-
tively control the input current harmonics to meet stringent specifications remains a
major challenge. It would be more difficult to meet the entire requirement simul-
taneously without reducing the other performances, such as converter’s voltage
transfer ratio, input power factor and robustness against unbalanced and/or
distorted voltage input. On the other hand, the requirements also present a major
design constraint for the filter design of the converters. A relatively small
second-order input filter is typically employed in the converters to smooth out the
switching ripple current at the power input lines. Alternative topologies for active
front-side control and input power filter designs to further reduce the weight and
size are also a big challenge.
MIL-STD-407F, US Military Standard, defines a standardized power interface
between a military aircraft and its equipment and carriage stores, covering such
topics as voltage, frequency, power factor, electrical noise and abnormal conditions
(overvoltage and undervoltage), for both AC and DC systems. Figure 11.6 shows
the limit for 115 V AC, 400 Hz and 270 V DC, 28 V DC. The envelope of the
nominal voltage transients is shown in Figure 11.6. It can be seen that in 270 V DC
system, the tolerable steady-state range is between 250 and 280 V, which indicates
the possibility of droop control (will be discussed in the following section).
The voltage at the transient can go up to 330 V and down to 200 V. Similarly, the
steady-state limit of 28 VDC system is 22–29 V. The overvoltage and under voltage
for 270 and 28 V DC system are illustrated in Table 11.3.
The high fundamental frequency (400 Hz in CF EPS and 360–800 Hz in
VF EPS) is also a challenge for aircrafts. Achieving low input current distortion and
unity input power factor at such high line frequencies require much wider control
bandwidth compared to the industry 50/60 Hz systems. Moreover, VF EPS will
350 330 50
50
300
280
DC voltage, V
0.0125 s
DC voltage, V
40
250
250 29
30
0.0825 s
0.015 s 22
200 20 18
150 10
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.0025 0.05 – 0.075 0.1 0.125 0.15 0.175
(a) Time, s (b) Time, s
Figure 11.6 Envelope of normal voltage transients for (a) 270 V DC system and
(b) 28 V DC system
Table 11.3 Voltage limits for 270 and 28 V DC system in aircraft DC MGs
Steady-state limit Limit for under Limit for Transient high

voltage overvoltage voltage level
270 V [250,280] 240 290 350 (t < 0.05 s)
28 V [22,29] 20 31.5 50 (t < 0.05 s)
have more challenges for power factor correction (PFC). It is because that, the
cable impedance will vary with the AC frequency in VF EPS. As a result, the PFC
needs to provide reactive power varying with the frequency. Another reason is that,
because the frequency range of VF EPS is typically 360–800 Hz, most of which is
higher than CF, the voltage drop of cables in VF EPS is higher than in the system of
CF EPS.
11.4 Aircraft starter/generator control

Electric starter/generator is beneficial to facilitate electric engine crank, assist to
the engine in reaching the desired idle speed, generate power to drive the onboard
auxiliaries and improve the overall system efficiency [17]. To ensure that an ade-
quate supply of electrical power is available under all operational conditions,
including those conducted in the presence of failures, the B787 is equipped with
four VF starter generators of 250 kV A capacity each. Figure 11.7 shows the
architecture of the electric starter/generator system. In the motoring mode, the
machine is driven by a bi-directional power converter and uses the power coming
from the supply bus to accelerate the aircraft engine. Since the control target in the
motoring mode is the machine speed, as it can be seen, the speed signal is sensored
and the control system regulates the machine speed to its reference value. This
mechanical energy can be harnessed to the engine shaft and cause the engine to
spool up to a speed where the combustion ignites and the engine becomes self-
sustaining. In the generating mode, the engine acts as a source of mechanical power
that is converted by the machine and converter into the electrical power to supply
the onboard loads.
The University of Nottingham contributes to ‘‘Aircraft Electrical Generation
with Active Rectification Technology’’ (AEGART) project, a part of the Clean Sky
JTI ITD EU FP7 framework program [18]. The project has conducted an intensive
trade-off study of potential S/G topology candidates and, due to the advantages of
the permanent magnet machine (PMM), the topology of surface-mounted PMM fed
by an active front end (AFE) rectifier has been chosen. As shown in Figure 11.8,
fundamental vector control is used. After transforming measured currents to
rotating frame, the linear control (conventional PI controller) adjusts current in the
synchronously rotating reference frame (dq domain) and outputs dq voltage
demands. Voltage demands are inversely transformed into three-phase demand
VAC,
Control constant frequency
system
Speed
VAC Control
variable frequency signals
Bus
Load
Bi-directional
Aircraft Starter/
power Load
engine generator Variable converter
frequency Load
Variable speed
Figure 11.7 Electric starter/generator system including loads and control
VC Idc
Cdc
SM PMM DC bus
Iabc mabc Edc

q abc Modulation
dq
abc
wm we ym Ed
dq
p/2
Vdref
L Vqref
–
–
Iqref
PI
VC mag
Idref –
PI
Figure 11.8 Inner current loop control of the PMM-based S/G [19]
modulation indexes for pulse width modulation. Then the core system can be fully
controlled by using both dq current demands.
Figure 11.9 shows the outer loop of the S/G. Since the designed PMM is for
high speed operations, the flux weakening controller needs to be taken into account
when the machine speed exceeds the base speed. For the starter mode, speed con-
troller and flux weakening controllers are employed to generate the q-axis current
reference and d-axis current reference. When the stator voltage is less than its
Flux weakening
VC max controller
Idref
1/√3
PI
VC –
wm Inner loop
Speed
controller max = √Imax–Id
Iqref 2 ref 2 as shown in
Figure 13.8
wmref –
PI S Iqref
G
Imax Idcref – Iq max TSR
Edc
–Imax PI limiter
Idc
Idc
controller
Figure 11.9 Outer loop control for Aircraft Electrical Generation with Active
Rectification Technology (AEGART) S/G system
Figure 11.10 Aircraft Electrical Generation with Active Rectification Technology

(AEGART) starter/generator system: (a) 45 kW 32 krpm PMM
starter/generator and (b) bi-directional power converter
maximum value, the output of the flux weakening controller is saturated to zero.
The controller starts to work when the stator voltage exceeds the maximum voltage
limit. With the outer voltage regulation loop (Vc loop), the flux-weakening control
is fulfilled by injecting the negative d-axis current. For the generator mode, the
conventional PI controllers are used to deflux the machine (d-axis) at high speed
and control the output DC current (q-axis). The stator current references in the d
and q axes are obtained from the output of the flux weakening controller and IDC
controller respectively. The reference of the AC voltage (Vc) is dependent on the
ref
DC voltage. The DC current reference (IDC ) is determined by the desired droop
characteristic.
Figure 11.10 shows a lab prototype for the electric S/G developed in Not-
tingham [20].
11.5 Control strategies in aircraft DC microgrids

A ‘‘single-bus’’ EPS architecture with multiple generators feeding to the same bus
is illustrated in Figure 11.11. The expected benefits of using parallel generation
with single bus include improvements in power system redundancy, i.e., the
parallel system can provide ‘no break’ power in the event of a generator failure.
As shown in Figure 11.11, two generators are connected directly to the turbine
shafts through active rectifiers and can operate in variable speed mode. The two
generators take power from the main engine through the high pressure and low
pressure shafts, guaranteeing an efficient exploitation of the power generated by the
engine, and operate in parallel to transfer the power to the DC bus. Normally, this
270 V DC power can be obtained from the main engine-driven generators or
through active rectification. Alternatively for ground starting of the engine, the
270 V could also be obtained from a ground-based APU.
In an example aircraft DC MG shown in Figure 11.11, the sources in a dis-
tributed control scheme operate cooperatively to regulate the bus voltage, but a load
sharing problem arises, where each source must provide power to the load pro-
portional to its power capacity. In such context, load sharing is critical; to avoid
that, some sources become overloaded, losing the reliability of distributed power
systems. Appropriate power sharing among the sources is of importance in multi-
source configuration.
Control strategies applied in utility DC MGs can be employed in aircraft DC
MG as well. Following the definition of hierarchical control in [21], the level of
control can be composed of primary control (inner control loop and primary power
sharing control), secondary control (voltage and frequency restoration) and tertiary
control (optimal load dispatch and system operation) which are explained in the
followings.
Power
Power electronic
G1 electronic
converter 1 HVDC bus G3
converter 3
Cable 1
Cable 3
C1
Power electronic C3
G2 converter 2 RL
Cable 2
Convention
C2 al loads
Constant power Power

CPLs electronic
loads Energy
converter 4 storage
Eb Cable 4
C4
Battery-converter
system
Network
Figure 11.11 A single DC-bus based aircraft DC microgrid

11.5.1 Primary control

For the voltage-source-converter based DC MGs, the basic droop control concept
can be implemented either as current/power mode droop, including current–voltage
(I–V) and power–voltage (P–V) strategies or as voltage mode droop, including V–I
and V–P strategies [18].
The basic concept of droop control, as discussed above, is implemented by
adding a ‘‘virtual resistance’’ into the existing local voltage controllers for the
sources, enabling their parallel operation on common load(s). Due to the mod-
ularity and reliability, droop control has been successfully applied in DC systems.
As discussed in previous sections, AFEs will be widely used in aircrafts rather
than TRU or ATRU due to the power quality requirements. In the AFEs-based DC
EPS, the basic droop control concept can be implemented either as current/power
mode droop, including I–V and P–V strategies or as voltage mode droop, including
V–I and V–P strategies [22]. The I–V and P–V droop methods are shown in Fig-
ure 11.12(a) and (b) [23–25]. In the implementation of these control methods, the
DC voltage is measured and the injected current/power is controlled according to
the droop characteristic. Alternatively, for the V–I and V–P methods shown in
Figure 11.12(c) and (d), current or power is measured and the DC voltage is
regulated accordingly [25].
Figure 11.13 shows the voltage current/power characteristics for both power
and current controlled terminals. It is obvious that both models are almost
equivalent with little difference when small voltage errors are considered. In [26],
I–V and P–V demonstrated similar performance, and for this reason only I–V droop
is considered in this chapter as a representative of the current/power mode strate-
gies. Similarly, only V–I droop is investigated within the voltage mode methods.
I–V droop P–V droop

characteristic I0 I0
characteristic
I* Io * P* *
V dc V dc P
V dc V dc
V V
V0 V0
{ {
(a) (b)
V–P droop
V–I droop I0
I0 characteristic
characteristic
*
V *
V* * P V dc
Io V dc V dc
V dc P
I
(c) (d)
Figure 11.12 Droop characteristic employed in active front ends (AFEs). (a) I–V
droop, (b) P–V droop, (c) V–I droop, (d) V–P droop [19]
1 1
0.8 0.8
Voltage
Voltage
0.6 0.6
0.4 0.4
V–I curve for P–V droop V–P curve for I–V droop
0.2 0.2
-----Equivalent I–V droop -----Equivalent P–V droop
–1 –0.5 0 –0.5 1 –1 –0.5 0 –0.5 1

Current Current
(a) Power voltage droop (b) Current voltage droop
Figure 11.13 Similarity validation of P–V droop and I–V droop [26] in DC MGs.
(a) Power voltage droop; (b) Current voltage droop
V–I droop Idc

characteristic
Vdc controller
V
Idc Vdcref Iqref Inner loop
I PI Figure 13.8 Vdc
Vdc
Figure 11.14 Voltage-mode droop control scheme of a single AFE module
The voltage-mode approach employs the V–I droop characteristic which uses
the measured branch current to generate the terminal voltage reference. The control
scheme of the voltage-mode droop-controlled AFE is shown in Figure 11.14. As
expressed in (1), the DC voltage reference is generated according to the branch
output DC current using the V–I droop characteristic.

VDC ¼ Vo kIDC (11.1)
The current-mode approach uses the measured voltage to calculate the desired
injecting DC current. The current-mode droop control scheme is shown in
Figure 11.15, with the current reference derived from the I–V droop characteristic,
based on the DC voltage measurement:
Vo VDC
IDC ¼ (11.2)
k
where Vo is the nominal bus voltage; k is the droop gain; VDC is the DC voltage

measurement; IDC is the generated DC current reference.
Figure 11.16(a) and (b) shows the primary control for current-mode and
voltage-mode droop controlled systems, respectively.
I–V droop Idc
characteristic
Idc controller
I
Vdc Idcref Iqref Inner loop
V PI Figure 13.8
Vdc
V0 Idc —
Figure 11.15 Current-mode droop control scheme of a single AFE module
DC bus
Module 1 I1
Cable
Idc controller
Vo I1ref Inner current
1/k1 PI ref
loop V1
I1 Iq1 Cable
Module n In
Cable
Idc controller
Inref Inner current
1/kn PI loop Vn
Vo In Iqnref Cable
{
(a)
DC bus
Module 1 I1
Cable
Vdc controller
ref Inner current
Vo V1
PI
loop V1
V1 Iq1ref Cable
k1
{
Module n In
Cable
Vdc controller
Vo Vnref Inner current
PI loop Vn
Vn Iqnref Cable
kn
{
(b)
Figure 11.16 Primary control for AFEs-based aircraft DC MGs (a) current-mode
droop control and (b) voltage-mode droop control
11.5.2 Secondary control

As discussed in [27], there is a trade-off between droop coefficients and voltage
regulation. High droop gain can guarantee the precise power sharing among the
sources, while the voltage regulation performance is poor, i.e., voltage drop is high
under high droop gains.
As abovementioned voltage profile in 270 V DC system (see Section 11.4), the
rated voltage of the main bus is 270 V, but the acceptable steady-state range is
between 250 and 280 V as depicted in the standard MIL-STD-704F.
In order to maintain the bus voltage for a droop-controlled aircraft DC MG
at the nominal value, a conventional method is to employ a secondary control to
compensate the voltage drop. The principle of the secondary control in DC MGs
is to restore the voltage drop due to the droop characteristic, as shown in Fig-
ure 11.17. When the primary control is implemented, the system operating point
will move from Vo (no load voltage) to OP1 under iDC1 load condition and from
Vo to OP2 under iDC2 load condition, respectively. After activating the secondary
control, the operating point will move from OP1 to OP1_new and OP2 to
OP2_new, i.e., the system always work at nominal voltage level.
Similar to the droop characteristic of each individual module at the DC term-
inal, the main bus of the multi-source system also has the droop characteristic for
the DC bus voltage and total load current. The global droop gain, which defines the
main bus V–I characteristic, will be derived as given and shown in Figure 11.18.
vdc
OP 1_ne w
OP2_new
vo
Secondary
Primary OP1
OP2
idc1 idc2 idc
Figure 11.17 Principle of secondary control on DC voltage restoration in

DC MGs
V1 I1
Cable
Vb
Ii IL
Cable
Vn In
Load
Cable
Figure 11.18 An equivalent circuit of a single bus aircraft DC MG

As one can see from Figure 11.18, at steady state, the bus voltage can be
expressed as follows:
V b ¼ V o I 1 ðk 1 þ R 1 Þ ¼ V o I 2 ðk 2 þ R 2 Þ ¼ ¼ V o I n ðk n þ R n Þ
(11.3)
The total load current can be written as
X
n
1
I L ¼ I 1 þ I 2 þ þ I n ¼ ðV o V b Þ (11.4)
i¼1
ki þ R i
Then (11.3) can be reformatted to

1
V b ¼ V o IL P n (11.5)
i¼1 ð 1=ðk i þ R i ÞÞ
and the main bus voltage-current characteristic can be formulated as a linear line
characterized by the droop gain as
1
k t ¼ Pn (11.6)
i¼1 ð 1=ðk i þ R i ÞÞ
The relationship between global droop gain and individual droop gain is depicted in
Figure 11.19. It can be seen that the global droop slope kt is stiffer than individual
droop gain ki. In other words, under the same load power, the voltage drop at the
main bus with multi-source operation is smaller than the voltage deviation under
the single source operation scenario.
In [27], a three-level control structure is proposed so that the load sharing and
voltage regulation can be achieved simultaneously. Secondary control is imple-
mented via a centralized voltage controller to solve the voltage deviation owing to
primary droop control. References [27–33] present several distributed secondary
control approaches in DC MGs. Reference [34] proposed a secondary control
method for DC bus voltage restoration without the need of communication line and
additional controllers.
According to the droop control approach, the DC bus voltage will reduce
according to the increase of the load power/current. The method restores the
terminal voltage to its nominal value autonomously. A feedforward term DV is
V0 kt
ki
k1 k2
I
Figure 11.19 Relationship between global droop gain and individual droop gain
in DC MGs
added to the terminal voltage reference for each module which can be expressed as
follows
DV ¼ ILt kt (11.7)
where ILt is the total load current and kt is the global droop gain. It is easily
implemented since only the load current needs to be measured and no DC voltage
controller is in need. If multiple loads are in presence and distributed among the
EPS, the total load current measurement can be obtained at the main feeders (if this
exists) which supplies the power to all loads.
However, the application is limited to small-scale DC MGs, i.e., the converters
need to be close to the main bus and an additional current sensor need to be
installed at the main bus to measure the total load current. It is not applicable for
large-scale MGs, where the converters and loads are distributed among the net-
work, and it is not practical to install a current sensor at the main bus for the total
load measurement. In this sense, a communication line is still required for the
information exchange, as shown in Figure 11.20.
11.6 Stability analysis
Control and stability are two important topics that interest researchers in aerospace
engineering. Typical loads within aircraft DC microgrids are tightly controlled by
power electronic converters and can often behave like constant-power loads
(CPLs).
Figure 11.21 shows an example of a tightly controlled motor drive system,
which behaves as a CPL. A DC/AC inverter drives an electric machine and tightly
regulates the rotating speed. Therefore the speed (w) is almost constant under this
rigorous regulation. Since the rotating load has a one-to-one torque–speed char-
acteristic, for every speed, there is only one corresponding torque. Hence, for a
given constant speed (w), torque (T) is constant and hence power (the multi-
plication of speed and torque) is also constant. Thus, the tightly controlled DC/AC
inverter and its associated motor drive present a CPL characteristic.
CPLs have negative incremental impedance characteristics which can lead to
interaction problems among subsystems, deteriorating system stability and also
power quality [35,36]. Thus, this undesirable destabilizing effect induced by CPLs
is a great concern for the MEA EPS. The candidate architecture should be carefully
examined for stability in order to guarantee safe EPS operation for a wide range of
operation scenarios.
Two types of stability analysis can be remarked here: large signal and small
signal analysis. Large signal stability analysis investigates the ability of a power
system to maintain stable operation and reach acceptable steady-state operating
points when subjected to large disturbances. The Lyapunov-based method has been
popular in terms of large signal stability analysis, where non-linear mathematical
modelling is used instead of small signal linearization [37,38]. The main advantage
of Lyapunov-based approaches is that a developed Lyapunov function allows the
DC bus
ILt Module 1
I1
kb Cable
Vdc controller
Vo V1ref Inner
PI current loop V1
Iq1ref Cable
Droop k V1
1 {
ILt Module 2 I2
kb Cable
Vdc controller
ref Inner
Vo V2
PI ref
current loop V
Iq2 2 Cable
Droop k2 V2
{
ILt Module n In
kb Cable
Vdc controller
ref Inner
Vo Vn
PI current loop Vn
Iqnref Cable
Droop kn Vn
{
(a)
DC bus
n Module 1
idc1 ∑ idcj ILt kb Vdc controller
I1
Cable
j=1
Vo V1ref Iq1ref Inner
PI current loop V1
idcn Cable
Droop k1 V1
{
Communication line
idc1
n Module 2
∑ idcj ILt kb Vdc controller
I2
Cable
Cable
j=1
idcn Vo V2ref Iq2ref Inner
PI current loop V2
Cable
Droop k2 V2
{
idc1
n
ILt k Module n In
∑ idcj b
Vdc controller Cable
idcn
j=1
Vo Vnref Iqnref Inner
PI current loop Vn
Cable
Droop kn Vn
{
(b)
Figure 11.20 Secondary control approach in aircraft DC MGs with multiple

controllers (a) without communication link and (b) with
communication link
Speed
Speed Regulator to
measurement
reference control the
speed
DC bus
ω
DC/AC Motor
inverter
TL = constant
ω = constant
Pin = constant
P = ω TL = constant
Figure 11.21 A typical presentation of CPLs in aircraft MGs (motor drives)
Vbus
Source Load
Vin Vo
converter converter
Zo_s Zin_L
Figure 11.22 Impedance-based method to assess system stability
estimation of the attraction region of a stable operating point, thus providing an

estimate of the size of disturbance that can be tolerated [39–42]. On the other hand,
small signal stability ensures that the system can revert to its original equilibrium
point when subject to a small signal disturbance. Thus, a linearized model around
the system equilibrium point is used to study the stability. Linear analysis tools
(Bode diagram, Nyquist diagram and Routh–Hurwitz criterion) can then be
employed to derive local stability conditions. These approaches are relatively easy
and familiar to applied engineers. Two dominant approaches for small signal sta-
bility analysis can be remarked on here. The first option is to consider the stability
problem with a general state space model, design the control and evaluate the
impact of different parameters by studying eigenvalues.
Alternatively, the impedance method, which is based on an analysis of the
output impedance of the source and equivalent load impedance, has also been
widely used. The contribution of each component to the system’s stability can be
assessed in the frequency domain. In addition, the output impedance of the com-
ponents or subsystems can be reshaped to stabilize the overall power system using
an active damping method, an impact of connection of new converters that can be
easily assessed. For these reasons, the impedance-based method has been suc-
cessfully applied in power system stability analysis [43,44]. Models of individual
components are assembled into a full system model which is then typically broken
down into two subsystems at a DC point of interest, i.e. a load subsystem and a
source subsystem, as shown in Figure 11.22. Consequently, analytical expressions
are derived for input impedance of the load Zin_L and output impedance of the
source Zo_s subsystems. The input of the full system is Vin and output is Vo. The
voltage at the breaking point for the subsystem is Vbus. The definition of the source
converter is the converter which regulates the bus voltage, and the load converter is
responsible of current control.
The total input-to-output transfer function of the cascaded subsystems can be
written by
Zin L Vin 1
Vo ¼ Vin ¼ (11.8)
Zin L þ Zo s Zin L 1 þ TMLG
where the ratio Zo_s/Zin_L, which is often referred to as the minor loop gain, TMLG
can be given by
Zo s
TMLG ¼ (11.9)
Zin L
If each of the two subsystems is individually properly designed with good dynamic
performance, the influence of their interaction can then be studied by looking into
the minor loop gain. In particular, in order to preserve the stability, it is mandatory
that minor loop gain meets the Nyquist stability criterion [45]. It should also be
noted that if the detailed information about source and load systems is not available
and the respective impedances cannot be analytically constructed, then they should
be measured online. Since the source and load converter are assumed to be
stable standalone, the minor loop gain term in (11.9) is the one responsible for
stability.
In the specific context of aircraft power systems, so far, several publications
have discussed the system stability in the MEA electric power systems. The sta-
bility of a switched reluctance motor-based 270 V DC power system has been
analysed in [46]. A permanent magnet synchronous generator-based aircraft DC
MG is investigated in [47–49], and the influence of parameter variation on system
stability is presented. Components (R, L, C parameters), operating points (voltage,
power) and control parameters (controller gains, droop gains, etc.) will influence
the system stability. Take the droop gain as an example, the Bode plot of the source
and load impedance in the voltage-mode droop-controlled system is shown in
Figure 11.23. The source impedance magnitude in the low-frequency range
increases with an increase in the droop gain, whilst the load impedance magnitude
decreases. Therefore, the droop gain not only affects the power sharing ratio but
also the system stability.
11.7 Chapter summary
This chapter provides an overview of architecture, power quality, control and sta-
bility in aircraft DC microgrids. First the evolution of aircraft DC microgrids
including power generation, distribution, utilization and energy storage is
reviewed. Then the chapter presents the power quality requirement in aircraft DC
microgrid including the AC current harmonics and DC voltage envelope. As an
Bode diagram
40
Load impedance
20
Magnitude (dB)
0
k=2
–20 k=1
k = 0.5
–40
Source impedance
–60
360
Source impedance
Phase (deg)
0
Load impedance
–360 –2
10 10–1 100 101 102 103 104
Frequency (Hz)
Figure 11.23 Bode diagram of source/load impedances with different values of

droop gain using the voltage-mode droop control approach
important part in aircraft power system, electric starter/generator system is intro-

duced and corresponding control in starter/generator mode is presented. In addition,
hierarchical control in the aircraft DC MG is presented. Since the bus voltage
envelop allows some voltage deviations, it provides the possibility of the voltage
droop control in a multiple source-based aircraft DC MGs. Primary control deals
with the inner current/voltage regulation and preliminary droop control. To over-
come the potential issue in the conventional droop control, distributed/decen-
tralized secondary control deals with the DC voltage restoration and improves the
power sharing performance in the meantime. At last, due to the presence of the
CPLs, stability is a concern in the aircraft DC MGs. To design a stable aircraft DC
MGs over a wide range of operation scenarios, eigenvalue-based and impedance-
based stability analysis methods have been discussed.
Acknowledgement
The authors acknowledge the support from CleanSky JTI Projects, FP7 European
Integrated Projects-http://www.cleansky.eu.
References
[1] P. Wheeler and S. Bozhko, ‘‘The more electric aircraft: technology and
challenge,’’ IEEE Electrif. Mag., vol. 2, no. 4, pp. 6–12, Dec. 2014.
[2] P. E. Gartz, ‘‘Commercial systems development in a changed world,’’ IEEE

Trans. Aerosp. Electron. Syst., vol. 33, no. 2, pp. 632–636, Apr. 1997.
[3] A. Eid, H. El-Kishky, M. Abdel-Salam, and T. El-Mohandes, ‘‘Modeling and
characterization of an aircraft electric power system with a fuel cell equip-
ped APU paralleled at main ac bus,’’ in Proc. IEEE Conf. Int. Power Mod-
ulator High Voltage, 2010, pp. 229–232.
[4] X. Roboam, B. Sareni, and A. D. Andrade, ‘‘More electricity in the air:
toward optimized electrical networks embedded in more-electrical aircraft,’’
IEEE Ind. Electron. Mag., vol. 6, no. 4, pp. 6–17, Dec. 2012.
[5] R. Abdel-Fadil, A. Eid, and M. Abdel-Salam, ‘‘Electrical distribution power
systems of modern civil aircrafts,’’ in Proc. 2nd International Conference on
Energy Systems and Technologies, Cairo, Egypt, Feb. 2013, pp. 1–10.
[6] M. Olaiya and N. Buchan, ‘‘High power variable frequency generator for
large civil aircraft,’’ in Proc. IET Elect. Mach. Syst. More Elect. Aircr.,
1999, pp. 1–4.
[7] B. H. Nya, J. Brombach, and D. Schulz, ‘‘Benefits of higher voltage levels in
aircraft electrical power systems,’’ in Proc. Electrical Systems for Aircraft,
Railway and Ship Propulsion (ESARS), Bologna, Oct. 2012, pp. 1–5.
[8] F. Gao, S. Yeoh, S. Bozhko, and G. Asher, ‘‘Coordinated control of a DC
electrical power system in the more electric aircraft integrated with energy
storage,’’ in ECCE 2015, IEEE Energy Conversion Congress & Expo,
Montreal, Canada, Sep. 2015, pp. 5431–5438.
[9] H. Zhang, F. Mollet, C. Saudemont, and B. Robyns, ‘‘Experimental valida-
tion of energy storage system management strategies for a local DC
distribution system of more electric aircraft,’’ IEEE Trans. Ind. Electron.,
vol. 57, no. 12, pp. 3905–3916, Dec. 2010.
[10] K. Jong-Yul, J. Jin-Hong, K. Seul-Ki, et al., ‘‘Cooperative control strategy
of energy storage system and microsources for stabilizing the microgrid
during islanded operation,’’ IEEE Trans. Power Electron., vol. 25, no. 12,
pp. 3037–3048, Dec. 2010.
[11] Y. Guan, L. Meng, C. Li, J. Vasquez, and J. Guerrero, ‘‘A dynamic con-
sensus algorithm to adjust virtual impedance loops for discharge rate
balancing of AC microgrid energy storage units,’’ IEEE Trans. Smart Grid,
To be published.
[12] Characteristics of aircraft electrical systems. ISO-1540, 2006.
[13] Environmental conditions and test procedures for airborne equipment.
DO-160F, 2010.
[14] Aircraft Electric Power Characteristics, American Military St. MIL-STD-
704F, Mar. 2004.
[15] T. Yang, S. Bozhko, and G. Asher, ‘‘Functional modeling of symmetrical
multipulse autotransformer rectifier units for aerospace applications,’’ IEEE
Trans. Power Electron., vol. 30, no. 9, pp. 4704–4713, Sep. 2015.
[16] X. Zheng, F. Gao, H. Ali, and H. Liu, ‘‘A droop control based three phase
bidirectional AC-DC converter for more electric aircraft applications,’’
Energies, vol. 10, no. 3, pp. 1–17, 2017.
[17] B. S. Bhangu and K. Rajashekara, ‘‘Electric starter generators: the integra-

tion into gas turbine engines,’’ IEEE Ind. Appl. Mag., vol. 20, no. 2, pp. 14–
22, Apr. 2014.
[18] Cleansky. AEGART Project Summary. 2015. Available: http://cordis.
europa.eu/result/rcn/143557_en.html. Accessed: 11/11/2015.
[19] F. Gao, S. Bozhko, A. Costabeber, et al., ‘‘Comparative stability analysis of
droop control approaches in voltage source converters-based DC micro-
grids,’’ IEEE Trans. Power Electron., vol. 32, no. 3, pp. 2395–2415, Mar.
2017.
[20] F. Gao, ‘‘Decentralized control and stability analysis of a multi-generator
based electrical power system for more electric aircraft,’’ PhD dissertation,
University of Nottingham, 2016.
[21] J. M. Guerrero, M. Chandorkar, T.-L. Lee, and P. C. Loh, ‘‘Advanced con-
trol architectures for intelligent microgrids—Part I: Decentralized and hier-
archical control,’’ IEEE Trans. Ind. Electron., vol. 60, no. 4, pp. 1254–1262,
Apr. 2013.
[22] F. Gao, Y. Gu, S. Bozhko, G. Asher, and P. Wheeler, ‘‘Analysis of droop
control methods in DC microgrids,’’ in Proc. 16th IEEE European Con-
ference on Power Electronics and Applications (EPE-ECCE’ Europe),
Lappeenranta, Finland, Aug. 2014, pp. 1–9.
[23] K. Rouzbehi, A. Miranian, A. Luna, and P. Rodriguez, ‘‘DC voltage control
and power sharing in Multiterminal DC grids based on optimal DC power
flow and voltage-droop strategy,’’ IEEE J. Emerg. Sel. Topics Power Elec-
tron., vol. 2, no. 4, pp. 1171–1180, Dec. 2014.
[24] W. Wang, M. Barnes, O. Marjanovic, and O. Cwikowski, ‘‘Impact of
DC breaker systems on multiterminal VSC-HVDC stability,’’ IEEE Trans.
Power Del., vol. 31, no. 2, pp. 769–779, Apr. 2016.
[25] X. Lu, J. M. Guerrero, K. Sun, J. C. Vasquez, R. Teodorescu, and L. Huang,
‘‘Hierarchical control of parallel AC–DC converter interfaces for hybrid
microgrids,’’ IEEE Trans. Smart Grid, vol. 5, no. 2, pp. 683–692, Mar. 2014.
[26] J. Beerten and R. Belmans, ‘‘Analysis of power sharing and voltage devia-
tions in droop-controlled DC grids,’’ IEEE Trans. Power Syst., vol. 28, no. 4,
pp. 4588–4597, Nov. 2013.
[27] J. M. Guerrero, J. C. Vasquez, and L. G. D. Vicuna, ‘‘Hierarchical control of
droop-controlled ac and dc microgrids—a general approach toward stan-
dardization,’’ IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 158–172, Jan.
2011.
[28] X. Lu, J. M. Guerrero, K. Sun, and J. C. Vasquez, ‘‘An improved droop
control method for DC microgrids based on low bandwidth communication
with DC bus voltage restoration and enhanced current sharing accuracy,’’
IEEE Trans. Power Electron., vol. 29, no. 4, pp. 1800–1812, Apr. 2014.
[29] S. Anand, B. G. Fernandes, and J. M. Guerrero, ‘‘Distributed control to
ensure proportional load sharing and improve voltage regulation in low
voltage DC microgrids,’’ IEEE Trans. Power Electron., vol. 28, no. 4,
pp. 1900–1913, Apr. 2013.
[30] N. Yang, D. Paire, F. Gao, A. Miraoui, and W. Liu, ‘‘Compensation of droop

control using common load condition in dc microgrids to improve voltage
regulation and load sharing,’’ Int. J. Elect. Power Energy Syst., vol. 64,
pp. 752–760, Jan. 2015.
[31] P. Wang, X. Lu, X. Yang, W. Wang, and D. Xu, ‘‘An improved distributed
secondary control method for DC microgrids with enhanced dynamic current
sharing performance,’’ IEEE Trans. Power Electron., vol. 31, no. 9,
pp. 6658–6673, Sep. 2016.
[32] T. Dragicevic, J. C. Vasquez, J. M. Guerrero, and D. Skrlec, ‘‘Advanced
LVDC electrical power architectures and microgrids: a step toward a new
generation of power distribution networks,’’ IEEE Electrifi. Mag. vol. 2,
no. 1, pp. 54–65, Mar. 2014.
[33] X. Lu, J. M. Guerrero, K. Sun, and J. C. Vasquez, ‘‘An improved droop
control method for dc microgrids based on low bandwidth communication
with dc bus voltage restoration and enhanced current sharing accuracy,’’
IEEE Trans. Power Electron., vol. 29, no. 4, pp. 1800–1812, Apr. 2014.
[34] F. Gao, S. Bozhko, G. Asher, P. Wheeler, and C. Patel, ‘‘An improved vol-
tage compensation approach in a droop-controlled DC power system for the
more electric aircraft,’’ IEEE Trans. Power Electron., vol. 31, no. 10,
pp. 7369–7383, Oct. 2016.
[35] A. Emadi, A. Khaligh, C. Rivetta, and G. Williamson, ‘‘Constant power
loads and negative impedance instability in automotive systems: definition,
modeling, stability, and control of power electronic converters and motor
drives,’’ IEEE Trans. Veh. Technol., vol. 55, no. 4, pp. 1112–1125, Jul. 2006.
[36] A. Rahimi and A. Emadi, ‘‘Active damping in DC/DC power electronic
converters: a novel method to overcome the problems of constant power
loads,’’ IEEE Trans. Ind. Electron., vol. 56, no. 5, pp. 1428–1439, May
2009.
[37] M. Kabalan, P. Singh, and D. Niebur, ‘‘Large signal Lyapunov-based sta-
bility studies in microgrids: a review,’’ IEEE Trans. Smart Grid, vol. 8,
no. 5, pp. 2287–2295, Sep. 2017.
[38] P. Kundur, J. Paserba, V. Ajjarapu, et al., ‘‘Definition and classification of power
system stability IEEE/CIGRE joint task force on stability terms and defini-
tions,’’ IEEE Trans. Power Syst., vol. 19, no. 3, pp. 1387–1401, Aug. 2004.
[39] P. W. Sauer and M. A. Pai, Power System Dynamics and Stability. Champaign,
IL, USA: Stipes, 2006.
[40] W. Du, J. Zhang, Y. Zhang, and Z. Qian, ‘‘Stability criterion for cascaded
system with constant power load,’’ IEEE Trans. Power Electron., vol. 28,
no. 4, pp. 1843–1851, Apr. 2013.
[41] A. Griffo and J. Wang, ‘‘Large signal stability analysis of ‘more electric’
aircraft power systems with constant power loads,’’ IEEE Aerosp. Electron.
Syst. Mag., vol. 48, no. 1, pp. 477–489, Jan. 2012.
[42] D. Marx, P. Magne, B. Nahid-Mobarakeh, S. Pierfederici, and B. Davat,
‘‘Large signal stability analysis tools in DC power systems with constant
power loads and variable power loads—a review,’’ IEEE Trans. Power
Electron., vol. 27, no. 4, pp. 1773–1787, Apr. 2012.
[43] M. Cespedes and J. Sun, ‘‘Impedance modeling and analysis of grid-
connected voltage-source converters,’’ IEEE Trans. Power Electron., vol. 29,
no. 3, pp. 1254–1261, Mar. 2014.
[44] A. Radwan and Y. Mohamed, ‘‘Assessment and mitigation of interaction
dynamics in hybrid AC/DC distribution generation systems,’’ IEEE Trans.
Smart Grid, vol. 3, no. 3, pp. 1382–1393, Sep. 2012.
[45] R. D. Middlebrook, ‘‘Input filer consideration in design and application of
switching regulators,’’ in Proc. IEEE IAS Annu. Meet., Chicago, USA, Oct.
1976, pp. 336–382.
[46] L. Han, J. Wang, and D. Howe, ‘‘Small-signal stability studies of a 270V DC
more-electric aircraft power system,’’ in Proc. Power Electronics Machines
and Drives 2006. The 3rd IET International Conference (PEMD), Apr. 2006,
pp. 197–201.
[47] K.-N. Areerak, T. Wu, S. V. Bozhko, G. M. Asher, and D. W. P. Thomas,
‘‘Aircraft power system stability study including effect of voltage control
and actuators dynamic,’’ IEEE Trans. Aerosp. Electron. Syst., vol. 47, no. 4,
pp. 2574–2589, Oct. 2011.
[48] K.-N. Areerak, S. V. Bozhko, G. M. Asher, L. De Lillo, and D. W. P. Thomas,
‘‘Stability study for a hybrid AC–DC more-electric aircraft power system,’’
IEEE Trans. Aerosp. Electron. Syst., vol. 48, no. 1, pp. 329–347, Jan. 2012.
[49] F. Gao and S. Bozhko, ‘‘Modeling and impedance analysis of a single DC
bus based multiple-source multiple-load electrical power system,’’ IEEE
Trans. Transp. Electrif., vol. 2, no. 3, pp. 335–346, Sep. 2016.
Chapter 12
Shipboard MVDC microgrids
Dong-Choon Lee1, Yoon-Cheul Jeung1, Dinh Du To1,
and Duc Dung Le1
12.1 Introduction
Electric power was successfully applied to marine vessels in 1880s. Due to the
advance of technology in batteries, motors, and engines, the primitive electric
devices were installed on the ship, and in the early twentieth century the electric
propulsion system was developed [1]. The electric propulsion has a lot of advan-
tages such as better boarding comport, design flexibility, low vibration and noises,
low fuel consumption and emission, etc.
In electric propulsion systems, the propulsion motors are supplied by the dedi-
cated generators and the other loads are supplied by the auxiliary generators. To avoid
the inefficient structure of system, multiple gen-sets are connected to one common
bus, which supply propulsion and service loads together. This system is called
as integrated propulsion system (IPS) [2]. When the load power demand in the IPS is
low, some of the prime movers are operated at high output power region and the others
are deactivated, which increases the overall system efficiency. Queen Elizabeth II
launched in 1987 was the first diesel-electric ship that embeds the IPS [3].
Due to the evolution of the power electronics technology, the concept of all-
electric ship (AES) has been established, where, needless to say about propulsion
and service loads, even the loads using hydraulic and compressed air power are
replaced by electric devices. In 2015, the world’s first purely battery-driven car and
passenger ferry Ampere was launched in Norway [4].
So far, the AC power distribution system has been commonly used in ship-
board power systems since the AC motors are mostly employed for electric pro-
pulsion. Recently, however, the shipboard medium voltage DC (MVDC) power
system is attracting a great attention since it has lots of advantages such as smaller
high-frequency transformer, easy connection and disconnection, no harmonic and
unbalance problems, no need of synchronization, no reactive power, high effi-
ciency, flexible design in space, easy monitoring of power quality, fuel saving by
variable speed operation, and so on [5].
1
Department of Electrical Engineering, Yeungnam University, South Korea
In this chapter, the shipboard MVDC microgrid system is introduced. First, the
architecture and components of the system are described, and then the modeling
and control are dealt with. Next, power management based on hierarchical control
is described and then fault and protection schemes are discussed. Some simulation
results for the case study are provided.
12.2 Architecture of shipboard MVDC power systems

The shipboard DC power system consists of generation, distribution, power con-
version, and ship loads. In the IPS, the gen-sets and ship loads are connected to the
common MVDC bus through the power converters. Electric generators are driven
by gas turbines (GTs) or diesel engines (DEs), which supply power to the bus. In
large ships, multiple generators are employed to supply the high power demand. In
addition, the ESS (energy storage system), which can compensate for the fluctua-
tions in the DC bus voltage due to load variations, is integrated. The ship loads are
usually composed of propulsion, services, and pulsed power loads (PPLs). The
optional configuration of the shipboard MVDC power system is shown in
Figure 12.1.
12.2.1 Structure of DC bus

The architecture of the MVDC distribution in ships is mostly of a ring or radial
type. The radial bus is an economic structure since it can supply power to various
loads through independent power lines. At fault conditions, however, the surviva-
bility of the ship is low. To overcome this weakness, the additional energy sources
can be installed for service loads, but it is not easy in ships due to the limitation of
the space.
Meanwhile, the ring bus fed by multiple power sources can supply the power
to the loads through the common power lines. Figure 12.2 shows the ring bus
structure of the MVDC distribution system with several zones [5]. This structure
has not only an intrinsic advantage of DC distribution but also an ability to supply
power continuously to the load under a certain fault condition. Furthermore, it is
Generation Power conditioning and distribution Power conversion Loads

Prime movers Generators AC/DC rectifier DC DC/AC Ship services
distribution converter
Gas-turbine Synchronous IGBT
generator Ring bus DC/DC Propulsions
(zonal) converter
SCR Induction
Diesel-engine PMSG motor
Radial bus
Diode
PMSM
Optional energy
Fuel sources
DC/DC Passive Synchronous
Photovoltaic converter filter motor
Energy Active
Fuel cells filter Pulsed load
storage Ship to grid
Figure 12.1 Configuration of shipboard MVDC power system

Shipboard MVDC microgrids 297
PM
SG1 GT1 DE1
SG1 ESS
M1
CB CB CB
Medium voltage DC bus CB
CB CB CB CB CB CB
CB CB CB CB CB
CB
Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 CB
load load load load load
CB CB CB CB CB
CB CB CB CB CB CB
CB
CB CB CB
M2 Pulsed
PM load
DE2 GT2 SG2
SG2
Figure 12.2 Ring bus structure of shipboard MVDC power system with zonal
distribution [5]
expected that fuel consumption and emission can be reduced by up to 20% and that
footprints and weight of electrical equipment can be reduced by up to 30% com-
pared with the AC distribution systems [6].
12.2.2 Prime movers

In ships, the main electric power sources are mostly synchronous generators (SGs)
which are driven by prime movers. The GT is preferred for the high power level of
ships, but the DE is often used below a medium power level. In certain cases, both
types of prime movers can be used at the same time.
Conventionally, the prime mover in ships is operated at a fixed speed for AC
distribution. In this case, the fuel efficiency of the prime movers is low at light
loads. Even though higher fuel efficiency can be achieved by a variable speed
operation, it is difficult to apply to the shipboard AC power system since the output
voltage frequency of the generators should be limited by the standards of AC
distribution.
In the DC power system, however, there is no critical hurdle to apply the
variable speed operation scheme since the AC voltage with a variable frequency is
rectified to the DC bus voltage. It has been reported that up to 15% of the fuel is
saved in the 5.17 MW GT system by the variable speed operation [7].
12.2.3 Components of shipboard MVDC power systems

12.2.3.1 AC generators
For high power ships, a wound-field SG is usually used, whereas a permanent
magnet SG (PMSG) is applied to the low-to-medium power capacity ships.
The SG is the most popular machine of which power rating is ranged from a
few kW to tens of MW. The stator of the SG has three-phase windings and the rotor
has a field winding for DC excitation. The SG produces AC voltages at 50 or 60 Hz
frequency at the rated speed. The shaft speed is typically ranged from 600 or
720 rpm to 3,000 or 3,600 rpm, respectively [8].
The stator of the PMSG is similar to that of the SG, but the rotor has permanent
magnets instead of windings. So, the efficiency of the PMSG is higher since the
copper loss of the rotor is eliminated. Furthermore, the weight of the PMSG is
reduced by up to 30%. There is a commercial product which has 3,000 rpm and
8 MW [9].
Recently, application of superconductor SGs is gaining attention due to high
efficiency and compact size [2].
12.2.3.2 Energy storage systems

In the shipboard MVDC power system, the response of the gen-sets as a main
power source is relatively sluggish. Therefore, the DC bus voltage is vulnerable to
variations of the load, which will drop much in the case of abrupt application of
high power loads. Thus, the ESS is indispensable to supply power quickly since the
voltage drop in the DC bus causes harmful influences on propulsion and service
loads. According to the IEEE Standard 1709–2010, the bus voltage tolerances
should be 10% in steady states [5].
Warships or other electric ships have PPLs like rail guns, high power radars,
cranes, etc., which demand a high power in a short interval. The ESS needs to cover
this requirement as well as supply loads for survivability in certain cases. When
determining the power capacity of the ESS, the initial installment cost, payback
time, weight, reserve margin of single failure, optimal efficiency operation of
turbines, and power supplying capability to PPLs should be considered [10,11].
12.2.3.3 Ship loads

Ship loads can be classified into four categories: propulsion, dedicated high-power
load, ship services, and PPL [5]. The electric propulsion forms 80% of the total ship
load power. The electric propulsion needs no speed reduction gears, which uses a
fixed pitch of propeller blades. Also, it has a lot of advantages such as small
volume, low maintenance, low fuel consumption, longer life, and flexibility of
design, low emission, and better boarding comport. The dedicated high-power load
includes military radars and cranes, which consume a few MW of power. The
service load includes small pumps and fan motor drives and hotel load. The rail gun
is a kind of PPLs which consumes a few MW of power in several seconds [5].
The load power of propulsion depends on speed and sailing mode of the ship.
For instance, the load profile of liquefied natural gas (LNG) carrier consists of
unloading, loading, loaded voyage, and ballast voyage modes [12].
12.2.3.4 Power converters

In the shipboard MVDC power system, different types of power converters are
employed such as AC/DC converters, DC/DC converters, and DC/AC inverters.
● AC/DC converters
In shipboard MVDC power systems, the output voltage of the main generators is
connected to the DC bus through the AC/DC converters that are diode rectifiers or
phase-controlled converters with silicon-controlled rectifier (SCR). In the case of diode
rectifiers, if the SG is used, its output voltage is controlled by an exciter, as shown in
Figure 12.3(a). If the PMSG is used, however, the back-end DC/DC converter is
required to match the MVDC bus voltage, as shown in Figure 12.3(b). If the SCR
rectifier instead of diode rectifier is employed, the back-end converter is not necessary.
On the other hand, multilevel converters have been so far used for medium
voltage applications, in which neutral-point clamped (NPC), flying capacitor,
cascade-H bridge converters are included. Recently, the modular multilevel converter
(MMC) topology has been applied to the high voltage DC (HVDC) and MVDC
systems, as shown in Figure 12.4. Due to the modular structure, the voltage rating of
MMC can be expanded easily.
● DC/DC converters
The DC/DC converter is required for connecting the ESS to the MVDC bus. A dual
active bridge (DAB) DC/DC converter is popularly used since it has advantages of
bidirectional power flow, high efficiency, modular structure, isolation, and soft
switching. By a multiple connection of the single DAB, the power capacity of the
converter can be expanded easily, which is shown in Figure 12.5 [13].
Similarly, modular multilevel DC/DC converters are also a candidate for high
power applications [14].
● DC/AC converters
In electric propulsion ships, DC/AC converters, that is, inverters are employed to
feed the propulsion motors and other AC loads. The two-level inverter is sufficient
for small pump and fan motor drives and hotel loads. For propulsion drives,
DC-DC
Rectifier converter DC bus
Rectifier DC bus P
6-Phase
wound-field 12-Phase
P
SG PMSG
+ Prime PM
+
Prime
SG
mover
Gear − mover SG −
box
N
Exciter
N
(a) (b)
Figure 12.3 AC/DC converter for MVDC shipboard system: (a) Wound-field SG
and (b) PMSG
DC bus
SM1 SM1 SM1 + P
Upper SM2 SM2 SM2 Sj

arm
isub +
+ C
SMn SMn SMn
iap −
Vsub Sj
A L L L
B –
C L
ian L L
SM1 SM1 SM1
Submodule
Lower SM2 SM2 SM2
arm
SMn SMn SMn − N
Figure 12.4 Topology of MMC
DAB converter module 1
+ +
− − +
+
LVDC MVDC
DAB converter module 2

–
–
DAB converter module n
Figure 12.5 Modular dual active bridge DC–DC converter
cycloconverters and load-commutated inverters have been conventionally adopted

for high power AC motor drives. For the shipboard MVDC power system, however,
voltage-source inverters are appropriate since the inverter is supplied from the DC
bus. For high power AC drives, multilevel inverters can be applied. It has been
reported that the NPC inverter available commercially can be ranged up to 33 MW

[15]. In the future, the MMC topology is a good candidate since it has a lot of
advantages such as modularity, fault tolerant capability, low-voltage device, high
efficiency, and better harmonic performance [16].
● DC circuit breakers (CBs)
In AC power systems, the alternating current naturally crosses zero at every half
cycle, which creates a condition for arc self-extinction to isolate the fault. However,
due to no zero-crossing of the direct current, it is more challenging to design the
DC CB in the DC power system.
The DC CB topologies are classified into the three categories: mechanical CBs,
solid-state CBs (SSCBs) and hybrid CBs [17]. The mechanical CB has an advan-
tage of low-contact resistance, but a long opening time (5–100 ms). In the SSCB,
the insulated-gate bipolar transistors (IGBTs) and integrated-gate commutated
thyristors (IGCTs) are the key components to break the fault current with a shorter
opening time (100 ms). The SSCB has some critical issues such as the high conduc-
tion loss of power switches, the expensive cooling system, and a lot of power switches
connected in series and/or parallel for high current and voltage ratings. A combination
of the mechanical CB and the power switches results in a hybrid CB, which is featured
with low conduction loss and short opening time. When a fault occurs, the fault cur-
rent is commutated from the mechanical CB to the power switches during the fault. In
the near future, wide band-gap devices (SiC or GaN) will be used to achieve the lower
conduction loss, the higher speed fault clearance as well as the cost reduction of
cooling system due to the wider range of operating temperature.
12.3 Control of shipboard MVDC power systems
12.3.1 Gen-sets
The gen-set consists of a prime mover and a SG. A classical gen-set is operated at a
constant speed to generate a fixed electrical frequency and output voltage. Although
such a gen-set is of low cost, its fuel efficiency is low at light load conditions, and the
output voltage and frequency are fluctuated when the load varies. If a variable speed
operation scheme is applied, a significant amount of fuel saving is achieved. The
optimal speeds of DEs can be derived from the specific fuel consumption curve [18].
Figure 12.6 shows the control block diagram of the gen-set system, which
consists of a GT, a wound-field SG with an exciter, and a diode rectifier. The
governor can utilize conventional PI or lead-lag controllers to regulate the speed of
the prime mover. Recently, some advanced control techniques like sliding mode
control, fuzzy logic control, and model predictive control have been applied [19].
In the wound-field SG, the magnitude of output voltage is controlled by an exciter,
so that the DC output voltage rectified by the diode rectifier is regulated indirectly.
In the case of PMSG, the thyristor phase-controlled rectifier or the diode rectifier
with back-end DC/DC converter are usually used.
Speed
detection
+ – Gas τout Vout
ω* Governor
turbine Vdc
Synchronous Diode
generator rectifier
+ Excitation Vf Iout
Vdc* Exciter
– controller
Figure 12.6 Control structure of gen-set system
Governor Prime mover

wr max
Ky tout
w* PI e−ts
1 + tcs
min
Figure 12.7 Speed control of prime mover
To simplify the design of the speed control loop, the dynamics of prime
movers are often modeled as a first-order or second-order delay. Figure 12.7 shows
the speed control block diagram of the prime mover with PI regulators, where t is a
time delay, tc is the time constant, and Ky is the torque constant. For better control
performance, the characteristics of the engine and turbine, effects of temperature,
shaft speed, combustion, and cooling air flow need to be included in the detailed
modeling [20].
For the long-term system analysis, the generator model is effectively simpli-
fied in a series circuit of variable AC voltage source, resistance, and inductance
[21]. The exciter system is often modeled as a first-order or second-order time
delay, which is usually controlled by thyristor phase-controlled converters [22].
12.3.2 Energy storage systems

In the shipboard MVDC power system, the ESS can be used to compensate for the
fluctuations of the DC bus voltage and to manage the number of generators being
operated for fuel savings.
For the operation of the ESS, the following control strategies are considered.
The ESS controller regulates the voltage and current of the ESS during charging
and discharging operations. If multiple ESSs are used, the output powers and state
of charge (SOC) values among the ESSs can be balanced by power sharing. If the
droop control is applied for power sharing, or if the power line impedance is high,
the DC bus voltage becomes lower than the nominal voltage. Then, the control
scheme which increases the output voltage of the ESS should be applied. Finally,
the power management controller produces a power reference for the ESS [23].
12.3.3 Power management control

The power management controller should be able to ensure the energy balance
between generated powers and load power consumptions both at steady-state and
transient-state conditions. For the shipboard power system, the same hierarchical
control structure as in the conventional power grid can be applied for power
management [24].
Figure 12.8 shows a hierarchical control structure of shipboard MVDC power
system. The primary controllers regulate voltages and currents of the local power
converters. The secondary controller performs the DC bus voltage control. The
tertiary controller produces a command for power management of the whole ship
with information of generated power and load power demand.
12.3.3.1 Local control

Local (primary) control requires the fastest responses among three hierarchical con-
trol levels. The primary controller regulates the output voltage at the coupling point
of the DC bus where the power converter is connected. The voltage controller is of
PI type, which usually has an inner current control loop and selectively a feedforward
control term. Also, the power sharing can be included in this control level.
The droop control is the most popular scheme for power sharing since it needs
no communication links, which results in a simple circuit and low cost. With this
scheme, each output current of the converters can be balanced by adjusting the DC
voltage reference, which can be expressed as
vref
i ¼ vnom Rii ii (12.1)
Tertiary control
Central control
Information
of energy sources
and loads
Secondary control
Communication line
Local control
Primary Primary Primary

controller 1 controller 2 controller n
PEBB PEBB PEBB

DC bus
Figure 12.8 Hierarchical control structure of shipboard power system

where vref
i is the voltage reference for ith unit, vnom is the nominal voltage, ii is the
output current, and Rii is the droop coefficient.
The droop gain can be obtained from the maximum output current (idc_max) and
the allowable voltage variation (Dvdc), as [25]
Dvdc
Rii ¼ : (12.2)
idc max
When the gen-sets and ESSs supply power to the DC grid, each local controller
regulates its output voltage and performs the power sharing. In the case of battery
ESS, the management of the SOC can be included. If only one generator is operated
or if the generator is operated at rated power, the droop control is not applied. When
switching the control modes, the higher-level controller sends a control mode
selection signal to the primary one via communication as shown in Figure 12.9.
12.3.3.2 Central control

The secondary control can be designed to meet the power quality requirements.
Figure 12.10 demonstrates the droop characteristic of the secondary control. If the
Rated power
Tertiary V *DC
mode
control VDC + Mode
selection
+
Output
Secondary voltage
control Power sharing V *DC reference
mode
+
+
–
Rd Output current
Figure 12.9 Hierarchical control block diagram

Converter output voltage reference
Nominal voltage
Pri δvi
ma Se
ry con
dar
y
Rii
Output current
Figure 12.10 Droop characteristic of the secondary control

output current in the primary level is increased according to the load, the output
voltage is decreased due to droop control. Then, the secondary control yields a
DC-bus voltage compensation term, dvi , by which the modified voltage reference is
obtained as [26]
vref
i ¼ vnom Rii ii þ dvi : (12.3)
The tertiary control, which is the highest-level control, decides whether the gen-sets
and ESSs are operated or not, depending on the load power demand. Furthermore,
the tertiary controller can perform the economic power dispatch control and energy
balancing control.
12.3.3.3 Case study

In this subsection, the simulation study of MVDC shipboard power system with
the hierarchical control is described. The bus structure and system power capacity
are selected referred to the IEEE Standard 1709–2010 [5]. The parameters of the
simulation model are listed in Table 12.1.
● Model and control of MVDC IPS
In simulation models, the gen-sets consist of two GTs with SGs and two DEs with
PMSGs. The prime movers are modeled as shown in Figure 12.7, where t ¼ 0.02,
tc ¼ 0.0025, and Ky ¼ 1 in p.u. For fuel savings, the optimal speed control of the
prime movers is applied [18]. In addition, the generators are modeled as a variable
voltage source with internal impedance.
The output of the SG is connected to the MVDC bus through diode rectifiers,
but that of the PMSG is connected through diode rectifiers and back-end DC/DC
converters. For speed and excitation controls of the gen-sets, the PI controllers are
used. The ESS is modeled as a current source with a slew rate of 100 MW/s, which
is operated in a bidirectional power flow mode. The operating mode and power
reference for the ESS are commanded by the tertiary control.
There are three kinds of loads applied, where the zonal load is modeled as a
resistance, the propulsion drive as a constant power load, and the PPL as a pulse
charging circuit with a large capacitance.
Table 12.1 Parameters of MVDC shipboard power system
Parameters Values
MVDC-bus voltage 6 kV
GT-SG set 36 MW 2
DE-PMSG set 4 MW 2
Propulsion (M1, M2) 36 MW 2
Zonal loads 5 MW
Pulsed power load 4 MW for 1 s
Energy storage system 4 MW h
The secondary controller regulates the MVDC bus voltage with information of
gen-sets. The output voltage references of each gen-set system are calculated in
(12.3), which will compensate for the variations of the MVDC bus voltage. Finally,
the tertiary controller selects the operating mode of the gen-sets according to the
load profile.
● Load profile
The load profile has four operating modes, as shown in Figure 12.11, which are
classified by the power demand of ship loads.
Mode I: The consumption of the load power is lower than 5 MW. The gener-
ated power is supplied only to services loads since the ship is in port. In this
mode, only DEs except GTs are operated.
Mode II: The consumption of the load power is ranged between 5 and 36 MW.
In this mode, only one GT can support the whole power demand for pro-
pulsion and service loads.
Mode III: The consumption of the load power is ranged between 36 and
72 MW. In this mode, both GTs are operated and DEs are not operated. The
PPL is applied in this mode.
Mode IV: The consumption of the load power is higher than 72 MW. The four
gen-sets should be operated to satisfy the power demand while the ship is on
the voyage at the maximum speed.
The operating modes of the four gen-sets are summarized in Table 12.2.
● Simulation results
Figure 12.12 shows the system performance in operating modes I and II. Initially,
the system is operating in mode I, where two DEs supply power to loads. At 10 s,
the propulsion drives are activated. Then, the load power is increased and the
operating mode is changed to mode II. The central control sends a start command to
the GT1. At the same time, the governor adjusts the mechanical torque to increase
80
Mode IV
70 72 MW
60
Load power (MW)
Mode III
50
40
36 MW
30
20 Mode II
10
5 MW
Mode I
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320
Time (s)
Figure 12.11 Load profile

Table 12.2 Control mode of shipboard power system
Operating Control mode of gen-sets Load power

mode
*DE1 *DE2 **GT1 **GT2
Mode I Droop control Droop control Off Off Lower than 5 MW
Mode II Off Off Voltage Off 5–36 MW
control
Mode III Off Off Droop Droop 36–72 MW
control control
Mode IV Rated power Rated power Droop Droop Higher than
control control control control 72 MW
*DE, diesel engine; **GT, gas turbine.
40
Pload Pgen PESS (MW)
Pgen
DE and GT1 speed
30 1
ωr_DE (1 and 2)
20 ωr_GT1
(p.u.)
Pload
10 PESS 0.5
0
–10 0
(a) 10 20 30 40 50 60 (d) 10 20 30 40 50 60
40 6.3
DC bus voltage (kV)
GT1 power (MW)
6.2
30
6.1
20 6.0
10 5.9
5.8
0
5.7
(b) 10 20 30 40 50 60 (e) 10 20 30 40 50 60
5 4
DE power (MW)
4
Operating mode
PDE (1 and 2) 3
3
2
2
1 1
0
0
10 20 30 40 50 60 10 20 30 40 50 60
(c) Time (s) (f) Time (s)
Figure 12.12 System performance in operating modes I and II: (a) generator,
ESS, and load powers, (b) output power of GT1, (c) output power
of DE1 and DE2, (d) speed of DE1, DE2, and GT1, (e) DC bus
voltage, (f) operating mode
the speed. During this interval, the load power exceeds the maximum power of two
DEs (DE1 and DE2). The central controller requests the ESS to support the
demanded power at 13 s. When the GT reaches to the optimal speed at 15 s, the
output voltage is regulated by the exciter system and the GT begins to support
the power. The three generators are still operated together until the load power
stops increasing at 25 s. After that, the central controller sends a stop command to
the DE1 and DE2 at 27 s. At 28 s, the charging operation of the ESS is activated.
At 45 s, the propulsion load begins to decrease (not shown). As it reaches 5 MW
at 55 s, the operating mode is changed back to mode I. The GT1 continues sup-
plying power to the load. If the propulsion stops completely, then, only zonal loads
consume the power. The DE1 and DE2 are requested to supply the power. When
the DE1 and DE2 can afford to supply the load demand, the central controller gives
a stop command to the GT1 at 57 s.
Figure 12.13 shows the system performance in operating modes II and III.
Initially, the system is operating in mode II, where one GT1 supplies the power of
25 MW to the loads. At 110 s, the propulsion loads begin to increase to accelerate
the ship. At 118 s, the load power demand becomes higher than 36 MW, from
80 Pgen
1
60
GT speed (p.u.)
ω r_GT1
40 PLoad ω r_GT2
0.5
20 PESS
0
0
(a) 110 120 130 140 150 160 170 180 (d) 110 120 130 140 150 160 170 180
50 6.2
DC bus voltage (kV)
GT1 power (MW)
40 PGT1 6.1
30 6.0
5.9
20
5.8
10 PGT2 5.7
0 5.6
(b) 110 120 130 140 150 160 170 180 (e) 110 120 130 140 150 160 170 180
5 4
Operating mode
DE power (MW)
4 3
3
2
2
1 1
0
0
(c) 110 120 130 140 150 160 170 180 (f) 110 120 130 140 150 160 170 180
Figure 12.13 System performance in operating modes II and III: (a) generator,
ESS, and load powers, (b) output power of GT1 and GT2, (c) output
power of DE1 and DE2, (d) speed of GT1 and GT2, (e) DC bus
voltage, (f) operating mode
which the operating mode is changed to mode III. The central controller requests
the ESS to support the demanded power at 118 s and at the same time sends a start
command to the GT2. After some delay, the output power of GT2 begins to
increase and the support of the ESS decreases. At 133 s, two GTs begin to operate
at a balanced power sharing mode without the support of the ESS. While the system
is operating in steady states, the ESS energy is recovered by a charging operation.
At 150 s, the propulsion load begins to decrease (not shown). As it reaches
36 MW at 159 s, the operating mode is changed back to mode II. Two GTs are still
operating in parallel until the load demand stops decreasing at 165 s. Then, the
central controller sends a stop command to the GT2. At 175 s, the GT2 is deacti-
vated completely and the GT1 alone supplies the power.
Figure 12.14 shows the system performance in operating modes III and IV.
Initially, the system is operating in mode III, where two GTs supply the power of
55 MW to the loads. At 230 s, the propulsion loads begins to increase for accel-
erating the ship. At 242 s, the load power demand becomes higher than 72 MW, DE and GT speed (p.u.)
100
Pgen ωGT1
80 1
60 ωGT2
Pload
40 0.5
PESS ωDE
20
0
0
230 240 250 260 270 280 230 240 250 260 270 280
(a) (d)
40 6.3
DC bus voltage (kV)
PGT2
GT power (MW)
35 6.2
PGT1 6.1
30 6.0
25 5.9
5.8
20 5.7
(b) 230 240 250 260 270 280 (e) 230 240 250 260 270 280
5 4
DE power (MW)
Operating mode
4
3
3 PDE (1 and 2)
2 2
1 1
0 0
230 240 250 260 270 280 230 240 250 260 270 280
(c) Time (s) (f) Time (s)
Figure 12.14 System performance in operating modes III and IV: (a) generator,
ESS, and load powers, (b) output power of GT1 and GT2, (c) output
power of DE1 and DE2, (d) speed of DE1 and 2 and GT1 and 2,
(e) DC bus voltage, (f) operating mode
5 PPLL PESS
PPPL PESS (MW)

4
3
2
1
0
–1
285 290 295 300 305 310 315
(a)
6.2
DC bus voltage (V)
6.1
6.0
5.9
5.8
5.7
285 290 295 300 305 310 315
(b) Time (s)
Figure 12.15 System responses for pulsed power load: (a) pulsed power load
and ESS power, (b) DC bus voltage
from which the operating mode is changed to mode IV. The central controller
requests the ESS to support the demanded power at 243 s and at the same time
sends a start command to the DE1 and DE2. At 247 s, both of the DE1 and DE2
generate their rated powers, from which the output power of the ESS is decreased.
During this interval, the output power of GTs varies to make a balance between the
total supplied power and the load power demand.
At 260 s, the propulsion load begins to decrease. As it reaches 72 MW at 263 s,
the operating mode is changed back to mode III. Two GTs and two DEs are still
operating in parallel until the load demand stops decreasing at 55 MW, at 271 s.
After a delay, the central controller sends a stop command to the DE1 and DE2.
Then, only the GT1 and GT2 supply the power to the loads.
Figure 12.15 shows the responses in the case of applying the PPLs while the
system is operating in mode III. The PPL consumes a high power of 4 MW for 1 s.
The central controller sends a command to the ESS, which supplies the demand of
the PPL quickly. It is seen that the MVDC bus voltage has a little fluctuations of
about 150 V, which is within the allowable range of 10%.
12.4 Stability analysis

In shipboard MVDC power systems, the stability analysis is needed to ensure
reliable operation and power quality [5]. The shipboard MVDC power systems
have multiple power sources and a lot of linear and nonlinear loads which are
complicatedly connected. Even though the operation of each unit is designed to be

stable, the system control performance may be unstable when they are integrated,
due to various operating modes, interferences between subsystems, and different
load power characteristics [27].
For system stability analysis, the simulation based on a precise model can be per-
formed in a time domain. However, it takes a long time to confirm the stable operation
over the wide operating range. To reduce the simulation time, the concept of average
model may be utilized [28].
On the other hand, there are analytic methods, which are based on large-signal
analysis and small-signal analysis. The large-signal stability analysis is based on
Lyapunov theory which gives the region of the asymptotic stability, where the
effects of large perturbations and nonlinearity are involved. Meanwhile, the stabi-
lity analysis based on small-signal analysis is used to investigate the system sta-
bility for small variations around the operating points.
For stability analysis of the shipboard MVDC power system, the Lyapunov
method is difficult to apply since the MVDC IPS is a complicated system. On the
other hand, the time-domain simulation can be employed, but it usually requires a
lot of repetitive tasks to find stable operating conditions. So, the small-signal-based
stability analysis is preferred, of which analysis procedure is described as the fol-
lowing [27–31]:
1. Modeling of shipboard MVDC power system
At first, it is necessary to derive the mathematical model of the power sys-
tem network, which is usually modeled in p-equivalent circuit. In the model,
the power line impedance can be neglected since the cable length is sufficiently
short in ships.
2. Simplification of system
For the analysis, each subsystem is classified into two categories as source
subsystem and load subsystem, from which a two-port system is obtained.
3. Determination of stability criteria
In general, the stability analysis of a two-port system is performed by
applying the Middlebrook criteria based on Nyquist plot. However, this
method is difficult to apply where the power flow direction changes after
system reconfiguration. However, the passivity-based stability analysis can be
applied to solve the problem. That is, the stability can be discerned from the
passivity of the bus impedance of one-port system which is transformed from
the two-port system.
4. Identification of bus impedance
The DC bus impedance should be known before applying the stability cri-
teria. The input and output impedances of the power converters which are
connected to the DC bus can be directly measured by digital network analy-
zers. Otherwise, they can be estimated from the transfer function of each
subsystem. For this, the closed-loop transfer functions including the controllers
of each subsystem are derived. From these transfer functions, the DC bus
impedance can be determined. An MVDC bus where m power sources and n 7
loads connected in parallel can be simplified by one-port system, which can be

expressed as
Vbus ðsÞ
Zbus ðsÞ ¼ ¼ Z1 ==Z2 ==. . .Zn ==. . .Znþm (12.4)
Iinj ðsÞ
where Zbus is the overall bus impedance, Z1 to Znþm are the impedances of the
subsystems.
5. Passivity-based stability analysis [30]
If the one-port system with Zbus can only absorb the energy, it is said that the
system is passive. This condition can be expressed as
ðT
vbus ðtÞiinj ðtÞdt 0; for all T (12.5)
1
For the linear time-invariant (LTI) system, the system is stable if and only if
the following conditions are satisfied:
(i) Zbus(s) includes no right-half plane poles.
(ii) Re{Zbus(jw)} 0, 8w
12.5 Faults and protection

According to the IEEE Std. 1709–2010, the ring bus structure with zonal dis-
tribution is recommended for the shipboard MVDC power system. This bus
structure provides a good survivability and reliability of ships since the faulty zone
can be isolated easily. In addition, it has an advantage that the load can be supplied
from the two longitudinal MVDC buses, resulting in redundancy. Therefore, the
system can be autonomously reconfigured to supply the power to the load at
healthy zones.
The most frequent faults in the shipboard MVDC power system are line-to-line
faults and line-to-ground faults. For the shipboard DC distribution system, the
ungrounded system is preferred since it has high survivability and reliability when
the line-to-ground fault occurs. Moreover, it is simple and economic [32].
To ensure the continuous operation of the critical loads at fault conditions, the
appropriate fault protection scheme is needed to isolate the fault location quickly.
Since there are no zero-crossing points in the DC system, it is more challenging to
develop the protection scheme including fault detection and fault location [33,34].
12.5.1 Architecture of protection schemes

There are two structures of protection schemes for the shipboard DC power system.
One is the circuit-breaker-based scheme and the other is the converter-based
method. The former scheme mostly employs the SSCB, which has been described
in Section 12.2.3 [32]. The latter scheme utilizes a current limit algorithm of the
power converters which can be regarded as a protective device to isolate faults.
This can reduce the weight and volume of the system significantly without any
Trip signal
generated
Voltage
Fault
Normal current System
operation limited restored
Current
Fault
t0 t1 t2 t3 t4 t5 t6 t7 Time
Figure 12.16 Time sequence for converter-based protection scheme [35]
additional devices. However, the reliability and survivability of the converter-based

scheme is not so high.
A time sequence for the converter-based protection scheme in the shipboard
MVDC power system is shown in Figure 12.16 [35]. When a fault occurs, the
protection schemes of all the power converters connected to the faulty bus are
triggered. At t0, a fault occurs and the current increases rapidly. At t1, a current
limiting mode is operated by limiting the currents of the converters which feed the
faulty bus. During this interval, the central controller detects the fault and identifies
the location.
After the fault is identified, a tripping signal is generated at t2 to block
the operation of power converters. After some delay time, the faulty bus is
de-energized by reducing the converter current at t3. When the currents of the power
converters are decreased to zero, at t4, the magnetic contactor opens to isolate the
faulty branch. At t5, healthy sections are reconfigured by closing relevant magnetic
contactors. At t6, the remaining systems are reenergized and resume the normal
operation at t7.
12.5.2 Fault detection and location

In this section, the fault detection and location are introduced, where the single
line-to-ground fault is investigated for a simplified DC power system of the
ungrounded port bus, as shown in Figure 12.17. An AC/DC converter is used for
the power conversion which feeds the 6 kV MVDC bus. The two DC/DC con-
verters and a DC/AC converter are connected to the MVDC bus, nominal voltages
6 kV
DC/AC
MVDC bus 2,300 V AC
+ – converter
Port propulsion
A1 B
AC/DC Icm_A
1 Sensing
DC/DC
converter Icm_B station 4
converter 1 kV DC
R2 Zone 1 load
Vn A A2 C
n R3
Icm_A Sensing
Icm_A 2 DC/DC Icm_C
R4 Icm_A station 5
3
converter 1 kV DC
R1 Vcm_off Sensing Zone 2 load
station 1 A3 D
C1
Sensing
GFDL Sensing Icm_D station 6
CR Sensing
controller station 3
Local station 2
area
network
(R1 = 25 Ω, C1 = 0.75 μF, R2 = R3 = 1 MΩ and R4 = 10 MΩ)
Figure 12.17 Configuration of ground-fault detection and location scheme
of which are 1 kV DC for zone 1 and 2 loads and 2,300 V AC for port propulsion
drives, respectively.
Depending on the location of the ground fault, the voltages at all interfaces
with respect to the ground shift by different levels [36]. For instance, if a ground
fault occurs at positive or negative DC buses, the voltages at all the interfaces shift
negatively or positively by a half of DC bus voltage, respectively. However, if the
ground fault occurs at AC bus, the voltages at all the interfaces shift positively and
negatively by a peak value of phase voltages in AC bus.
Figure 12.18 shows the line-to-ground voltages at each interface when the
ground fault occurs in the system of Figure 12.18. If the ground fault occurs at the
negative bus of 1 kV DC in zone 1, all the output voltages of power converters
are shifted by 500 V, as shown in Figure 12.18(a). For example, the positive
MVDC bus voltage shifts from 3 to 3.5 kV DC and the negative bus voltage shifts
from 3 to 2.5 kV DC. In the case of ground fault at one phase of 2,300 V AC
bus, the voltages at all the interfaces are shifted by a peak phase voltage, 1,878 V,
shown in Figure 12.18(b).
To protect the equipment, a strategy for detecting and locating the ground fault
is needed, which is discussed as follows:
● Ground-fault detection
A ground fault condition is checked by measuring a common-mode offset voltage,
Vcm_off, at the output terminal of the AC/DC converter. Vcm_off is the voltage across
a resistor, R4, which is connected from the mid-point of two identical resistors, R2
and R3, to the ground, as shown in Figure 12.17.
MVDC port 6 kV DC bus voltage MVDC port 6 kV DC bus voltage
5k 5k
Voltage (V)
Voltage (V)
0 0
–5k –5k
Propulsion motor 2,300 V AC bus voltage Propulsion motor 2,300 V AC bus voltage
2k 2k
Voltage (V)
Voltage (V)
0 0
–2k –2k
Zone 1 load 1 kV DC bus voltage Zone 1 load 1 kV DC bus voltage
1.5k 1.5k
Voltage (V)
Voltage (V)
0 0
–1.5k –1.5k
Zone 2 load 1 kV DC bus voltage Zone 2 load 1 kV DC bus voltage
1.5k 1.5k
Voltage (V)
Voltage (V)
0 0
–1.5k –1.5k
0.76 0.78 0.8 0.82 0.84 0.76 0.78 0.8 0.82 0.84
Time (s) Time (s)
(a) (b)
Figure 12.18 Line-to-ground voltages when a ground fault occurs at (a) negative
bus of the 1 kV DC zone 1 load, (b) one phase of 2,300 V AC bus
In the ground-fault detection and location (GFDL) controller, the rms and
average values of Vcm_off are calculated. If both of them are higher than the
thresholds for an interval long enough not to be considered as a transient state, it is
concluded that a DC ground fault has occurred. On the other hand, if only its rms
value is higher than the threshold, it is known that an AC ground fault occurs.
● Ground-fault location
To locate the ground fault, the series R1C1 circuit which is connected to the neutral
point of the three-phase power source is utilized. If the line-to-ground fault is
detected, the controller gives a command to close the contactor relay (CR). Then, a
third-order harmonic current is conducted from the chassis ground to the fault point
through the CR. At all the interfaces of power converters, the common-mode cur-
rents, Icm_x, which are the sum of currents through positive and negative buses are
measured. Then, each sensing station calculates the rms value of its Icm_x and sends
to the GFDL controller through communication links in a delay time.
If the DC ground fault occurs, the common-mode currents of AC/DC and DC/
DC converters are examined. Then, it is identified that the ground fault occurs in
the converter where the highest common-mode current flows. On the other hand, if
the AC ground fault occurs, it is decided that the inverter where the highest com-
mon-mode current flows has a ground fault.
Figure 12.19 shows the waveforms of Vcm_off and Icm_x at the same fault con-
dition as in Figure 12.19. If there is no ground fault in the system, the common-
mode voltage and current are zero. If the ground fault occurs at the negative DC bus
of zone 1 load (point C), the rms and average values of Vcm_off reach to 500 V, from
which the DC ground fault is detected, as shown in the top of Figure 12.19(a).
After a delay, the fault location is found from the Icm_A and Icm_C. If the
ground fault occurs at one phase of inverter (point B), the rms and value of Vcm_off
is 500 V, but its average value is almost zero, as shown in Figure 12.19(b). In this
Vcm_off_RMS Vcm_off_avg. Vcm_off_RMS Vcm_off_avg.

Voltage (V)
Voltage (V)
1k 1k
0 0
–1k –1k
Icm_A Icm_A
0.6 0.6
Current (A)
Current (A)
0.4 0.4
0.2 0.2
0 0
Icm_B Icm_B
0.6 0.6
Current (A)
Current (A)
0.4 0.4
0.2 0.2
0 0
Icm_C Icm_C
0.6 0.6
Current (A)
Current (A)
0.4 0.4
0.2 0.2
0 0
Icm_D Icm_D
0.6 0.6
Current (A)
Current (A)
0.4 0.4
0.2 0.2
0 0
0.76 0.78 0.8 0.82 0.84 0.76 0.78 0.8 0.82 0.84
Time (s) Time (s)
(a) (b)
Figure 12.19 Common-mode offset voltages (in rms) and currents (in rms) when a
ground fault occurs at (a) negative DC bus of zone 1 (b) one phase
of 2,300 VAC bus at B
case, it is diagnosed that the AC ground fault has occurred. It is seen that the
common-mode currents of Icm_A and Icm_B have a significant value.
12.6 Summary
As the environmental regulations have become stricter and stricter recently, the
demand on higher fuel efficiency and lower emission for the ship is more pressing.
Therefore, the AES, which has high survivability and reliability, is predicted to be a
future trend of the ship. Since the MVDC IPS well fits with the AES, more research
efforts should be devoted to the development of this technology up to the maturity
stage.
References
[1] E. Skjong, R. Volden, E. Rødskar, M. Molinas, T. A. Johansen, and
J. Cunningham, ‘‘Past present and future challenges of the marine vessel’s
electrical power system,’’ IEEE Trans. Transport. Electrific., vol. 2, no. 4,
pp. 522–537, 2016.
[2] M. Patel, Shipboard propulsion and power electronics and ocean energy,
USA, FL, Boca Raton, CRC Press, 2012.
[3] ‘‘Queen Elizabeth 2,’’ 2002. Available from: https://en.wikipedia.org/w/
index.php?title=Queen_Elizabeth_2&offset=20041114142442&limit=500&
action=history.
[4] O. Alnes, S. Eriksen, and B.-J. Vartdal, ‘‘Battery-powered ships: A class
society perspective,’’ IEEE Electrific. Mag., vol. 5, no. 3, pp. 10–21, 2017.
[5] IEEE Std. 1709-2010, ‘‘IEEE recommended practice for 1 kV to 35 kV
medium voltage DC power systems on ships,’’ in IEEE Std. 1709-2010, Nov.
2010, pp. 1–54.
[6] J.-F. Hansen, J. Lindtjoern, T. Myklebust, and K. Vanska, ‘‘Onboard DC
grid: The newest design for marine power and propulsion systems,’’
ABB Rev., no. 2, 2012. Available from: https://library.e.abb.com/public/
b4f3f099e9d21360c1257a8a003beac2/ABB%20Generations_20%20Onboard%
20DC%20grid.pdf.
[7] D. Li, R. A. Dougal, E. Thirunavukarasu, and A. Ouroua, ‘‘Variable speed
operation of turbogenerators to improve part-load efficiency,’’ in Proc. IEEE
Electr. Ship Technol. Symp., 2013, pp. 353–359.
[8] Energy and Environmental Analysis, Inc. an ICF Company, Technology
characterization: Reciprocating engines, Environmental Protection Agency,
Combined Heat and Power Partnership Program, 2008.
[9] ABB, Generators for wind power, proven generators – reliable power, ABB
Motors and Generators, 2010.
[10] Y. Tang and A. Khaligh, ‘‘Bidirectional hybrid battery/ultracapacitor energy
storage systems for next generation MVDC shipboard power systems,’’ in
Proc. IEEE VPPC, 2011, pp. 1–6.
[11] S. Kim, S. Choe, S. Ko, and S. Sul, ‘‘A naval integrated power system with a
battery energy storage system: Fuel efficiency reliability and quality of
power,’’ IEEE Electrific. Mag., vol. 3, no. 2, pp. 22–33, 2015.
[12] J. F. Hansen and R. Lysebo, Comparison of electric power and propulsion
plants for LNG carriers with different propulsion systems, ABB AS, 2007.
[13] M. Stieneker and R. W. D. Doncker, ‘‘Dual-active bridge dc-dc converter
systems for medium-voltage DC distribution grids,’’ in Proc. IEEE Power
Electronics Conference and 1st Southern Power Electronics Conference
(COBEP/SPEC), 2015, pp. 1–6.
[14] R. Mo and H. Li, ‘‘Hybrid energy storage system with active filter function
for shipboard MVDC system applications based on isolated modular multi-
level DC/DC converter,’’ IEEE J. Emerg. Sel. Top. Power Electron., vol. 5,
no. 1, pp. 79–87, 2017.
[15] P. Manuelle, B. Singam, and S. Siala, ‘‘Induction motors fed by PWM
MV7000 converters enhance electric propulsion performance,’’ in Proc.
IEEE Power Electronics and Applications, 2009. EPE’09. 13th European
Conference on, 2009, pp. 1–9.
[16] M. Spichartz, V. Staudt, and A. Steimel, ‘‘Modular multilevel converter for
propulsion system of electric ships,’’ in Proc. IEEE Electric Ship Technol.
Symp, 2013, pp. 237–242.
[17] X. Pei, O. Cwikowski, M. Barnes, A. C. Smith, and R. Shuttleworth, ‘‘A
review of technologies for MVDC circuit breakers,’’ in Proc. Industrial
Electronics Society, IECON 2016—42nd Annual Conference of the IEEE,
2016, pp. 3799–3805.
[18] S.-H. Lee, J.-S. Yim, J.-H. Lee, and S.-K. Sul, ‘‘Design of speed control loop
of a variable speed diesel engine generator by electric governor,’’ in Proc.
IEEE IAS Annu. Meet., 2008, pp. 1–5.
[19] S. Di Cairano, D. Yanakiev, A. Bemporad, I. V. Kolmanovsky, and
D. Hrovat, ‘‘Model predictive idle speed control: Design analysis and
experimental evaluation,’’ IEEE Trans. Control Syst. Technol., vol. 20, no. 1,
pp. 84–97, 2012.
[20] S. K. Yee, J. V. Milanovic, and F. M. Hughes, ‘‘Overview and comparative
analysis of gas turbine models for system stability studies,’’ IEEE Trans.
Power Syst., vol. 23, no. 1, pp. 108–118, 2008.
[21] M. Farasat, A. S. Arabali, and A. M. Trzynadlowski, ‘‘A novel control
principle for all-electric ship power systems,’’ in Proc. IEEE Electric Ship
Technologies Symposium (ESTS), 2013, pp. 178–184.
[22] S. D. Umans and D. J. Driscoll, ‘‘Excitation system for rotating synchronous
machines,’’ U.S. Patent 6 362 588, Mar. 26, 2002.
[23] O. Palizban, K. Kauhaniemi, and J. M. Guerrero, ‘‘Microgrids in active
network management—Part I: Hierarchical control energy storage virtual
power plants and market participation,’’ Renewable Sustainable Energy
Rev., vol. 36, pp. 428–439, 2014.
[24] Z. Jin, G. Sulligoi, R. Cuzner, L. Meng, J. C. Vasquez, and J. M. Guerrero,
‘‘Next-generation shipboard DC power system: Introduction smart grid and
DC microgrid technologies into maritime electrical networks,’’ IEEE Elec-

trific. Mag., vol. 4, no. 2, pp. 45–57, 2016. Available from: https://ieeex-
plore.ieee.org/document/7480937/.
[25] T. Dragičević, J. M. Guerrero, J. C. Vasquez, and D. Škrlec, ‘‘Supervisory
control of an adaptive-droop regulated DC microgrid with battery manage-
ment capability,’’ IEEE Trans. Power Electron., vol. 29, no. 2, pp. 695–706,
2014.
[26] J. M. Guerrero, J. C. Vasquez, J. Matas, L. G. de Vicuna, and M. Castilla,
‘‘Hierarchical control of droop-controlled AC and DC microgrids: A general
approach toward standardization,’’ IEEE Trans. Ind. Electron., vol. 58, no. 1,
pp. 158–172, 2011.
[27] A. Riccobono, M. Cupelli, A. Monti, E. Santi, T. Roinila, H. Abdollahi,
S. Arrua, and R. A. Dougal, ‘‘Stability of shipboard DC power distribution:
Online impedance-based systems methods,’’ IEEE Electrific. Mag., vol. 5,
no. 3, pp. 55–67, 2017.
[28] J. Shi, R. Amgai, and S. Abdelwahed, ‘‘Modelling of shipboard medium-
voltage direct current system for system level dynamic analysis,’’ IET Electr.
Syst. Transp., vol. 5, no. 4, pp. 156–165, 2015.
[29] W. Tang and R. H. Lasseter, ‘‘An LVDC industrial power supply system
without central control unit,’’ in Proc. IEEE Power Electron. Spec. Conf.,
2000, vol. 2, pp. 979–984.
[30] A. Riccobono and E. Santi, ‘‘Stability analysis of an all-electric ship MVDC
power distribution system using a novel passivity-based stability criterion,’’
in Proc. IEEE ESTS, 2013, pp. 411–419.
[31] U. Javaid, F. D. Freijedo, D. Dujic, and W. van der Merwe, ‘‘Dynamic
assessment of source-load interactions in marine MVDC distribution,’’ IEEE
Trans. Ind. Electron., vol. 64, no. 6, pp. 4372–4381, 2017.
[32] R. M. Cuzner and V. Singh, ‘‘Future shipboard MVDC system protection
requirements and solid-state protective device topological tradeoffs,’’ IEEE
J. Emerg. Sel. Top. Power Electron., vol. 5, no. 1, pp. 244–259, 2017.
[33] E. Christopher, M. Sumner, D. Thomas, and F. De Wildt, ‘‘Fault location for
a DC zonal electrical distribution systems using active impedance estima-
tion,’’ in Proc. IEEE Electric Ship Technologies Symposium (ESTS), 2011,
pp. 310–314.
[34] P. Cairoli and R. A. Dougal, ‘‘New horizons in DC shipboard power sys-
tems,’’ IEEE Electrific. Mag., vol. 1, no. 2, pp. 38–45, 2013.
[35] Q. Deng, X. Liu, R. Soman, M. Steurer, and R. A. Dougal, ‘‘Primary and
backup protection for fault current limited MVDC shipboard power sys-
tems,’’ in Proc. IEEE Electric Ship Technologies Symposium (ESTS), 2015,
pp. 40–47.
[36] R. M. Cuzner, T. Sielicki, A. E. Archibald, and D. A. McFarlin, ‘‘Manage-
ment of ground faults in an ungrounded multi-terminal zonal DC distribution
system with auctioneered loads,’’ in Proc. IEEE Electric Ship Technologies
Symposium (ESTS), 2011, pp. 300–305.
Chapter 13
DC-based EVs and hybrid EVs
Ruoyu Hou1, Jing Guo1, and Ali Emadi1
13.1 Introduction
The recent combination of environmental, geopolitical, economic, and health

concerns related to internal combustion engine powered transportation has created
fertile ground for renewed interest and investment in electric drive vehicles and
electrified transportation [1]. The paradigm shift has already begun and is gaining
momentum. Figure 13.1 shows the U.S. plug-in electric vehicle sales by model
from 2011 to 2015 [2]. Note that the listed vehicles include plug-in hybrid electric
vehicles (PHEV) and range extended electric vehicles (REEV). Only full-sized
vehicles sold in the United States and capable of 60 mph are listed. It is clear that,
in recent years, more car manufacturers have started to enter the competitive
market of PHEV and REEV. In addition, sales for most vehicle models keep
increasing.
13.1.1 Power electronic system in electrified vehicles

The power electronics system in the electrified vehicle is DC-based. Its possible
architecture is shown in Figure 13.2. It has a high-voltage (HV) battery and a low-
voltage (LV) battery. The HV battery is mainly for the traction system where the
voltage is usually much higher than the vehicular electrical system’s voltage. Based
on its hybridization factor and the type of its powertrain, the HV battery voltage range
varies accordingly. Typically, the LV mild hybrid obtains an HV battery voltage
lower than 60 V. This is mainly because the precautions required for HV systems do
not need to be applied here, as HV standards are applied over 60 V [3]. For example,
in belt-driven starter generator applications, traction power is usually supplied by
lead-acid batteries and the voltage is set around 48 V. Currently, for HV mild hybrids,
the HV battery voltage could be around 110 V. For full hybrid, PHEV, or REEV, the
HV battery voltage is rated from 200 to 650 V, or even higher.
The HV battery charging method can be classified as inductive charging and
conductive charging. The inductive charging is also known as wireless charging
1
Electrical and Computer Engineering Department, McMaster Automotive Resource Centre (MARC),
McMaster University, Canada
140
120
100
Thousand vehicles
80
60
40
20
0
2011 2012 2013 2014 2015
Year
Volvo XC90 Honda Fit EV Mercedes B-Class E BMW i3
Mercedes S550 Plug in Kia Soul EV BMW i8 Ford C-MAX Energi PHEV
Tesla Model X Porsche Cayenne S E Chevrolet Spark Ford Fusion Energi
Smart ED Porsche Panamera S E VW e-Golf Toyota Prius Plug-in
BMW X5 Mitsubishi i-MiEV Smart for Two EV Tesla Model S
BMW Active E Cadillac ELR Fiat 500E Chevrolet Volt PHEV
Honda Accord Toyota RAV4 EV Ford Focus EV Nissan LEAF
Figure 13.1 2011–2015 U.S. plug-in electric vehicle sales by model [2]
Off-board
HV battery
charger
Charge port
LV DC-link
LV battery APM dc
DC
Grid
LV On-board ac
AC
accessories HV
DC/DC boost HV DC-link battery
charger
Electric HV battery
motor DC/AC inverter
Inductive charger
Inductive charger
Figure 13.2 Power electronic system architecture of an electrified vehicle

powertrain
and transfers power magnetically. The AC/DC power factor correction (PFC)
converter, high-frequency DC/AC primary side, and the primary transducer are all
located away from the vehicle. The secondary transducer and AC/DC rectifier are
installed inside the vehicle. The conductive charging systems use direct contact
DC-based EVs and hybrid EVs 323
between the connector and the charge inlet [4]. These can be categorized as
onboard and off-board battery chargers. The onboard charger is a must for elec-
trified vehicles needing external charging. It has power limitations because of cost,
weight, and space constraints. The onboard charger connects to the grid through
the AC charge port from the vehicle and creates the AC/DC power conversion
inside the vehicle. The off-board charger has higher power rating and thus reduces
the charging time. It has the AC/DC power conversion away from the vehicle and
charges the HV battery directly through the DC charge port of the vehicle.
Typically, the propulsion system is an electric motor with a three-phase DC/
AC inverter as the power processing unit between the HV battery and the motor.
Some electrified powertrains apply an additional DC/DC boost converter in
between the HV battery and the inverter to improve the system’s efficiency by
increasing the DC-link voltage and providing an adjustable DC-link voltage for the
traction inverter.
The LV vehicular system voltage is typically 12 V; therefore, the HV battery
cannot be utilized to supply power directly to the vehicular loads. For this reason,
power converters are required, which convert HV from the traction battery to a
lower voltage in order to supply power to the LV vehicular loads, and to charge the
LV battery, which is usually a lead-acid battery. This power converter is called an
auxiliary power module (APM).
In this chapter, several converter topologies of PFC converter, isolated DC/DC
converter, APM, and active power filter (APF) in electrified vehicle applications
are presented. The practical design considerations, including power switch,
DC-link capacitor, and DC bus bar, are given. Finally, the state-of-the-art topolo-
gical system reconfigurations in electrified vehicles are presented.
13.1.2 DC auxiliary loads in electrified vehicles

The loads that are not directly related to the powertrain of the vehicle are called
auxiliary loads. The auxiliary loads in vehicle can be mainly categorized as air
conditioning, heating, ventilation, lighting, wiper and window system, electronic
accessory, and others. Figure 13.3 shows the auxiliary load distribution in the
electrified vehicles. Among them, air conditioning, heating, and ventilation are the
largest auxiliary loads, which are also highly dependent on the ambient tempera-
ture. At extreme temperature, these loads can have up to 32% impact on the range
of the vehicle. Therefore, these loads are, most likely, directly connected to the HV
battery due to the potential high-power demand. The rest loads are all connected to
the 12-V DC-link and are all powered by either the 12-V battery or the APM
converter.
Most of the auxiliary loads in the LV system are resistive DC loads. The
resistance that is seen from the supply side changes with the current drawn by the
load. The total power of a typical LV auxiliary load is around 2 kW. A typical LV
auxiliary load power distribution is shown in Figure 13.4. Lighting loads consume
around 32% of the total power. They are composed of headlights, fog lamps, park
lamps, flashers, turn signals, etc. Among these, back lights, headlamps, and fog
Boost
DC DC
HV Traction motor
DC AC
battery
DC LV auxiliary system
DC
APM 12 V LV battery
Air Wiper and window Electronic Autonomous

Lighting Others
conditioning, system accessory load system
heating,
ventilation Headlights, Computer processing, RADAR,
displays, Power steering,
fog lamps, camera,
power outlets, motor engaging
park lamps, LiDAR,
CD player, park brake
flashers, IMU
turn signals, bluetooth,
back lights GPS
Auxiliary loads
Figure 13.3 Possible DC auxiliary load distribution in the electrified vehicle
11%
Lighting
32% Wiper and window system
23% Electronic accessory
Other loads
Autonomous system
14%
20%
Figure 13.4 Typical LV auxiliary load power distribution in the electrified vehicle
lamps draw most of the power. Wiper and window system related loads draw
around 14% of the total power. Electronic accessory loads include the computer,
displays, power outlets, CD player, Bluetooth, and Global Positioning System.
Electric power steering and motor engaging park brake are some other loads. Other
than that, in vehicles where additional luxury loads are requested, such as power
sunroof, active suspension system, or entertainment systems, the total power level
could be higher.
Besides these conventional auxiliary loads, with the recent development on the
autonomous vehicle, the needed power from autonomous system in the vehicle appli-
cation has also been evaluated. The sensors could be used in the autonomous system are
camera, radio detection and ranging (RADAR), light detection and ranging (LiDAR),
and inertial measurement unit. LiDAR is a sensor used to detect the presence of objects
round the vehicle, so the computer can interpret the vehicle’s relation to its surround-
ings. Besides these sensors, extra computer processing is required for the autonomous
system. The total amount of power occupied by this system is around 200 W.
13.2 Converter topologies in electrified vehicles

13.2.1 Conductive HV battery chargers
AC conductive charging topologies can be categorized into two types: single-stage
and two-stage chargers. The single-stage charger is compact and combines both
AC/DC and DC/DC into one stage. It can achieve smaller size, lower cost, and
higher efficiency. The drawbacks are that it might not provide galvanic isolation
and usually contains large low-frequency current ripple in the converter. The two-
stage charger is an AC/DC PFC converter followed by either a nonisolated or an
isolated DC/DC converter. The role of the front-end AC/DC converter is to convert
the supplied AC voltage into DC voltage and to also act as an input current shaper
for PFC and harmonic reduction. Power factor and total harmonic distortion are the
major factors that influence the selection of specific PFC rectifier topology [5].
Boost type topologies and their variants are widely used for PFC rectification
purposes. A conventional PFC circuit is shown in Figure 13.5(a). It consists of a
full bridge rectifier and a boost preregulator. The boost stage can be continuous
conduction mode (CCM), discontinuous conduction mode (DCM), or critical
LH DH5
DH1 DH3 LH1 DH1 DH2
Load
Load
~ SH1
~ LH2
Cdc Cdc
DH2 DH4 SH1 SH2
DH3 DH4
(a) (b)
SH2 DH1 SH1 SH3

Load
Load
LH ~ LH
~
Cdc Cdc
SH1 DH2 SH2 SH4
(c) (d)
Figure 13.5 Single-phase PFC converter: (a) boost PFC converter. (b) 2-Phase
bridgeless PFC converter. (c) Bridgeless totem-pole PFC converter.
(d) GaN-based bridgeless totem-pole PFC converter with SR
conduction mode (CRCM) to improve the efficiency. However, large portions of

system loss are in the diode bridge and cannot be avoided even with zero-voltage
switching (ZVS) at the boost stage. This inherently limits the peak efficiency of a
conventional PFC. The typical forward voltage of a rectifier diode is 1 V, and there
are two diodes in the current path, which could reduce the total efficiency by 2%.
The efficiency of a well-designed PFC can probably achieve about 97% to even
98%, but efficiency higher than 98% becomes relatively challenging for a standard
PFC due to the fixed diode bridge loss.
In a bridgeless PFC, the diode losses can be eliminated, so efficiencies of 99%
or higher are made possible to meet highest efficiency standards. Various bridge-
less PFC topologies have been proposed to overcome the high diode bridge losses
[6]. Among all the bridgeless PFC topologies, the two-phase bridgeless PFC and
the bridgeless totem-pole PFC (BTPPFC) are the most popular topologies.
The topology of the two-phase bridgeless PFC is shown in Figure 13.5(b). This
topology is essentially two boost legs with each one taking control during half of
the AC cycle. SH1/SH2 are typically MOSFETs and DH1/DH2 can be diodes, or for
higher efficiency, silicon carbide (SiC) diodes. In past years, this has been the most
popular bridgeless PFC topology on the market because it is easy to implement
using conventional Si MOSFETs with the control algorithm’s being similar to a
standard PFC circuit, and the efficiency is improved as it eliminates one diode from
the current path. However, it has several drawbacks. (1) It has low-power density
and component utilization, as it doubles the part counts, and each boost stage only
works during half a cycle. (2) It needs additional return diodes DH3/DH4 to provide
a return current path and a reference DC-link ground to AC line to reduce the
common mode noise. (3) The diode DH1/DH2 need to be fast SiC diodes whose cost
is relatively higher than AC rectifier diodes. (4) This topology is not bidirectional,
which limits the HV battery charger to be unidirectional.
Figure 13.5(c) shows the topology of BTPPFC. It can be considered as a
conventional boost PFC in which half of the diode bridge is replaced by active
switches SH1 and SH2, hence the name ‘‘totem pole.’’ However, its application has
been very limited until recently. The major challenge is the poor reverse recovery
performance of the conventional silicon MOSFETs in the half-bridge configura-
tion, which makes CCM operation impractical due to excess reverse recovery Qrr
loss during turn-on transient. To avoid body diode conduction, BTPPFC with sili-
con MOSFETs must work in DCM or CRCM modes, which is only suitable for
lower power and with more complicated control. The absence of a body diode (zero
reverse recovery charge Qrr) and the fast switching nature of gallium nitride (GaN)
make a GaN-based BTPPFC a good fit for a CCM hard switching half-bridge
power stage. In addition, to further improve the converter efficiency, the diode DH1/
DH2 in the 50/60-Hz line-frequency leg can be replaced by low-Rds(on) synchro-
nous MOSFETs as shown in Figure 13.5(d). The GaN-based CCM BTPPFC
overcomes many issues which exist in the two-phase bridgeless PFC and has the
following advantages: (1) Efficiency improvement. The main current only flows
through two switches at a time. SH1/SH2 is driven with complimentary pulse width
modulation (PWM) signals and the SH3/SH4 on the line-frequency leg is synchronous
MOSFETs which further reduce the conduction loss. (2) Lower part counts, higher
power density, and lower bill of material cost. It uses fewer parts and has a simpler
circuit. It needs only one inductor and neither SiC diodes nor AC return diodes are
required. (3) Bidirectional power flow. BTPPFC is inherently capable of bidirec-
tional operation, which is ideal as the application requires power flow in both
directions. Therefore, the BTPPFC is a promising bidirectional PFC candidate for
achieving bidirectional power flow from grid to vehicle and from vehicle to grid.
The main function of the second-stage DC/DC converter is to offer regulated
voltage and current to better charge the HV battery. Two-stage topology provides
high-power factor, wide line regulation performance and clean charge current.
However, cost also needs to be taken into account, especially for the onboard
charger. The most widely used unidirectional isolated topologies are the phase-shift
full bridge and LLC resonant converter as shown in Figure 13.6. As a frequency-
modulated resonant converter, the LLC converter has a resonant tank between the
primary and secondary sides. The capacitor Cs and the inductors Ls and Lm form
the resonant tank. The Lm is the transformer magnetizing inductor and Ls can be the
transformer leakage inductor or an additional auxiliary inductor. The phase shift
full bridge is a PWM-modulated converter. It utilizes the transformer leakage
inductor and the parasitic capacitors from the primary switches to form a resonant
circuit and creates a switching pattern that utilizes the lagging current in the
inductor to fully discharge the parasitic capacitor before turning on the/a switch.
As shown in Figure 13.6, two contactors T1 and T2 need to be installed on the
DC terminals of a HV battery. They provide the isolation between the battery and
T1
S1 S3
D1 D3 +
Ls
HV battery
Vdc n:1
+
Cdc Co V
– Ho
–
S2 S4 D2 D4
T2
(a)
T1
S1 S3
D1 D3 +
Ls n:1
HV battery
Vdc
+
Cdc Lm Co
– Cs VHo
–
S2 S4 D2 D4
T2
(b)
Figure 13.6 Second-stage unidirectional isolated DC/DC converter: (a) phase-

shift full bridge converter and (b) LLC resonant converter
remaining vehicle systems. Among them, T1 is the main contactor and T2 is the
backup contactor from a safety consideration.
13.2.2 Active power filters in HV battery chargers

In a typical 50/60 Hz single-phase AC HV battery charger for vehicle applications,
as the input current is forced to vary sinusoidally in phase with the input voltage,
the pulsating power at twice the line frequency will be seen on the DC-link after the
single-phase AC/DC PFC converter. In the meantime, by applying a wide band-gap
power device, such as GaN or SiC, higher switching frequency and higher effi-
ciency can be achieved. Therefore, the size of the most passive components in the
power electronic system can be shrunk.
However, large capacitance CDC is still required to balance out the instanta-
neous current difference between the pulsating input and the constant output current.
In this case, the corresponding required capacitance is dependent on the harmonic
energy rather than the switching ripple. In other words, the bulky capacitor is mainly
used to store the low-frequency second-order harmonic energy, whose frequency is
100/120 Hz. If the average output voltage of the HV DC-link is 400 V and the input
AC line frequency is 60 Hz, the relation between the power level, capacitance, and
its voltage ripple is shown in Figure 13.7. For example, if the requirement of the
output peak-to-peak voltage ripple is limited to 10%, the required output capaci-
tance will be about 4 mF for a 6.6-kW HV battery charger. Hence, the output
capacitor bank becomes a major barrier in terms of power density.
9
8
7
Capacitance Cdc (mF)
6
5
4
3 5
2
10
V)
1
(
ge
ta
0
ol
15
kv
1,000
ea
2,000 3,000
-p
4,000
-to
5,000 20
6,000
ak
Output power (W) 7,000

Pe
Figure 13.7 Relation among the power level, capacitance, and its voltage ripple
To increase the power density and reliability of the power electronic system
in vehicle applications, capacitor-less designs have been proposed. APF can be a
potential solution to reduce the required capacitance. The HV battery charger
structure with additional APF is shown in Figure 13.8.
Four conventional single-phase APF converter topologies are shown in
Figure 13.9. They are a voltage source inverter, a current source inverter, a
Idc+ir Idc
AC +
~
DC HV DC
AC
port ir port
APF
(a)
Idc+ir Idc
AC DC AC +
~
DC AC DC HV DC
AC
port ir port
APF
(b)
Figure 13.8 HV battery charger structures with additional APF: (a) single-stage
charger, (b) two-stage charger with isolated DC/DC converter
S1 S3 S1 S3
Lr D1 D3
Cr Cr Lr
S2 S4
S2 S4
D2 D4
(a) (b)
S1 S1
Lr Lr
Cr
Cr S2 S2
(c) (d)
Figure 13.9 Four typical conventional APFs. (a) Voltage source inverter.
(b) Current source inverter. (c) Bidirectional buck converter.
(d) Bidirectional boost converter
bidirectional buck converter, and a bidirectional boost converter. With these addi-
tional APF circuits, the required capacitance can be reduced significantly. How-
ever, by adding an extra APF circuit, additional power electronic components are
needed, and thus the system’s complexity is also increased.
13.2.3 LV auxiliary power modules

In the vehicle, the LV battery is in parallel with all the auxiliary loads. The
topology of the LV APM must incorporate galvanic isolation to protect the LV
electronic system from the potentially hazardous HV [7]. In addition, there is a HV
ratio between the HV and LV battery. These requirements restrict the available
topologies to those containing a transformer. Two popular topologies are shown in
Figure 13.10. Figure 13.10(a) shows the topology which combines the phase shift
full bridge with a synchronous rectification current doubler. Synchronous rectifi-
cation is critical for this application as the conduction loss becomes a significant
percentage of overall loss. The synchronous MOSFET rectifier operates in the third
quadrant. The body diode of the MOSFET conducts prior to the turn-on of the
switch. In other words, the conduction loss of the body diode is generated just
before the synchronous MOSFET turns on. However, it can be turned on in ZVS,
which results in negligible switching loss at turn-on. During the turn-off transient,
the MOSFET stops conducting prior to the body diode, which means that the
synchronous rectifier still has reverse recovery loss from its body diode. If the APM
requires bidirectional operation, the dual-active-bridge (DAB) converter is a
S5
LV battery and aux load
T1
S1 S3 +
+ Ls L1
n:1
HV battery
VHo Co
Cdc –
– L2
S2 S4 S6
T2
(a)
T1
LV battery and aux load
S1 S3 S5 S7
+ Ls n:1 +
HV battery
VHo
Cdc Co
–
– VLo
S2 S4 S6 S8
T2
(b)
Figure 13.10 LV APM isolated DC/DC converter candidates (a) phase-shift full
bridge with current doubler, (b) DAB converter
promising candidate. The DAB converter has active full bridges on both primary
and secondary sides. DAB can provide ZVS on both primary and secondary sides
by utilizing the leakage inductor. Its symmetry and fixed switching frequency make
it a robust converter.
13.3 Practical design considerations
13.3.1 Selections of switching devices

In the converter design, switching devices are selected according to the DC-link
voltage and the peak operating current. The voltage and current ratings of a device
cannot be determined to be the same as the system requirements. Certain margins
need to be allowed since current and voltage spikes exist in the device during turn-
on and turn-off transients, as shown in Figure 13.11. The figure shows the current
and voltage waveforms of an insulated-gate bipolar transistor (IGBT). The current
spike in Figure 13.11(a) is caused by the diode reverse recovery and the parasitic
capacitor charge during the switch turn-on period. The voltage spike in
Vce Ic
400 650
350 550
Switch voltage Vce (V)
Switch current Ic (A)

300
450
250
200 350
150 250
100
150
50
0 50
–50 –50
0 100 200 300 400 500 600 700 800 900 1,000 1,100
(a) Time (ns)
Ic Vce
450 450
400 400
Switch voltage Vce (V)
Switch current Ic (A)
350 350
300 300
250 250
200 200
150 150
100 100
50 50
0 0
0 100 200 300 400 500 600 700
(b) Time (ns)
Figure 13.11 Current through and voltage across an IGBT during turn-on and
turn-off transients: (a) during turn-on transient and (b) during
turn-off transient
Figure 13.11(b) is generated by the voltage across the circuit stray inductance
during the switch turn-off period. Generally speaking, the voltage and current
ratings of the device are chosen to be about 1.5–2 times the values required by
the system.
In addition, other parameters of the switching devices must also be considered,
such as equivalent turn-on resistance, series equivalent inductance, switching time,
and switching energy loss. The large equivalent resistance, switching time, and
energy loss result in high conduction and switching power loss, while the large
series equivalent inductance contributes to the total circuit stray inductance and
leads to a HV spike that may damage the switching device [8].
13.3.2 Selections of DC-link capacitors

It has been proven that the power switch devices and capacitors are the most vul-
nerable components in power electronic systems [9]. The reliability of the system
can be improved by minimizing the number of power switching devices and
capacitors. In power electronics converters, the DC-link capacitor can be electro-
lytic, film, or ceramic capacitors. Ceramic capacitors can be operated at a relatively
high temperature; however, they are brittle. One typical failure mechanism of the
ceramic capacitor is cracking when the PCB is bent.
Meanwhile, the HV electrolytic capacitor is not promising in electrical systems
for automotive applications due to its short lifetime and safety issues. One issue is
the electrolyte leakage which poses several potential safety hazards: human contact
with electrolyte and electrolyte residue, and the short-circuiting of adjacent elec-
tronic systems. This could become a disaster for both the automakers and users.
The merits of film capacitors are low ESR and ESL, no liquid electrolyte, and
strong overvoltage capability. In addition, a film capacitor also has self-healing
capability. Therefore, film capacitors are preferred for installation on HV DC-link
in electrified vehicles rather than electrolytic capacitors. However, compared to
electrolytic capacitors, the capacitance to volume ratio of film capacitors is much
smaller. A size comparison between a 500-mF/450-V film capacitor and a 330-mF/
450-V electrolytic capacitor can be seen in Figure 13.12, regardless of the ripple
current rating. By adding the bulky film capacitor, it results in larger converter
volume and lower power density. This issue becomes more stringent for the
onboard power electronics system, which is preferred to be light weight, small size,
and low cost. Therefore, the type of capacitor in the vehicle is dependent on the
required capacitance. For a relatively small capacitance requirement, a film capa-
citor should be applied, especially on the HV DC-link. However, for a relatively
large capacitance requirement, an electrolytic capacitor or other alternative solution
(APF, etc.) might be applied in the vehicle.
The DC-link capacitors are selected based on the converter system require-
ments. The capacitance, rated maximum continuous current, and rated voltage
are three important parameters in DC-link capacitor selection. First, the capaci-
tance can be calculated according to the DC-link voltage ripple. The voltage ripple
required by most converter systems is 5%–10% of the DC-link voltage. Second,
Figure 13.12 A comparison between a 500-mF/450-V film capacitor and a

330-mF/450-V electrolytic capacitor
l r
I cto r
ondu n laye
w C tio r
ula cto
t Ins ondu
C
d
t
Figure 13.13 A laminated bus bar example where t, w, l, and d are conductor
thickness, width, length along the current direction, and the
distance between two conductors
in order to prevent overheating and the lifetime degradation of the capacitor, the
rated current of the selected capacitor must be higher than the root mean square
(RMS) value of the DC-link ripple current. Finally, the rated voltage of the film
capacitor could be selected as the same as the DC-link voltage, since its over-
voltage capability is high and can reach up to twice the rated voltage.
13.3.3 Design considerations of DC bus bars

In the high-power converter systems, bus bars are usually utilized instead of cables
to simplify the assembly. The shape of a bus bar must be designed based on the
conducting current requirements, the switching device packages and the DC-link
capacitor numbers and packages. An example of a bus bar structure is shown in
Figure 13.13, in which the positive and negative conductor plates of the DC-link
are laminated. The width and length of the conductor are usually determined by the
packages and locations of the converter’s main components, such as switching
devices, DC-link capacitors, and inductors, while the thicknesses of the bus bar
conductors and insulation layer are designed according to the conducting current
and voltage.
Generally, there are no upper limits of the conductor and insulation layer
thickness; however, there are lower limits. For the thickness of the insulation layer,
in order to sustain the operating DC-link voltage, the value should be larger than a
minimum value. The dielectric strength of various insulation materials with dif-
ferent thicknesses is not a constant value, thus the insulation layer thickness should
be defined depending on the materials and the practical dielectric strength. For
conductor thickness, to avoid overheat on the entire bus bar or at some hot spots,
the current density in each conductor must meet the requirement of 5 A/mm2 [10].
The cross-section area A in the current direction is defined by (13.1), and the
current density J is the amount of current I flowing through the unit area of this
cross section, (13.2).

A ¼ w t mm2 (13.1)
I
J¼ A=mm2 (13.2)
A
There are two types of currents flowing in the DC-link bus bar conductors, DC
current and AC current. The DC current flows from the input terminals to the
switching device terminals, while the AC current is the current ripple that flows
through the DC-link capacitor. During the design, both DC and AC current
densities need to be considered, and the conductor thickness and width can be
determined by the required conducting current and the DC-link current ripple.
Finally, the total current density must satisfy the conductor current’s density
requirement [11].
DC current on the bus bar spreads out evenly in the conductor. As a result, DC
current flows through the entire cross section area of the conductor. Here, the
thickness of the current layer is equal to the thickness of the conductor. However,
the AC current tends to be concentrated near to the conductor’s surface due to the
skin effect that is shown in Figure 13.14. Thus, the thickness of the current layer,
J[A_per_m2]
4.9397e+006
4.4779e+006
4.1862e+006
3.8945e+006
3.6028e+006
3.3111e+006
3.0195e+006
2.7278e+006
2.4361e+006
2.1444e+006
1.8527e+006
1.5610e+006
1.2693e+006
9.7764e+005
6.8595e+005
3.9426e+005
d d
1.0257e+005
Figure 13.14 AC current on a square and a round cross section in conductors,

where d is the skin depth
which is called skin depth d, is smaller than the thickness of the conductor. Hence,
adjusting the thickness of the conductor only might not be an effective solution to
improve the AC current density, which can be calculated by replacing t by d in
(13.1). To satisfy the current density requirement, the conductor width also needs to
be considered under AC current conditions.
In addition, when more than one DC-link capacitor is utilized, the current
distribution on the bus bar needs to be balanced to avoid imbalanced current
sharing among the capacitors. Unbalanced current sharing between DC-link capa-
citors may result in higher current stress and a shorter lifetime for some capacitors
than others. This will also impact the reliability of the system. To achieve identical
impedance from different current paths on the DC bus bar and balanced current
sharing between capacitors, a symmetric structure of the bus bar is usually
preferred.
In the bus bar design, besides the requirements of current density and current
distribution, the bus bar parasitic parameters should also be considered.
The parasitic parameters include bus bar resistance, stray inductance, and stray
capacitance. When current flows through a bus bar, power loss is mainly generated
by the resistance. Voltage spikes during switching device turn-off transients are
caused by stray inductance. Besides, high-frequency noise can be filtered by stray
capacitance. As a result, large stray capacitance, small resistance, and small stray
inductance are preferred. On the one hand, the bus bar resistance should be
designed to be as small as possible and is usually under 1 mW. Then, the amount of
power loss caused by this resistance could be several watts. On the other hand, the
thickness of dielectric material layer should be designed to be as thin as possible to
achieve relatively large stray capacitance. Nevertheless, the dielectric layer thick-
ness also affects the insulation ability of the bus bar; to sustain the required DC-link
voltage, the thickness of the dielectric layer must be above a minimum value. In
fact, other than the optimization of resistance and stray capacitance, most efforts
are usually focused on minimizing the bus bar’s stray inductance during the design;
this is because the ohmic losses generated by the bus bar resistance is negligible
compared to the total converter power losses, and there are limited improvements
that could be achieved in stray capacitance maximization.
Total bus bar stray inductance Ltotal can be estimated by the summation of
conductor self-inductance Lself and mutual-inductance LM. Self-inductance is
determined by the shape of the conductor, while mutual-inductance is influenced
by both the shape and the overlapping area of the two conductors. When the current
directions are opposite to each other in two DC-link bus bar conductors, mutual-
inductance is negative. Thus, the total stray inductance of the bus bar can be
obtained as (13.3). In order to minimize the bus bar’s stray inductance, the over-
lapping area of the two conductors should be maximized. By laminating two con-
ductors of the bus bar, the stray inductance can be reduced significantly. That is the
reason why the laminated bus bar structure is popular in various bus bar designs,
seen in Figure 13.13.

Ltotal ¼ 2 Lself LM (13.3)
The designed bus bar can be analyzed by FEA simulations, which is a straight-
forward method for estimating the parasitic parameters. Then, the voltage spike
could be evaluated with the estimated bus bar stray inductance. An equivalent
circuit of a converter phase-leg during the switching transient is shown by Fig-
ure 13.15, where the upper switch S1 is being turned off, while its corresponding
lower diode is turning on. The direction of current i does not change during this
time period, and the current in the high-side switch is transferring to the low-side
diode. During this turn-off period, voltage spike occurs as it is plotted by Fig-
ure 13.11. It is obvious that the voltage spike across switch S1 is caused by the
released energy from the inductances in the circuit, including series equivalent
inductance Lc of the DC-link capacitor, series equivalent inductance L1 of the
switching device, and the bus bar stray inductance Lbusbar. Thus, the voltage spike
and the peak voltage across switch S1 can be calculated by (13.4)–(13.6).
dia1 dia1 dia2 dia1

Vce; peak ¼ UDC Lc L1 þ L2 Lbusbar (13.4)
dt dt dt dt

dia1 dia1 dia2 dia1
Vspike ¼ Lc þ L1 þ L2 þ Lbusbar (13.5)
dt dt dt dt
Vce; peak ¼ UDC þ Vspike (13.6)
During the switching period shown by Figure 13.15, ia1 is reducing and ia2 is
increasing; therefore, the derivative of ia1 with respect to time t is negative and that
of ia2 is positive. Finally, the voltage spike is represented by (13.5), and the peak
voltage across S1 is obtained as (13.6). To prevent the device being damaged, the
peak voltage across the switch should be lower than its rated voltage. Otherwise,
the bus bar design must be revised.
+ Vl,busbar – ia,1
Lbusbar +
L1 Vl,1
– –
Vl,c +
Lc S1
+ Vce
–
i
+
–
Udc C
L2 Vl,2
– +
ia,2
Figure 13.15 The equivalent circuit of a converter phase-leg during the turn-off
transient of switch S1
13.4 Topological reconfigurations of DC systems

in electrified vehicles
The integration of power electronics is critical, as it can reduce the size, weight, and
cost of the overall system. A simple type of system-level integration combines
several parts to better utilize the space in the three dimensions. For example, the
APM converter and traction inverter in the vehicle can be placed together in one
enclosure to share the DC-link capacitors, controller, heat sink, and housing cover
as presented in [12]. More advanced system-level integrated converters also allow
sharing of the power switches and gate drivers. However, these converters usually
either employ a relatively complex control strategy or apply a complicated topol-
ogy to regulate the power flows.
Figure 13.16 shows an integrated converter topology that achieves the bidir-
ectional power flows among four DC terminals [13]. The converter has the func-
tions of one DAB and two two-phase bidirectional buck converters but only need
eight switches. Two coupled inductors are needed so that different inductances can
be obtained according to different current flow directions for different power
conversions. The duty cycle is used to control the buck converter, and the phase
shift angle between the primary and secondary side is applied to control the DAB.
Therefore, the integrated converter can work independently and simultaneously
with multiple functions. However, there is a trade-off between the duty cycle and
phase shift angle which might limit the performance of both power conversions,
especially for the conversion with wide voltage and wide power range, such as a
battery charger. In addition, those integrated multifunctional switches need to
carry the overall power for both power conversions, leading to a higher current
rating on the power switches. This might become an obstacle for the power
switches and cooling system to reach higher power. Eventually, the reduced part
might be the gate-driver circuit only, as the required die size in the power switch
devices and the amount of heat-dissipation are still the same compared to the
nonintegrated method.
The operation of electrified vehicle has two modes. These are HV battery
charging and vehicle running mode. This makes electrified vehicle a unique
application. In other words, the role of operating time can also be considered as the
S1 S3 S5 S7
La Lc
DC DC
port 1 Cin1 Co1 port 3
S2 S4 Lb Ld S6 S8
DC DC
Cin2 Co2
port 2 port 4
Figure 13.16 An integrated multiport converter

fourth dimension for the system-level integration. Therefore, in the recent years,
the more integrated power electronic systems in vehicle applications have been
evaluated and proposed with an attempt to further reduce the size, weight, and cost.
These systems are dual-mode, which are application-oriented and fully utilize the
characteristics of the vehicle application.
13.4.1 Topological reconfigurations in HV charging

and propulsion systems
The first type of dual-mode system integration is between the HV battery charging
system and the traction drive system as shown in Figure 13.17 [14,15]. The system
shown in Figure 13.17(a) utilizes the traction motor and its inverter as the PFC for
the HV battery charging. If a second-stage DC/DC is required in the HV battery
charging system, a DC/DC converter can be added and the two contactors beside
the HV battery pack are utilized to provide a path to bypass this second-stage
DC/DC converter in the vehicle’s running mode. Therefore, the efficiency of
the traction system will not be sacrificed by this system integration.
Depending on the powertrain requirements, the design parameters of the PFC
and traction drive system might not be matched. Another system integration
approach is shown in Figure 13.17(b). In this integrated system, the secondary-
stage isolated DC/DC converter is shared between the traction drive system and
HV battery charging system. Its DC/AC stage and the magnetizing inductor of the
transformer create a two-phase interleaved boost converter for the traction drive
system. Therefore, no additional power switch devices and inductors are needed to
T1
Electric motor
T3
AC AC DC +
~ HV DC
port DC DC port
T2
(a)
T1
T3 PFC
AC AC DC AC +
~ HV DC
port DC
DC AC DC port
Electric AC
M T2
motor DC
(b)
HV battery charging Propulsion
Figure 13.17 Topological reconfigurations between HV charging and propulsion

systems
build the boost converter. As with the previous case, contactors T1 and T2 are
employed for reconfiguration between the two functions and isolation between the
battery and remaining vehicle systems.
13.4.2 Topological reconfigurations in dual-voltage charging

systems
Other than the integration between the HV battery charging system and propulsion
system, dual-mode integration can also be achieved between the HV battery char-
ging system and LV APM charging system. Several integrated multifunctional
dual-mode dual-voltage charging configurations have been proposed in [16–20].
Basically, they can be summarized and mainly classified into three types as shown
in Figure 13.18.
As shown in Figure 13.18(a), the first system integrates the primary DC/AC
side of the APM with the secondary AC/DC side of the second-stage DC/DC
converter of the HV battery charger. The transformer is also shared between the
APM and second-stage DC/DC converter.
T1
T3 PFC
AC ~ AC DC AC +
HV DC
port DC AC DC port
T2
+ DC
LV DC
port AC
(a)
T1
T3
AC DC +
AC ~ HV DC
port DC DC port
T2
T4
AC +
LV DC
DC port
(b)
T1
T3 PFC
AC ~ AC DC +
HV DC
port DC DC port
T2
APF
+ DC
LV DC AC
AC DC
port
(c)
HV battery charging LV battery charging
Figure 13.18 Topological reconfigurations in dual-voltage DC systems

Figure 13.18(b) shows the second system which integrates the primary DC/AC
side of the APM with the first-stage AC/DC PFC converter of the HV battery
charger. An extra inductor is still needed for the PFC, and the extra relay T4 is
required to achieve the reconfiguration.
The third type is shown in Figure 13.18(c). It integrates the primary DC/AC
side of the APM with the APF to alleviate the second-order harmonic current from
the single-phase grid line for the HV battery charger. As a result, the additional
power switches and their gate drivers, heat sink, and extra inductors can be elimi-
nated. However, the extra second-order harmonic energy storage capacitor is still
required for the APF.
13.5 Conclusions
In this chapter, the power electronics converters and inverter in vehicle application
are introduced. Several possible topologies for PFC, isolated DC/DC converter,
APM, and APF are presented. The practical design of DC bus bars and the selection
of power switch devices and DC-link capacitors are given. Finally, the state-of-the-
art topological reconfigurations of DC systems in electrified vehicles are presented.
One of the biggest challenges for the next generation of power electronic
systems in vehicle application will be the cost reduction to provide more affordable
solutions. This has become one of the major barriers for electrified vehicle for mass
commercialization. These dual-mode integration approaches are dedicated to
vehicle application and can reduce the cost, size, and weight of the system.
Therefore, they will also help promoting and accelerating the paradigm shift to the
transportation 2.0.
References
[1] A. Emadi and J. G. Petrunic, Chapter 1: Automotive Industry and Elec-

trification, in Advanced Electric Drive Vehicles, Edited by A. Emadi,
pp. 1–13. Boca Raton, FL: CRC Press, ISBN: 978-1-4665-9769-3, 2014.
[2] U.S. Department of Energy Alternative Fuels Data Center, U.S. Plug-in
Electric Vehicle Sales by Model, 2016. [Online]. Available: http://www.
afdc.energy.gov/data/.
[3] S. G. Wirasingha, M. Khan, and O. Gross, Chapter 10: 48-V Electrification
Belt-Driven Starter Generator Systems, in Advanced Electric Drive Vehi-
cles, Edited by A. Emadi, pp. 331–367. Boca Raton, FL: CRC Press, ISBN:
978-1-4665-9769-3, 2014.
[4] M. Yilmaz and P. T. Krein, ‘‘Review of battery charger topologies, charging
power levels, and infrastructure for plug-in electric and hybrid vehicles,’’
May 2013.
[5] A. Khaligh and S. Dusmez, ‘‘Comprehensive topological analysis of con-
ductive and inductive charging solutions for plug-in electric vehicles,’’ IEEE
Transactions on Vehicular Technology, vol. 61, no. 8, pp. 3475–3489,

Oct. 2012.
[6] L. Huber, Y. Jang, and M. M. Jovanovic, ‘‘Performance evaluation of
bridgeless PFC boost rectifiers,’’ IEEE Transactions on Power Electronics,
vol. 23, no. 3, pp. 1381–1390, May 2008.
[7] R. Hou, P. Magne, and B. Bilgin, Chapter 9: Low-Voltage Electrical Systems
for Nonpropulsion Loads, in Advanced Electric Drive Vehicles, Edited by
A. Emadi, pp. 317–328. Boca Raton, FL: CRC Press, ISBN: 978-1-4665-
9769-3, 2014.
[8] J. Guo, Modeling and Design of Inverters using Novel Power Loss Calcu-
lation and DC-Link Current/Voltage Ripple Estimation Method and Bus Bar
Analysis. Ph.D. Thesis, McMaster University, Hamilton, Canada, 2017.
[9] Y. Song and B. Wang, ‘‘Survey on reliability of power electronic systems,’’
IEEE Transactions on Power Electronics, vol. 28, no. 1, pp. 591–604, Jan. 2013.
[10] J. M. Allocco, ‘‘Laminated bus bars for power system interconnects,’’ in
Proc. 1997 IEEE Applied Power Electronics Conference and Exposition,
Atlanta, GA, Feb. 1997, pp. 585–589.
[11] A. D. Callegaro, J. Guo, M. Eull, et al., ‘‘Bus bar design for high-power
inverters,’’ IEEE Transactions on Power Electronics, vol. 33, no. 3,
pp. 2354–2367, Mar. 2018.
[12] Delphi, Delphi Integrated Inverter, Converter and Controller, 2013. [Online].
Available: http://www.delphi.com/manufacturers/auto/hevevproducts/
integrated-solutions/integrated-i-c-c.
[13] K. Itoh, M. Ishigaki, N. Yanagizawa, S. Tomura, and T. Umeno, ‘‘Analysis
and design of a multiport converter using a magnetic coupling inductor
technique,’’ IEEE Transactions on Industry Applications, vol. 51, no. 2,
pp. 1713–1721, Mar. 2015.
[14] G. J. Su and L. Tang, ‘‘An integrated onboard charger and accessory power
converter using WBG devices,’’ in Proc. 2015 IEEE Energy Conversion
Congress and Exposition, Montreal, QC, Sep. 2015, pp. 6306–6313.
[15] S. Anwar, W. Zhang, F. Wang, and D. J. Costinett, ‘‘Integrated DC–DC
converter design for electric vehicle powertrains,’’ in Proc. 2016 IEEE
Applied Power Electronics Conference and Exposition, Long Beach, CA,
Mar. 2016, pp. 424–431.
[16] A. Khaligh and Y. Tang, ‘‘An integrated dual-output grid-to-vehicle (G2V)
and vehicle-to-grid (V2G) onboard charger for plug-in electric vehicles,’’
US patent US20160016479A1, Jun. 15, 2015.
[17] J. G. Pinto, V. Monteiro, H. Goncalves, and J. L. Afonso, ‘‘Onboard recon-
figurable battery charger for electric vehicles with traction-to-auxiliary
mode,’’ IEEE Transactions on Vehicular Technology, vol. 63, no. 3,
pp. 1104–1116, Mar. 2014.
[18] R. Hou and A. Emadi, ‘‘Applied integrated active filter auxiliary power
modules for electrified vehicles with single-phase onboard chargers,’’ IEEE
Transactions on Power Electronics, vol. 32, no. 3, pp. 1860–1871, Mar.
2017.
[19] R. Hou and A. Emadi, ‘‘A primary full-integrated active filter auxiliary
power module in electrified vehicle applications with single-phase onboard
chargers,’’ IEEE Transactions on Power Electronics, vol. 32, no. 11,
pp. 8393–8405, Nov. 2017.
[20] Y. Kim, C. Oh, W. Sung, and B. K. Lee, ‘‘Topology and control scheme of
OBC–LDC integrated power unit for electric vehicles,’’ IEEE Transactions
on Power Electronics, vol. 32, no. 3, pp. 1731–1743, Mar. 2017.
Chapter 14
DC data centers
Enver Candan1 and Robert C.N. Pilawa-Podgurski2
As high-performance computing and data storage transition toward becoming

Internet-based services, the world has witnessed an ever-increasing demand for
both size and capacity of data centers. The growth of cloud-based services and
applications shows no sign of slowing down, with additional custom-hardware for
machine learning algorithms beginning to be deployed at scale in dedicated data
centers. Today’s data centers accommodate many pieces of information technology
(IT) equipment such as data-processing units, data storage units, and communica-
tion devices. A recent report estimated the energy usage of data centers in the
United States (US) alone at 70 billion kW h in 2014, corresponding to 1.8% of the
total electric energy consumed in the country [1]. Since the IT equipment requires
low DC voltage (typically ranging from a few volts to a few dozen volts) to operate,
various power delivery architectures are established to provide low DC voltage
from utility and renewable resources. In this case, the power delivery infrastructure
in data centers can be considered as a microgrid due to the high installed power
capacity and dynamic loads. However, data centers are also quite different than
typical DC microgrids in many regards, both in the characteristics of the loads
(extraordinarily rapid transients, but all controlled/managed from a central load
scheduling interface) and the extreme up-time requirements. This chapter addresses
major aspects of power-delivery architectures in data centers such as efficiency,
reliability, integration with renewable resources, and protection. A critical evaluation
of the technical and commercial barriers to widespread DC power distribution in data
centers will be performed, along with a few examples of existing DC data centers.
14.1 Introduction
In data centers, since the major energy supply is the AC utility and the primary
power consumers (i.e., IT equipment) require low voltage DC, both AC and DC
power infrastructures are concurrent. As the power is delivered from the utility to
the low voltage DC loads, rectification (power conversion from AC to DC form)
1
Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, USA
2
Department of Electrical Engineering and Computer Sciences, University of California, USA
can be performed at various points, resulting in different power architecture con-

figurations [2].
Traditionally, the utility power is distributed in AC form inside the data center,
and then, rectification and voltage step-down conversion are performed at the load
end of the power architecture by a dedicated power supply unit (PSU) for each
piece of IT equipment as shown in Figure 14.1(a). Conversely, the utility power can
be rectified at the source end of the power architecture (i.e., data center input), and
480 VAC 400 VDC 480 VAC 208 VAC 12 VDC

Rack
AC AC DC AC AC
IT Eq.
DC
AC DC AC AC Blade
Site
substation
{ UPS
Power
distribution
transformer
Grid Data center

(a)
480 VAC 380–400 VDC 48 VDC 12 VDC

Rack
AC AC DC DC
IT Eq.
DC
AC DC DC Blade
Site Central High-voltage Bus
substation rectifier step-down converter
converter
Grid Data center
(b)
480 VAC 208 VAC 48 VDC

Rack
AC AC AC DC
IT Eq.
DC
AC AC DC Blade
Site Power Rack-level Bus
substation distribution rectifier converter
transformer
Grid Data center
(c)
Figure 14.1 Simplified drawings of common power delivery architectures in data

centers: (a) AC power distribution in data centers, (b) DC power
distribution in data center, and (c) AC power distribution in data
center, DC power distribution in the rack or within a few racks
DC data centers 345
distributed in high DC voltage form within the data center. Then, as shown in
Figure 14.1(b), a dedicated DC–DC converter per IT equipment steps down the
high DC voltage at the IT equipment input. Alternatively, the AC power can
be distributed to racks that host the IT equipment, and rectifiers that are in the same
rack as the IT equipment (or in another rack that is in close proximity) provide DC
power to the nearby IT equipment as shown in Figure 14.1(c). Simplified diagrams
of these common data center power distribution architectures are depicted in
Figure 14.1. For prioritizing transition from AC to DC power distribution in
Figure 14.1, isolation transformers, protection equipment, and cooling devices are
not depicted in these figures but of course exist in practical designs and may
introduce additional power conversion stages.
Because of the extensive background, acceptability and well-established
standardization of the AC distribution in many other applications, a high percen-
tage of the existing data centers use derivations of the power delivery architectures
depicted in Figure 14.1(a) or (c), which are fundamentally inherited from telecom
applications. Recently, DC power delivery architecture, as depicted in Figure 14.1
(b), has gained attention, mainly because it involves fewer conversion stages and
could potentially simplify the integration of ancillary distributed energy resources,
such as solar PV and fuel cells (FC).
14.2 Development of DC power distribution in data centers

The technological developments in the early 2000s led to rapid expansion of data
centers, and as a result, their increasing energy consumption has been noted in a
several reports [3,4]. A well-cited 2008 report has underlined the benefits of high
voltage DC power distribution in data centers to date [5], although the idea of using
a higher voltage level than 48 V for data center power distribution appears in the
literature as early as 1999 [6].
Over the years, the literature has addressed various high DC voltage levels
such as 270 [6], 300 [7], 325 [8], 380 [9,10], and 400 V [9,11]; however, the
consensus for high voltage DC distribution in data centers appears to eventually
have become 380 and 400 V. The protracted discussion on voltage levels over
15 years, combined with relatively slow development of standards for the IT
equipment and power distribution such as ETSI EN 300 132-3-1 [12] for 400 V DC
bus voltage, ETSI EN 301 605 [13] for grounding, and IEC 61643-21 [14] for
protection, has arguably discouraged short-term adoption of DC distribution in data
centers.
Following the potential of DC distribution for data centers noted by [5], AC
and DC power distribution architectures for data centers have been both quantita-
tively and qualitatively compared by both academia and industry [15–19]. In
addition, [20] has reviewed some of the highly cited comparison reports and stated
that the results of the studies significantly vary and overstate the benefits of the DC
power architectures. In order to fairly evaluate the comparison studies, it should be
noted that over the years, significant barriers such as lack of standardization,
market share, and compatibility with the IT equipment have prevented the wide-
spread adoption of high voltage DC in data centers, while already well-established
AC distribution architectures have kept developing to meet expectations. This
chapter assesses the state of high voltage DC power delivery architecture for data
centers and points out its unique advantages and challenges but does not attempt
to provide a quantitative comparison between the AC and DC distributions.
14.3 Efficiency
By far, the largest operating cost of data centers is the cost of electricity. For this
reason, improving the energy efficiency of the computing and power delivery is
critical. As shown in Figure 14.1, regardless of the preferred distribution archi-
tecture, the utility power has to go through several cascaded power conversion
stages before it reaches the IT equipment. Since the power consumed by the IT
equipment must be processed by each power converter, the overall power infra-
structure efficiency is limited by the stage with the lowest efficiency. Power
delivery studies must thus consider the entire power conversion stage, from the
high voltage AC input to the building, all the way down to the CPU and memory
voltages, around 1 V.
While at a high-level, it may appear that simply reducing the number of con-
version stages would yield increased efficiency. However, one must be careful to
consider that for a constant power converter volume, a high-step down power
converter generally has lower efficiency than a converter with a modest voltage
step-down. This is particularly apparent in the case of the recently proposed 48 V
to the point-of-load concept, where the conventional architecture shown in
Figure 14.2 involves first a 48 to 9–12 V conversion, followed by (typically sev-
eral) 9–12 V to the point-of-load (e.g., 1–2 V) converters. While it may be tempting
to simply eliminate the two-stage conversion and design a single 48–1 V converter,
such a converter is significantly more difficult to design to be highly efficient and
highly power dense. For example, consider the reference design of [21], which
represents a single-stage buck converter, achieving a peak efficiency of 84%.
48 VDC 9–12 VDC
DC 1V
DC
CPU
DC
DC 1.2 V
DC
Memory
To rack-level converter DC
( AC–DC or DC–DC ) 3.3 V
DC
Disk
DC
Point-of-load IT equipment
converter
Figure 14.2 48 V to point-of-load architecture

DC data centers 347
Other, more complex topologies, likewise achieve limited performance in both

efficiency and power density. In comparison, recent hybrid switched-capacitor
power converters have been shown to achieve near 99% efficiency with extra-
ordinarily high power density (106.5 kW/L) for 48–12 V conversion [22], and
separate 12–1 V converters can similarly achieve high overall efficiency and power
density [23–25]. While there may be other considerations (such as reliability, cost,
reduced complexity) to prefer fewer numbers of stages, the above discussion
highlights that increased efficiency is not necessarily a given outcome of this
approach. Regardless of the power distribution architecture, power electronics
plays a critical role in achieving high power conversion efficiency in data centers.
An in-depth review of power electronics, ranging from utility-scale to chip-level
converters in data centers, can be found in [26].
However, one key approach to increasing the system-level conversion effi-
ciency is to eliminate, to the greatest extent possible, any double conversion, where
power is stepped up and down (or converted from AC to DC or DC to AC) more
times than the absolute minimum. An example of such a double conversion is
the back-to-back AC–DC and DC–AC conversion of the centralized uninterrupted
power supplies (UPS) approach shown in Figure 14.1(a), which is one reason why
it is no longer a preferred approach. The DC power distribution for data centers
offers some opportunities to reduce the number of double conversion power stages.
● A facility-level battery system for power backup in a DC system does not

require an inverter stage while providing backup power to both the IT equip-
ment and auxiliary loads such as lighting and cooling. Since electrochemical
energy storage is inherently in DC form, a battery bank for a utility outage or
failure scenario can be connected to the high voltage DC bus. Of course,
additional circuitry may be needed to connect the battery banks to the high
voltage DC bus for regulatory or other operating reasons; however, such cir-
cuitry does not process the requisite power similar to an inverter that outputs a
tightly regulated AC waveform and synchronizes with multiple converters
across the bus.
● Twice-line frequency energy buffering is a well-known issue in single phase
rectifiers [27]. The recent Google Little Box Challenge [28] has accelerated
research efforts in the area of twice-line energy buffering, and as a result,
extremely high efficiencies for twice-line frequency energy buffering have
been reported in the literature [29]. Nevertheless, moving from single phase
rectification at the server input to a centralized three-phase rectification at
the high voltage DC bus eliminates the need for twice-line frequency energy
buffering and the associated circuitry from the cascaded power chain. It should
be noted, however, that similar benefits can be realized if three-phase rectifiers
are employed in an AC distribution approach, generally at the rack level (i.e.,
rack-level three-phase rectifiers). Generally, however, it is likely less costly
and more practical to achieve very high efficiency when designing fewer large
rectifiers at the data center input (DC distribution), than at each rack or server
(AC distribution).
● Similar to the twice-line energy buffering, power factor correction (PFC) cir-
cuitry is an essential requirement in both single and multiphase high power
rectifiers. The most common PFC architecture is the boost-type converter,
meaning the output of the PFC circuit is at a higher voltage than its input. Since
the IT equipment requires low voltage DC, employing PFC closer to the low
voltage load requires a back-end, high conversion ratio, voltage step-down
converter. Although the centralized three phase rectification does not eliminate
the PFC circuitry, it can remove a cascaded voltage step up and down con-
version at the IT equipment input, which is fundamentally counterintuitive
since the load is at low DC voltage. Here, recent developments on step-down
(e.g., buck-derived) PFC rectifiers [30] show promise to achieve increased
system-level efficiencies.
Another potential advantage of the reduced number of cascaded stages in DC
data centers is the increase in available volume inside the IT equipment blade or
rack for more data processing and/or storage units. For example, the AC to DC
conversion stage in the PSU of the IT equipment requires twice-line frequency
energy buffering and PFC circuitries (as explained above), which consist of bulky
passive components since they need to be designed for grid frequency. Removal of
the AC to DC stage from the PSU thus allows more space in the IT equipment blade
or rack.
Since all the power consumed by the IT equipment must be processed through
the cascaded power stages, the system-level efficiency is still limited by the lowest
efficient power stage. Thus, research efforts have focused on the efficiency
improvements of each major power electronics converter type for high voltage DC
data center applications. Development and commercial availability of wide band-
gap devices have been leveraged in converter designs. Consequently, high-90%
efficiencies were achieved in three-phase rectifiers with 400 V DC output by
eliminating the boost PFC stage [30,31], and mid-90% efficiencies in 400–12 V or
48 V DC to DC converters by using resonant topologies [32–34].
Entirely new power delivery architectures also hold great promise for dramatic
increase in power-delivery efficiency. Instead of focusing on the efficiency
improvements of the individual power converters in the DC to DC conversion stage,
a series-stacked architecture (Figure 14.3) that focuses on reducing the amount of
DC bus
Rack or blade
AC
or DC DC
AC or DC IT Eq.
DC
power supply
DC DC IT Eq.
DC
Central
converter
Differential DC IT Eq.
DC
converter
Figure 14.3 Series-stacked power architecture

DC data centers 349
power processed between the high voltage DC bus and multiple IT equipment has
been proposed in [35, 36]. By electrically connecting IT equipment of the same type
(servers in these works), the series-stacked architecture greatly reduces the requisite
power processed while achieving inherent high voltage step-down. The bulk power to
the servers is delivered by the DC bus current that passes through all the servers
without being processed. Since the series connected loads (servers) must conduct the
same amount of current, their input voltages drift in the case of a power consumption
mismatch. Bidirectional differential converters are employed at intermediate nodes to
maintain voltage balance under all circumstances. These converters are able to
exchange the difference between the server currents in the case of a power mismatch
between the servers. Through series stacking and active voltage balancing using dif-
ferential power processing, only the difference power between the servers needs to be
processed. Therefore, the amount of processed power is decoupled from the total
power delivered, yielding greatly reduced power loss during the power conversion in
comparison to the conventional power-delivery systems where the full server power
must be processed by the converters. Experimental verification of the series-stacked
architecture has demonstrated higher than 99.8% peak efficiency for the steady-state
operation of the servers, corresponding to an up to 40 reduction in average power
loss in comparison to a state-of-the-art PSU to date [36]. In addition, practical chal-
lenges of a series-stacked architecture such as load balancing in software [37],
hot-swapping [38], and unregulated bus operation [39] have been analyzed and
experimentally addressed without sacrificing high power conversion efficiency.
Although experimental verification of this architecture is limited to a stack of four
servers to a 48 V, its superior performance has been analyzed in case studies for 32
series-stacked servers connected to a high voltage (380 V) DC bus [35,36,38].
14.4 Reliability
Maintaining high reliability in data centers is crucial because of our society’s
dependence on the uninterrupted IT services. Nowadays, any outage of the IT
services can have a large impact, both financial and in terms of societal impact.
A typical desired target reliability for a data center is 99.99% uptime (often called
as ‘‘four nines’’), which corresponds to 52.5 min downtime per year [40]. This
requirement, combined with maintaining high efficiency, makes data center power
delivery architecture design challenging. Fortunately, similar to improving the
power delivery architecture efficiency, reducing the number of cascaded power
stages typically reduces the overall risk of system failure and the mean time
between failures. Therefore, the opportunities to reduce the number of power
conversion stages in DC data centers have been also considered to facilitate higher
reliability and uptime. Elimination of the conversion stages such as power
distribution transformers and the inverter at the UPS output in high voltage DC
distribution has been considered as the main advantage for the DC data centers
[11,41,42]. However, the analysis in these works is qualitative, and the details are
unclear. A 2010 study quantitatively compared the reliability of AC and DC power
distribution for data centers with emphasis on UPS and concluded that the DC
distribution would be more reliable than the AC distribution without supplementary
effect of redundancy [16]. To the best of our knowledge, a more recent quantitative
reliability analysis for the DC data centers, which considers UPS, power converter
failure mechanisms and inclusion of different redundancy options, is missing from
the literature on this topic. Nevertheless, reducing the number of conversion stages
alone is not sufficient to assure reliable power distribution; back-up power and fault
tolerance must be incorporated in the data center power architecture design to
achieve the desired reliability level.
14.4.1 Fault tolerance
Fault tolerance is the ability of a system to maintain its operation in the case of
component failures. A fault tolerant system must be able to detect faults, protect the
system from the failed component, and provide redundancy. Fault detection and
redundancy for DC data centers is explained here, while protection methods are
explained later in Section 14.7.
There are two main types of faults to be detected in DC power delivery
architectures: over voltage and over current. Over voltage detection can be rela-
tively easily implemented in DC systems using voltage dividers and comparators.
However, over current detection is nontrivial in DC architectures because of in-
rush current phenomena and limited converter power ratings. Moreover, in data
centers, the rapidly changing loads may appear as an over current situation if load-
scheduling causes many servers to initiate computing at the same time. In-rush,
current can be thought of as an instantaneous and unusual rise at the terminal
current of a DC source or load and is present in all DC systems. Fundamentally, the
in-rush current can occur when a DC source or load connects to another DC source
or load due to the initial voltage difference between their filter capacitances.
The magnitude of the in-rush current depends on the voltage difference, filter
capacitance size, series inductive elements, and the contact resistance. If not
managed, in-rush currents may lead to equipment damage and voltage droops,
which can interfere with the converter control algorithm. Leveraging the fact that
in-rush currents occur during physical connection of different DC sources or loads,
in-rush current limitation techniques such as precharge circuits and soft-start rou-
tines have been developed. However, from a fault detection perspective, differ-
entiating between in-rush currents and an over current due to a component failure is
the main challenge in over current detection in DC data centers. Similarly, the
limited converter power rating complicates over current detection due to the faults.
Because of the finite power rating of the converters in a DC bus, a fault current due
to a component failure is limited by the rated converter power. A fault current may
also result in reduced voltage in the DC bus. Especially in data centers where the
instantaneous load current changes dynamically, differentiating between the full
load current and fault current requires careful analysis and tuning.
Redundancy is typically achieved through the incorporation of additional and
separate power conversion stages, UPS and power distribution paths in the data
center power infrastructure. The redundant components may be operated at all
DC data centers 351
Table 14.1 Tier certificate requirements summary
Tier I Tier II Tier III Tier IV

Redundancy level N Nþ1 Nþ1 N After any failure
Distribution path 1 1 1 Active, 1 Alternate 2 Active
times (e.g., each running at partial load to increase peak efficiency) but, strictly
speaking, only needed to meet the power demand of the load in case of failures.
Typical redundancy levels for data centers are N þ 1, 2N, and 2N þ 1, where
N represents the exact number of power converters or UPS systems in parallel to
meet the load demand. Uptime Institute defines Tier Classification levels for data
centers depending on the redundancy level of the data center [43]. Tier I represents
basic data center infrastructure without any redundancy. Tier II certification
requires redundant power stages and UPS; however, the power distribution path is
not redundant. Tier III certification requires the data center to have both redundant
power stages and multiple independent power distribution paths, although only
one distribution path is actively used at any time, while the other is for maintenance
purposes. Tier IV certification requires both redundant power stages and multiple
active power distribution paths configured to serve the entire data center under any
infrastructure failure. Tier Certificate requirements are summarized in Table 14.1,
and the details can be found in [43].
The DC power distribution in a data center simplifies redundant power archi-
tecture design. Since there is no synchronization and frequency control needed,
replacing redundant power stages is straightforward, given in-rush currents are
managed. In addition, if modular and redundant design is preferred, DC–DC con-
verters, UPS, and even the centralized rectifiers can be replaced without disrupting
the IT equipment operation. In addition to contributing to the overall reliability
through improving redundancy level, such a feature intuitively leads to reduced
equipment repair time in data centers, which further improves availability.
14.4.2 Back-up power

Utility power loss is considered as an expected scenario in data centers instead of a
failure. Data centers employ UPS, backup generators, and recently FCs at specific
points in the power delivery architecture to be able to compensate for both utility
loss and power stage failure. Various possible configurations are depicted in
Figure 14.4.
In the case of a utility loss, the facility level UPS must be able to provide
enough back-up power to the critical loads until the back-up generators can initi-
alize and output sufficient energy to the facility. On the other hand, UPS placement
close to the load (IT equipment) can compensate for any component failure
between the utility and the IT equipment. Uptime Tier Certification requires data
centers to have on-site fuel storage for at least 12 h of utility loss [43].
Rack
AC DC IT Eq.
or DC DC
Blade
DC IT Eq.
DC DC
AC or DC Blade
source Rack-level
UPS DC IT Eq.
DC
(a) Blade
Rack Rack
AC DC
IT Eq. IT Eq.
or DC +
_ Blade
Blade
DC
IT Eq. IT Eq.
DC
AC or DC Blade Blade
source
Distributed UPS Facility-level
IT Eq. UPS IT Eq.
Blade Blade
(b) (c)
Figure 14.4 Various UPS configurations for data centers. (a) Rack-level UPS.
UPS is located inside the racks and provides back-up power for
multiple IT equipment. (b) Distributed UPS. Each IT equipment has a
dedicated UPS. This configuration offers uninterruptible power in
the case of any converter failure. (c) Facility-level UPS. UPS is
located outside of the rack and provides back-up power for
multiple racks
The high voltage DC bus is the prime location for UPS and backup generators
to interface. Due to the DC voltage output of UPS, its connection to the high
voltage DC distribution architecture is straightforward. However, any backup
generator requires a rectifier stage to be able to supply power to the high voltage
DC bus. Although this rectifier is an additional power conversion stage, since it
only operates during a utility level power loss, the system-level efficiency and
reliability are not considerably penalized. In addition, connection of backup gen-
erators to high voltage DC bus does not require synchronization and frequency
control which are essential in AC architectures. Control methodologies are studied
for the DC data centers that employ diesel generators and battery banks for utility
power loss scenarios as early as 2008 [44].
14.5 Integration with other DC sources and loads

DC data centers have a clear advantage over their AC counterparts in interfacing
renewable and/or distributed energy sources and modern auxiliary electric loads
such as cooling and lighting to the power infrastructure.
DC data centers 353
14.5.1 Renewable and distributed energy sources

Photovoltaics (PVs) [45] and FC [46] are two energy sources for data centers
commonly preferred by the industry. Since both PV and FC are inherently DC
energy sources, their integration into the DC power infrastructure is simplified.
While it is theoretically possible to design a PV array to directly interface with the
high voltage bus, typically it is preferred to do so through a dedicated DC–DC
converter that ensures that the PV system operates at its maximum power point,
which varies with irradiation and temperature. Alternatively, multiple panel-
embedded power converters [47,48] can be employed to provide this voltage
matching, in addition to improved PV array performance.
Importantly, the integration ease reduces the power infrastructure control
complexity since frequency control and phase synchronization are not required
when DC sources and loads are connected. Power infrastructure control objec-
tives thus can focus purely on maximizing available energy from renewable
resources and minimizing energy drawn from the utility. A recent study investi-
gated energy savings and carbon dioxide emission reductions when a solar power
generation system is connected to a 380 V DC data center power distribution
architecture [49].
While from a conceptual point, interconnection of a PV array (DC source) to a
high voltage DC bus is simpler than having to go through intermediate AC vol-
tages, one must carefully weigh the availability and cost of commercial power
converters. Here, the rapidly growing solar PV industry has driven down the cost of
high voltage DC–AC inverters due to economies of scale, whereas high-voltage
DC–DC converters for solar PV are less prevalent. Consequently, the apparent
advantage of DC data centers for renewable integration may not be so clear when
considering the low-cost availability of inverters driven by the much larger solar
PV industry.
14.5.2 Cooling
In addition to the computational data storage and networking load, cooling requires
substantial electrical energy in data centers. Since the IT equipment mainly consists
of digital circuits, combined with the losses in the power distribution architecture,
almost all energy consumed in data centers results in heat that must be dissipated.
In order to maintain safe operation of the IT equipment, temperature control is
critical at all times. Therefore, the reliability discussion above also applies to the
cooling system, and back-up power should be designed to support the cooling load
in the case of a utility outage.
In order to provide sustainable heat removal from the IT equipment, a typical
data center cooling infrastructure includes air conditioning equipment and chilled
water system, which require pumps and fans. Such equipment involves electric
motors which are typically controlled by adjustable speed drives (ASD) for max-
imum efficiency and performance. Similar to UPS, DC data centers require only an
inverter stage for ASD, compared to AC-driven ASDs, which first perform recti-
fication, followed by the adjustable frequency and voltage inverter. Overall, one
would expect increased higher efficiency and reliability [50] owing to the reduced
number of stages (and conversions) in the direct DC approach.
14.5.3 Lighting
Although the lighting load in data centers is small in comparison to the computa-
tional and cooling load, DC data centers present unique advantages if modern and
efficient lighting technologies such as LEDs are used. Because of the DC nature of
the modern lighting loads, power conversion stages like rectification, twice-line
energy buffering, and PFC are not required when DC power distribution archi-
tecture is employed. As discussed in the efficiency section, reducing the number of
power conversion stages reduces the power conversion losses for the lighting loads
as well. Recent literature has demonstrated LED drivers with 380 V input for DC
microgrids, featuring dimming and short circuit protection [51].
14.6 Installation
Early papers promoting high voltage DC distribution in data centers noted famil-
iarity of the IT equipment with high voltage DC since the PFC stage of the PSU
typically involves a 400 V DC intermediate voltage. However, the installation
of high voltage DC distribution in data centers requires more than familiarity with
400 V DC. Major installation considerations include isolation, grounding, wiring,
connectors, and total cost of ownership (TCO).
14.6.1 Isolation
Electrical (galvanic) isolation has been an essential part of data center power-
delivery architectures. Provided by transformer usage, the electrical isolation offers
to filter grid disturbances, harmonic currents, and electrical noise. Also, the elec-
trical isolation limits ground loops and circulation of DC currents between the IT
equipment and racks [2]. Because of these benefits, the electrical isolation through
isolation transformers is a recommended practice by IEEE STD 1100 in AC power
distribution architectures [52].
Inherited from the AC power delivery, the DC power delivery architecture in
data centers also employs electrical isolation between the high voltage DC bus and
the IT equipment input [11]. The electrical isolation in DC data centers can be
achieved through incorporation of transformers in the DC to DC conversion stage
as the high DC voltage is stepped down to 12 or 48 V for the IT equipment input.
Recent developments in transformer design and wide bandgap devices [33,34] and
optimization approaches [53] enabled high efficiency and compact converters for
this DC to DC conversion stage. Also, in order to provide only electrical isolation
(without voltage conversion), unity transformation ratio is demonstrated in DC to
DC converters for data center applications at 400 [54] and 48 V [55].
Although electrical noise, ground loops and circulation of DC currents may
still be present in DC data centers, enforcing the electrical isolation may not be the
DC data centers 355
only and must-have practice to overcome such issues. For example, modern com-
munication links typically provide inherent isolation, either through the medium
itself (fiber optics), signal isolation transformers (Ethernet), or AC coupling capa-
citors (high speed serial links). Provided adequate system grounding, high safety
may be obtainable without requiring inherent electrical isolation. An example of
this transition is the case of grid-tied PV inverters, which until recently were
required to have galvanic isolation in the United States, while transformer-less
inverters were adopted earlier in Europe since they offered higher power efficiency
and density at a lower cost [56]. Similar efforts may lead to the elimination of
electrical isolation from data center power distribution architectures in the future.
14.6.2 Grounding
Proper grounding is an essential requirement in power infrastructure both for safety
and signal integrity. Grounding contributes to the safety from electrical hazards by
routing damaging currents away from the IT equipment and personnel. Although
safety is the primary driver for choosing a grounding configuration, proper
grounding also enables a common voltage reference for the overall electrical sys-
tem in the data center including the power infrastructure and communication
equipment.
A review of grounding concepts for DC data centers can be found in [57], and
ETSI EN 301605 [13] defines the grounding standard for high voltage DC dis-
tribution in data centers. Both recommend high resistive mid-point ground as
depicted in Figure 14.5 for DC data centers. This configuration keeps the fault
current flow at a level that prevents both electric shock of the human body and fire
hazard to the IT equipment.
14.6.3 Wiring
Wiring is an important consideration when transferring energy from sources
to loads in data centers. Cable size for wiring in data centers is determined by
considering maximum expected current, allowed voltage drop between the source
and load, and target worst case power transfer efficiency. Moreover, electrical
Protective earthing
Line + conductor
Rack
DC
+ IT Eq.
_
Blade
DC
Line –
Main earthing terminal
{
{
{
Source Distribution Load
Figure 14.5 Mid-point high resistive grounding in DC data centers

insulation to protect from short circuit faults must be considered, with some dif-
ferences between AC and DC rating. Because of the difference in the peak and idle
power of the IT equipment, the maximum expected current in data centers results in
oversizing the cables for the average operation. In addition, allowed voltage drop is
restricted by standards and also depends on line regulation capabilities of the front-
end power electronics converter of the load. On the other hand, although the wiring
power efficiency has little effect on overall power infrastructure efficiency, the
target worst-case power transfer efficiency may result in increasing the cable size.
Here, the differences between AC and DC data center wiring are not viewed as
significant, and DC wiring does not appear to a significant hindrance to DC data
center adoption. A study of wiring practices for high voltage DC data centers can
be found in [58].
14.6.4 Connectors
Connectors are another key element for transferring energy from the cables to the
IT equipment or power conversion stages efficiently and reliably, which are the two
most important aspects of data center power infrastructure. The efficient energy
transfer through connectors requires minimal contact resistance; otherwise, sub-
stantial energy loss may occur which conflicts with the high-efficiency target of
data centers as mentioned in Section 14.3. In addition, the connectors must also be
able to sustain possible arcs and in-rush currents as the IT equipment or power
converters are connected and disconnected during faults or maintenance. The reli-
able connectors are thus important to meet high uptime expectations in data centers
as discussed in Section 14.4.
In [59], the basic characteristics of a 400 V 10 A connector are provided.
Ongoing research on connectors for high voltage DC data centers focuses on arc
and in-rush current limitation in plugs and sockets by employing permanent mag-
nets [60] and embedded solid state electronics [61,62] while providing mechanical
isolation. Here, it is likely that research and product development in other DC
power transmission applications, such as solar PV and electric vehicle charging
will help drive down costs and provide market acceptance.
14.6.5 Total cost of ownership

TCO is an important consideration in data center design because building a data
center is a substantial investment for businesses. TCO extends beyond the cost of
power equipment and electricity, but as far as power architecture is a concern, there
are two main expenses: capital and operational expenses. The capital expenses
include up-front costs such as installation and equipment purchases, while the
operational expenses include the electricity and maintenance cost and therefore
involve efficiency and reliability aspects. A detailed explanation of TCO beyond
the power infrastructure can be found in [40,63].
Unfortunately, TCO is rather overlooked in the literature, as the primary
motivation for most work is efficiency and reliability. A 2011 white paper quan-
tifies the oversizing of the power infrastructure as the primary cost driver of data
DC data centers 357
center TCO and suggests measuring the TCO on a per-rack basis [64]. A three
phase rectifier for DC data centers is optimized for TCO in [30]. In [42], the cost
advantage of high voltage DC data centers is reported as 15% less in capital cost
and 36% less in lifetime cost, but the detailed analysis is missing. It should be noted
that improvements in efficiency and reliability do not translate to business deci-
sions unless the TCO analysis is incorporated into the benefits.
Critically, as TCO analysis requires accurate field data regarding reliability,
it is likely that estimates of hypothetical designs without empirical results vary
widely in their estimates. Hence, the TOC aspects of DC and AC data centers are
difficult to assess at this time. Moreover, the major corporation that carefully track
these metrics (e.g., Facebook, Amazon, Google, etc.) generally view these numbers
as key competitive features of their respective designs and are unlikely to share
them with researchers for dissemination.
14.7 Protection
The IT equipment and power infrastructure in data centers represent large invest-
ments; therefore, any damage due to the power system faults must be prevented by
protection equipment such as fuses, relays, and circuit breakers.
Following the fault detection as mentioned in Section 14.4.1, the protection is
activated to isolate the failed IT equipment or section of the power infrastructure
from the rest of the system. Therefore, the location of the protection equipment in
the power infrastructure is vital to enable isolation of any IT equipment and power
stage whenever needed. The lack of a periodic zero-crossing of DC voltage and
current complicates the protection equipment design for DC data centers. In addi-
tion, although sometimes overlooked, abrupt interruption of a DC current results
in high current slew rates (di/dt), which induces high voltages in any parasitic
inductive loop along the power path. This may endanger semiconductors and
capacitor voltage ratings if not considered.
The periodic zero-crossing is inherent in AC distribution architectures as the
polarity of voltage and current alternates 50 or 60 times per second; however, in
DC distribution, the voltage and current are controlled to be constant values and do
not naturally cross zero. Therefore, self-sustaining arcs are more likely to be pre-
sent in DC power architectures. The basic operation principle of the protection
equipment (i.e., moving electrical contacts away from each other when triggered) is
similar in AC and DC systems, but with added challenges for DC protection. In
order to successfully extinguish an arch in DC distribution architectures, the elec-
trical contacts must move not only further away from each other but also faster than
in AC distribution. Alternatively, protection equipment involving electronics to
force the DC current toward zero can also be used. Details of the operating prin-
ciples for DC protection equipment can be found in [65]. Various aspects of pro-
tection in DC microgrids, from modeling DC arcs [66] to implementation of hybrid
(both electronic and mechanical) circuit breakers [67], have been investigated in
the literature. We note that the availability of low-cost, standards-compliant DC
protection circuitry is typically viewed as a major impediment to practical imple-

mentation to DC data centers.
14.8 Power quality
Power quality issues such as phase imbalance, frequency drifting, and harmonic
currents at the multipliers of fundamental grid frequency are native to ac distribu-
tion architectures but are not present in the DC distribution architectures. However,
harmonics due to specifically ASD presence for cooling loads can be severe in DC
data centers. In addition, excessive use of filter capacitance to overcome electro-
magnetic interference in power electronics converters may result in increased
in-rush currents, which reduces the power quality in DC data centers. A detailed
discussion of such power quality issues in DC microgrids (with an example of DC
data centers) can be found in [68].
14.9 Stability
The DC power distribution architectures in data centers can also present stability
challenges due to rapid variation of the IT equipment workload. The IT equipment
in data centers interfaces with the power distribution system through point-of-load
converters and thus acts as constant power loads. Dynamic utilization of the IT
equipment at different workloads results in rapid transitions between idle and peak
power consumption, resulting in challenging the power electronics converter filter
and controller design for stable operation. In addition, since high efficiency is the
primary objective in DC data centers, every power electronics converter is designed
to be highly efficient which results in little damping in the power flow path.
Therefore, the stability of DC distribution architectures is an ongoing research topic
in the literature. A comprehensive examination of stability issues in DC microgrids
with a focus on constant power loads can be found in [69]. Recently, DC electric
springs have been proposed as an active suspension system to stabilize the DC bus
in microgrids. Theoretical analysis and an experimental proof-of-concept in a 48 V
DC electric spring implementation can be found in [70]. In addition, another stabi-
lity analysis with a focus on droop-controlled DC microgrids can be found in [71].
14.10 Existing high voltage DC data centers

A survey of the key data centers or telecom facilities that use high voltage DC
power distribution in 2014 is provided in [72]. In this section, different types of the
high voltage DC installations ranging from laboratory prototypes to extensions of
existing data centers are given as examples. Note that these DC data centers are not
completely configured as DC, and some of them are even limited to a few rows of
racks. Although this means the high voltage DC distribution is not throughout the
entire data center facility, such installations represent a significant step toward
DC data centers 359
Table 14.2 Summary of example DC data centers
Name Location Power Purpose Date

Duke Energy Charlotte, NC, USA 100 kW Demonstration 2011
NTT Cloud Tokyo, Japan 500 kW Cloud 02/2014
Green Zurich, Switzerland 1 MW Expansion 05/2012
Hikari Austin, TX, USA 208 kW Supercomputer 08/2016
Level 3 Broomfield, CO, USA 120 kW Lab evaluation 06/2016
completely DC powered data centers. Table 14.2 summarizes the examples given in
this section.
● Duke Energy Data Center, Charlotte, NC, USA
Duke Energy Data Center in Charlotte, NC is one of the earliest demonstra-
tions of distributing 380 V DC to the IT equipment racks in the United States.
In [17], 15% reduction in energy consumption compared to the conventional
AC power distribution system is reported. The system includes AC to DC and
DC to DC converters from Delta Products Corporation for the power infra-
structure, and DL 385 G7 servers from Hewlett-Packard (HP), Power795
servers from IBM, and Clarion CX4-960 data storage units from EMC as the
IT equipment. Total compute power was 100 kW, and the power delivery
architecture achieved 72.04% efficiency in 2011 [17].
● NTT Cloud, Tokyo, Japan
NTT facilities has installed a 500 kW system with 380 V high voltage DC
distribution for 4-MW scale cloud facility in February 2014. In this system,
N þ 1 redundant rectifier modules (each rated at 100 kW) provide the high
voltage DC bus to 80 server racks. The system spans 250 m2, has 5 min of
backup power provided by 380 V batteries, and uses mid-point high impedance
grounding. Further details of the implementation including safety and protec-
tion practices can be found in [72].
● Green Datacenter, Zurich, Switzerland
Collaborating with ABB, Green Datacenter installed a 1 MW 380 V DC power
distribution architecture as an extension to an existing data center in May 2012
in Zurich, Switzerland [73]. They reported 15% investment cost reduction in
comparison to an AC architecture. The overall architecture, including both
power distribution and cooling, is estimated to provide up to 20% energy
savings. The IT equipment used is compatible with high voltage DC distribu-
tion and includes HP X1800 G2 Network Storage System, HP DL385 servers,
and the HP BladeSystem c3000.
● Texas Advanced Computing Center, Austin, TX, USA
A supercomputer called ‘‘Hikari’’, which consists of 432 HP Enterprise
Apollo 8000 XL730f servers, has been installed by NTT at Texas Advanced
Computing Center (TACC) in August 2016. The Hikari supercomputer is
energized through to a 380 V DC bus which can be supplied by the solar panels
at the TACC parking lot up to 208 kW. The 380 V DC bus can also be con-
nected to the AC utility using rectifiers to be functional at night [74].
● Level 3, Broomfield, CO, USA
Level 3, a US-based service provider, has recently published their transition
from 48 to 380 V DC distribution [75]. The laboratory implementation is
considered as a proof-of-concept for 380 V DC distribution. The installed
capacity is 60 kW, although extension up to 288 kW is mentioned. Currently,
two 60 kW rectifiers supply the high voltage DC bus at 380 V, followed by two
30 kW 380 V to 48 V DC to DC converters. In addition, the current imple-
mentation has 15–30 min of backup power provided by 380 V batteries, and
preferred þ/190 V with mid-point high impedance grounding as recom-
mended by the ETSI standard. The system was put in service in June 2016.
14.11 Key obstacles to widespread adoption of DC data centers

Despite some key benefits associated with DC distribution over AC distribution in
data centers, commercial adaptation of this technology has been slow. Below, we
provide our view of some of the key obstacles for market acceptance.
14.11.1 Overly optimistic claims

As with most emerging technologies, the proponents of DC power distribution in
data centers overstated the advantages compared to the incumbent (AC) solution. In
particular, early studies that highlighted the benefit of no double conversion
(AC–DC to UPS to DC–AC) compared a hypothetical best case DC distribution
system to a below average AC system, especially in terms of efficiencies.
Moreover, early studies also failed to account for the fact that AC–DC and DC–AC
converter efficiencies (and power densities) have continued to improve, as is evi-
dent by recent commercial and research prototypes [76].
14.11.2 Emergence of rack-level UPS

A key selling feature of the DC distribution data center in the early days was the
elimination of the double conversion penalty associated with first AC to DC then
DC to AC conversion for the central UPS. While this remains a significant
advantage (both efficiency and reliability) compared to central UPS AC distribu-
tion systems, today’s design are moving away from such system toward a single-
conversion AC distribution system, with rack-level UPS [e.g., Figure 14.1(c)] that
does not suffer from double-conversion in the UPS stage. While there are some
benefits in terms of maintenance and costs associated with a central UPS system,
the rack-level UPS can also help mitigate power distribution faults in the data
centers, as each rack can operate directly from its own UPS.
14.11.3 Protection and DC circuit breakers

One area where the DC data center has a distinct disadvantage is in the protection,
circuit breakers, and fuses. Owing to the increased difficulty of extinguishing any
arcing in DC systems, the circuit breakers and other protective equipment will
DC data centers 361
necessarily be bulkier and more expensive. In addition to the increased cost asso-
ciated with more advanced circuit breaker technology, the cost for DC breakers is
also higher due to the smaller market size than that for AC breakers.
14.11.4 Incumbent cost and familiarity advantages

Arguably the biggest obstacle to overcome for DC data centers is the fact that they
represent a new entry, with many unknowns. Generally, in order to replace any
incumbent technology, the new technology must provide sufficient advantages
compared to conventional techniques to warrant the increased risk with an untested
design. This is particularly difficult in data center applications, where stringent up-
time requirements call for a conservative approach to introducing new components.
Here, one must compare the DC power distribution to an AC power distribution
with rack-level UPS, which represents arguably the best AC power distribution
solution. In this comparison, the DC solution must then promise sufficiently large
improvements (in efficiency, cost, up-time, etc.) to warrant the transition.
Moreover, in this case, the AC power distribution can leverage significant cost
and scaling advantages associated with being widely adopted in other markets
(commercial building power distribution, factories, etc.). Solutions for DC data
centers that require custom hardware (e.g., protection circuitry, power converters,
etc.) that do not have widespread use in other markets are unlikely to realize the
low cost of AC distribution systems.
14.12 Conclusion
Data centers consume a significant amount of electric energy; therefore, their power
delivery architectures require critical consideration for their sustainable growth. In
this chapter, we discussed the efficiency, reliability, integration with renewable
resources, installation, and protection in the DC power delivery architecture in data
centers. We also provided a short list of some existing DC data center installations and
our thoughts on the critical obstacles for widespread acceptance of this architecture.
Although high voltage DC distribution throughout the entire data center facility is still
not widely utilized, we expect that the performance outcome of the early adopters of
this architecture will promote the use of entirely DC power distributed data centers.
References
[1] Shehabi A, Smith SJ, Sartor DA, et al. United states data center energy usage
report. Ernest Orlando Lawrence Berkeley National Laboratory; 2016.
LBNL-1005775.
[2] Rasmussen N. AC vs. DC power distribution for data centers [White Paper];
2011. Available from: http://www.apc.com/salestools/SADE-5TNRLG/
SADE-5TNRLG_R6_EN.pdf.
[3] Koomey JG. Estimating total power consumption by servers in the US and
the world; 2007. Available from: http://www-sop.inria.fr/mascotte/Contrats/
DIMAGREEN/wiki/uploads/Main/svrpwrusecompletefinal.pdf.
[4] Brown R, Masanet E, Nordman B, et al. Report to congress on server and

data center energy efficiency: public law 109-431; 2007. Available from:
http://eta-publications.lbl.gov/sites/default/files/pdf_3.pdf.
[5] Ton M, Fortenbery B, Tschudi W. DC power for improved data center
efficiency; 2008. Available from: http://www.chip2grid.com/docs/DCDemo
FinalReport.pdf.
[6] Yamashita T, Muroyama S, Furubo S, Ohtsu S. 270 V dc system – a highly
efficient and reliable power supply system for both telecom and datacom
systems. In: Proc. IEEE Int. Telecommun. Energy Conf.; 1999. p. 1–6.
[7] Carlsson U, Flodin M, Akerlund J, Ericsson A. Powering the internet –
broadband equipment in all facilities – the need for a 300 V dc powering and
universal current option. In: Proc. IEEE Int. Telecommun. Energy Conf.;
2003. p. 164–169.
[8] Sannino A, Postiglione G, Bollen MHJ. Feasibility of a dc network
for commercial facilities. IEEE Trans Ind Appl. 2003 Sep;39(5):1499–1507.
[9] Bodi F, Lim EH. 380/400 V dc powering option. In: Proc. IEEE Int. Tele-
commun. Energy Conf.; 2011. p. 1–8.
[10] Salato M, Zolj A, Becker DJ, Sonnenberg BJ. Power system architectures
for 380V dc distribution in telecom datacenters. In: Proc. IEEE Int.
Telecommun. Energy Conf.; 2012. p. 1–7.
[11] Pratt A, Kumar P, Aldridge TV. Evaluation of 400V dc distribution in
telco and data centers to improve energy efficiency. In: Proc. IEEE Int.
[12] Final draft ETSI EN 300 132-3-1 environmental engineering (EE); power
supply interface at the input to telecommunications and datacom (ICT)
equipment; part 3: operated by rectified current source, alternating current
source or direct current source up to 400 V; sub-part 1: direct current source
up to 400 V; 2011. Available from: https://www.etsi.org/deliver/etsi_en/
300100_300199/3001320301/02.01.01_40/en_3001320301v020101o.pdf.
[13] Draft ETSI EN 301 605 environmental engineering (EE); earthing and
bonding of 400 VDC data and telecom (ICT) equipment; 2013. Available
from: https://www.etsi.org/deliver/etsi_en/301600_301699/301605/01.01.01_
20/en_301605v010101a.pdf.
[14] IEC 61643-21 low voltage surge protective devices – part 21: surge pro-
tective devices connected to telecommunications and signalling networks –
performance requirements and testing methods; 2012. Available from:
https://webstore.iec.ch/preview/info_iec61643-21%7Bed1.2%7Db.pdf.
[15] Aldridge T, Pratt A, Kumar P, Dupy D, AlLee G. Evaluating 400 V direct-
current for data centers a case study comparing 400 Vdc with 480–208
Vac power distribution for energy efficiency and other benefits [White
Paper]. Available from: https://blogs.intel.com/wp-content/mt-content/com/
research/Direct%20400Vdc%20White%20Paper.pdf.
[16] Sithimolada V, Sauer PW. Facility-level dc vs. typical ac distribution for
data centers: a comparative reliability study. In: TENCON - IEEE Region 10
Conf.; 2010. p. 2102–2107.
DC data centers 363
[17] Kintner D. Duke energy – EPRI dc powered data center demonstration

executive summary; 2011. Available from: http://portal.emergealliance.org/
DesktopModules/Inventure_Document/FileDownload.aspx?ContentID=20674.
[18] Rasmussen N, Spitaels J. A quantitative comparison of high efficiency ac vs.
dc power distribution for data centers [White Paper]; 2012. Available from:
http://www.apc.com/salestools/NRAN-76TTJY/NRAN-76TTJY_R4_EN.pdf.
[19] Szpek M, Sonnenberg BJ, Lisy SM. 400 VDC distribution architectures for
central offices and data centers. In: Proc. IEEE Int. Telecommun. Energy
Conf.; 2014. p. 1–6.
[20] Rasmussen N. Review of four studies comparing efficiency of ac and
dc distribution for data centers [White Paper]; 2012. Available from: http://
www.apc.com/salestools/VAVR-8Q7K7N/VAVR-8Q7K7N_R0_EN.pdf.
[21] Reusch D, Glaser J. Dc–dc converter handbook – a supplement to GaN tran-
sistors for efficient power conversion. Power Conversion Publications; 2015.
[22] Ye Z, Lei Y, Pilawa-Podgurski RCN. A resonant switched capacitor based 4-to-
1 bus converter achieving 2180 W/in3 power density and 98.9% peak efficiency.
In: Proc. 33rd Annu. IEEE Appl. Power Electron. Conf. Expo.; 2018. p. 1–6.
[23] Reusch D, Lee FC, Gilham D, Su Y. Optimization of a high density gallium
nitride based non-isolated point of load module. In: Proc. IEEE Energy
Conversion Congr. Expo.; 2012. p. 2914–2920.
[24] Rhea BK, Jenkins LL, Abell WE, et al. A 12 to 1 V five phase interleaving
GaN POL converter for high current low voltage applications. In: Proc. IEEE
Workshop Wide Bandgap Power Devices and Appl.; 2014. p. 155–158.
[25] Shenoy PS, Amaro M, Morroni J, Freeman D. Comparison of a buck
converter and a series capacitor buck converter for high-frequency, high-
conversion-ratio voltage regulators. IEEE Trans Power Electron. 2016
Oct;31(10):7006–7015.
[26] Krein PT. Data center challenges and their power electronics. CPSS Trans
Power Electron Appl. 2017;2(1):39–46.
[27] Krein PT, Balog RS, Mirjafari M. Minimum energy and capacitance
requirements for single-phase inverters and rectifiers using a ripple port.
IEEE Trans Power Electron. 2012 Nov;27(11):4690–4698.
[28] Detailed inverter specifications, testing procedure, and technical approach and
testing application requirements for the little box challenge, 2015. Available
from: https://littleboxchallenge.com/pdf/LBC-InverterRequirements-20150717.
pdf.
[29] Qin S, Lei Y, Barth C, Liu WC, Pilawa-Podgurski RCN. A high power density
series-stacked energy buffer for power pulsation decoupling in single-phase
converters. IEEE Trans Power Electron. 2017 Jun;32(6):4905–4924.
[30] Schrittwieser L, Kolar JW, Soeiro TB. 99% Efficient three-phase buck-type
SiC MOSFET PFC rectifier minimizing life cycle cost in DC data centers.
CPSS Trans Power Electron Appl. 2017;2(1):47–58.
[31] Xu F, Guo B, Tolbert LM, Wang F, Blalock BJ. An all-SiC three-phase buck
rectifier for high-efficiency data center power supplies. IEEE Trans Ind
Appl. 2013 Nov;49(6):2662–2673.
[32] Biela J, Badstuebner U, Kolar JW. Design of a 5-kW, 1-U, 10-kW/dm3

resonant dc–dc converter for telecom applications. IEEE Trans Power
Electron. 2009 Jul;24(7):1701–1710.
[33] Huang D, Ji S, Lee FC. LLC resonant converter with matrix transformer.
IEEE Trans Power Electron. 2014 Aug;29(8):4339–4347.
[34] Seeman MD, Bahl SR, Anderson DI, Shah GA. Advantages of GaN in a
high-voltage resonant LLC converter. In: Proc. 29th Annu. IEEE Appl.
Power Electron. Conf. Expo.; 2014. p. 476–483.
[35] McClurg J, Pilawa-Podgurski RCN, Shenoy PS. A series-stacked archi-
tecture for high-efficiency data center power delivery. In: Proc. IEEE Energy
Conversion Congr. Expo.; 2014. p. 170–177.
[36] Candan E, Shenoy PS, Pilawa-Podgurski RCN. A series-stacked power
delivery architecture with isolated differential power conversion for data
centers. IEEE Trans Power Electron. 2016 May;31(5):3690–3703.
[37] McClurg J, Zhang Y, Wheeler J, Pilawa-Podgurski R. Re-thinking data
center power delivery: regulating series-connected voltage domains in soft-
ware. In: Proc. IEEE Power Energy Conf. Illinois; 2013. p. 147–154.
[38] Candan E, Heeger D, Shenoy P, Pilawa-Podgurski R. Hot-swapping analysis
and implementation of series-stacked server power delivery architectures.
IEEE Trans Power Electron. 2017;32(10):8071–8088.
[39] Candan E, Shenoy PS, Pilawa-Podgurski RCN. Unregulated bus operation of
server-to-virtual bus differential power processing for data centers. In: Proc.
32nd Annu. IEEE Appl. Power Electron. Conf. Expo.; 2017.
[40] Barroso LA, Clidaras J, Hölzle U. The datacenter as a computer: an intro-
duction to the design of warehouse-scale machines. 2nd ed.; San Rafael,
CA: Morgan & Claypool Publishers, 2013.
[41] Maiquet D, Kervarrec G. New flexible powering architecture for integrated
service operators. In: Proc. IEEE Int. Telecommun. Energy Conf.; 2005.
p. 575–580.
[42] AlLee G, Tschudi W. Edison redux: 380 Vdc brings reliability and efficiency
to sustainable data centers. IEEE Power Energy Mag. 2012 Nov;10(6):50–59.
[43] Tier standard: topology; 2012. Available from: https://uptimeinstitute.com/
publications/asset/tier-standard-topology.
[44] Salomonsson D, Soder L, Sannino A. An adaptive control system for a dc
microgrid for data centers. IEEE Trans Ind Appl. 2008 Nov;44(6):1910–1917.
[45] Sverdlik Y. Apple reaches 100% renewable energy across all data centers;
2013. Available from: http://www.datacenterdynamics.com/content-tracks/
design-build/apple-reaches-100-renewable-energy-across-all-data-centers/
74708.fullarticle.
[46] Sverdlik Y. eBay’s utah data center offers a glimpse into the future; 2013.
Available from: http://www.datacenterdynamics.com/news/ebays-utah-data-
center-offers-a-glimpse-into-the-future/83495.fullarticle.
[47] Pilawa-Podgurski RCN, Perreault DJ. Submodule integrated distributed
maximum power point tracking for solar photovoltaic applications. IEEE
Trans Power Electron. 2013 Jun;28(6):2957–2967.
DC data centers 365
[48] Qin S, Cady ST, Domı́nguez-Garcı́a AD, Pilawa-Podgurski RCN. A dis-

tributed approach to maximum power point tracking for photovoltaic sub-
module differential power processing. IEEE Trans Power Electron. 2015
Apr;30(4):2024–2040.
[49] Yajima H, Usui K, Hayashi T, et al. Energy-saving effects of super com-
puters by using on-site solar power and direct HVDC feeding systems. In:
Proc. IEEE Int. Telecommun. Energy Conf.; 2016. p. 1–4.
[50] Kwasinski A. Analysis of electric power architectures to improve avail-
ability and efficiency of air conditioning systems. In: Proc. IEEE Int. Tele-
[51] Alonso JM, Perdigão M, Costa MAD, Martı́nez G, Osorio R. Analysis and
experiments on a single-inductor half-bridge LED driver with magnetic
control. IEEE Trans Power Electron. 2017 Dec;32(12):9179–9190.
[52] 1100 IEEE recommended practice for powering and grounding electronic
equipment; 2015. Available from: https://ieeexplore.ieee.org/stamp/stamp.
jsp?tp=&arnumber=1638205.
[53] Badstuebner U, Biela J, Kolar JW. Design of an 99%-efficient, 5 kW, phase-
shift PWM dc–dc converter for telecom applications. In: Proc. 25th Annu.
IEEE Appl. Power Electron. Conf. Expo.; 2010. p. 773–780.
[54] Simanjorang R, Yamaguchi H, Ohashi H, et al. High-efficiency high-power
dc-dc converter for energy and space saving of power-supply system in a
data center. In: Proc. 26th Annu. IEEE Appl. Power Electron. Conf. Expo.;
2011. p. 600–605.
[55] Vicor BCM bus converter BCM48Bx480y300A00 isolated fixed ratio dc-dc
converter; 2016. Available from: http://cdn.vicorpower.com/documents/
datasheets/BCM48B_480_300A00.pdf.
[56] Araujo SV, Zacharias P, Sahan B. Novel grid-connected non-isolated con-
verters for photovoltaic systems with grounded generator. In: Proc. IEEE
Power Electron. Specialists Conf.; 2008. p. 58–65.
[57] Hirose K, Tanaka T, Babasaki T, et al. Grounding concept considerations
and recommendations for 400 VDC distribution system. In: Proc. IEEE Int.
[58] Tanaka T, Hirose K, Marquet D, Sonnenberg B, Szpek M. Analysis of wiring
design for 380-VDC power distribution system at telecommunication sites.
In: Proc. IEEE Int. Telecommun. Energy Conf.; 2012. p. 1–5.
[59] Iino T, Hirose K, Noritake M, Nakamura A, Kiryu K, Sekikawa J. Char-
acteristics of 400 V dc plug and socket-outlet for dc distribution systems.
In: Proc. Int. Conf. Renewable Energy Res. Appl.; 2012. p. 1–6.
[60] Iino T, Hirose K, Noritake M, Yuba T, Miyazawa H, Sekikawa J. Devel-
opment of 400 Vdc plug and socket-outlet for dc distribution system. In:
Proc. IEEE Int. Telecommun. Energy Conf.; 2012. p. 1–6.
[61] Tan K, Huang AQ, Martin A. Development of solid state arc-free socket for
dc distribution system. In: Proc. 29th Annu. IEEE Appl. Power Electron.
Conf. Expo.; 2014. p. 2300–2305.
[62] Tan K, Peng C, Liu P, Song X, Huang AQ. Zero standby power high effi-
ciency hot plugging outlet for 380 VDC power delivery system. In: Proc.
31st Annu. IEEE Appl. Power Electron. Conf. Expo.; 2016. p. 132–137.
[63] Koomey J. A simple model for determining true total cost of ownership
for data centers [White Paper]; 2007. Available from: https://www.
missioncriticalmagazine.com/ext/resources/MC/Home/Files/PDFs/(TUI3011B)
SimpleModelDetermingTrueTCO.pdf.
[64] Rasmussen N. Determining total cost of ownership for data center and net-
work room infrastructure [White paper]; 2011. Available from: http://www.
apc.com/salestools/CMRP-5T9PQG/CMRP-5T9PQG_R4_EN.pdf.
[65] Salomonsson D, Soder L, Sannino A. Protection of low-voltage dc micro-
grids. IEEE Trans Power Delivery. 2009 Jul;24(3):1045–1053.
[66] Uriarte FM, Gattozzi AL, Herbst JD, et al. A dc arc model for series faults in
low voltage microgrids. IEEE Trans Smart Grid. 2012 Dec;3(4):2063–2070.
[67] Meyer JM, Rufer A. A dc hybrid circuit breaker with ultra-fast contact
opening and integrated gate-commutated thyristors (IGCTs). IEEE Trans
Power Delivery. 2006 Apr;21(2):646–651.
[68] Whaite S, Grainger B, Kwasinski A. Power quality in dc power distribution
systems and microgrids. Energies. 2015 May;8(5):4378–4399. Available
from: http://dx.doi.org/10.3390/en8054378.
[69] Kwasinski A, Onwuchekwa CN. Dynamic behavior and stabilization of dc
microgrids with instantaneous constant-power loads. IEEE Trans Power
Electron. 2011 Mar;26(3):822–834.
[70] Mok KT, Wang MH, Tan SC, Hui SYR. Dc electric springs – a technology
for stabilizing dc power distribution systems. IEEE Trans Power Electron.
2017 Feb;32(2):1088–1105.
[71] Anand S, Fernandes BG. Reduced-order model and stability analysis of low-
voltage dc microgrid. IEEE Trans Ind Electron. 2013 Nov;60(11):5040–5049.
[72] Inamori J, Hoshi H, Tanaka T, Babasaki T, Hirose K. 380-VDC power
distribution system for 4-MW-scale cloud facility. In: Proc. IEEE Int.
[73] Schmidt T, Ligi A, Zangger E, Monachesi A. World’s most powerful dc
data center online [Press Release]. Available from: http://www04.
abb.com/global/CNABB/CNABB050.NSF!OpenDatabase&db=/global/cnabb/
cnabb054.nsf&v=BF6&e=us&url=/global/seitp/seitp202.nsf/0/C30D32AC816
B2D5BC1257A140026A31F!OpenDocument.
[74] Salazar J. New Hikari supercomputer starts solar HVDC; 2016. Online.
Available from: https://www.tacc.utexas.edu/-/new-hikari-supercomputer-
starts-solar-hvdc.
[75] Ambriz R, Kania M. A service provider’s decision to move from 48 V to
380 V powering: the problem statement, technical assessment, financial
analysis and practical implementation plan. In: Proc. IEEE Int. Tele-
[76] Lei Y, Barth C, Qin S, et al. A 2-kW single-phase seven-level flying capa-
citor multilevel inverter with an active energy buffer. IEEE Trans Power
Electron. 2017 Nov;32(11):8570–8581.
Chapter 15
DC microgrid in residential buildings
Rajeev Kumar Chauhan1, Francisco Gonzalez-Longatt2,
Bharat Singh Rajpurohit1, and Sri Nivas Singh3
15.1 Introduction
A microgrid is an innovative control and management architecture at the distribu-
tion level, which makes it easy to implement smart grid techniques at power dis-
tribution level [1]. Depleting of fossil fuels, reducing the emission of greenhouse
gases and increasing energy demand are main factors responsible for the growing
penetration of photovoltaic (PV) generators to power distribution grids [2]. The
distribution generation (DG) has the potential to provide ancillary services under
these circumstances. For example, they may have many functions such as instan-
taneous power reserve, emergency supply and peak power saving. The PV system
with DG is a good solution for electrification to poorly grid connected or isolated
area. This concept easily works in both the AC and DC microgrid system [3].
However, the implementation of the DC microgrid simplifies and provides the
opportunity to integrate the renewable energy resource (RES) such as PV system,
and fuel cells (FC) without converting into AC [4]. These are intrinsically DC
power source at higher efficiency. Additionally, the loads used in buildings may be
DC or AC load which can be converted into DC. The rapid developments in the DC
technologies, the DC loads, are more energy efficient than the AC loads [5]. The
DC output power of PV systems can be used directly without conversion into the
low voltage direct current (LVDC) distribution system [6].
There are no inductive, capacitive and skin effects in the DC system. The DC
system has a better voltage regulation than the AC system because there is no
inductive effect on steady state [3]. Additionally, the power loss due to charging
and discharging of the capacitor is eliminated in the DC system. So overall, there is
less power loss in the DC system [7]. Due to the absence of the skin effect, the
entire cross-section area of the line conductor is used in the DC system. It means a
thinner conductor could be used for DC systems, and it reduces the line conductor
weight. Additionally, as the cross-sectional area is inversely proportional to the
1
School of Computing and Electrical Engineering, Indian Institute of Technology Mandi, India
2
Department of Electrical Engineering, University of Loughborough, United Kingdom
3
Department of Electrical Engineering, Indian Institute of Technology Kanpur, India
conductor resistance, the DC system has less line resistance than the AC system [7].
In this way, the direct current distribution system improves the system efficiency.
Additionally, the local utilisation of the energy reduces the transmission and dis-
tribution losses and may reduce the shortage of electricity. In this chapter, analysis
and comparison between AC and DC microgrid in residential buildings have been
done based on appliances, converters and their power losses in both systems. The
layouts for LVDC distribution network have been discussed. Both unipolar and
bipolar layouts of LVDC system have been discussed. Two microgrid system
configurations have been discussed: AC residential building [i.e. AC distribution
system (ACDS) with DC appliances] and DC residential building (i.e. distribution
network with DC appliances).
15.2 Conceptualisation: DC microgrids in buildings

A microgrid refers to a power distribution system integrated with distributed energy
resources (DERs) and controllable loads, which can either operate with the main
grid (i.e. grid-connected mode) or use the DERs to supply the loads without the
main grid (i.e. islanded mode) [6]. It can be viewed as a small-scale grid, and it is
widely considered as one of the key components to integrate renewable DERs and
save transmission losses for efficiency.
In a conventional power system, the power flow is a single directional from
bulk generation, e.g. a power plant, to load via a transmission and distribution
network. As more DERs (such as roof-top PV system, small wind turbines – WTs,
battery system, etc.) are becoming available on the distribution side, the bi-directional
power flow is now possible, and it makes the conventional power distribution system
into a microgrid which can use only the DERs as the power supply without the main
grid (i.e. islanded mode) or even sell back its surplus power to the main grid. The size
of a microgrid may vary depending on the application. For example, the concept of
microgrid can be implemented as smart home, smart building, even a smart campus.
Figure 15.1 shows a microgrid example consisting of PVs, WTs, electrical vehicle,
batteries and loads. Two-way communications are available between the local con-
trollers and the microgrid central controller.
15.3 Classification of microgrids

The microgrids can be classified into three categories based on the type of supply
system used by the distribution grid.
15.3.1 AC microgrid system

A typical structure of an AC microgrid system interconnected with medium voltage
(MW) system at the point of common coupling (PCC) is shown in Figure 15.2. The
main system might be an AC or DC bulk system. The DG units and energy storage
system (ESS) are connected at some points within the distribution networks. Part of
DC microgrid in residential buildings 369
MGCC
LC LC LC LC LC
Electric Wind Batteries Photovoltaic Loads

vehicle turbines
Figure 15.1 Conceptual layout of a microgrid including microgrid central

controller (MGCC) and local controllers (LCs)
AC-1
system
Hydro- DG DG PV
unit-4 unit-3 arrays
turbine
AC
loads
AC
loads
LVAC line LVAC line
LVAC line LVAC line

Sensitive
Public
load
utility
PCC
DG DG
unit-1 unit-2
DC ESS WECS
AC-2 loads
system
LVAC: Low voltage alternating current; ESS: Energy storage system;
WECS: Wind energy conversion system; PCC: Point of common coupling
Figure 15.2 The general structure of AC microgrid with DG units and mixed types
of loads
the network consisting of the DG units and load circuits can form a small isolated
AC electric power system, i.e. an ‘AC microgrid’. During normal operating
conditions, the two networks are interconnected at the PCC, while the loads are
supplied from the local sources (e.g. the RES-based DG units) and if necessary
from the utility. If the load demand power is less than the power produced by DG
units, excess power can be exported to the public utility (PU).
15.3.2 Hybrid AC–DC microgrid system

The general structure of a hybrid AC–DC microgrid is depicted in Figure 15.3.
After staying on AC technology in the area of electric power supply, the DC power
systems join it due to technological advancement in DC technology for power
conversion, generation, transmission, and consumption. However, there are many
challenges in DC technologies. Therefore, DC technologies should be integrated
into the power system by applying the algorithms at some places [8]. The scope to
explore the DC technologies with its specific advantages is the microgrid. The
hybrid AC–DC microgrids facilitate benefits of both the technologies, and these
are having the integration of AC technology with the DC technology. In a leading
hybrid AC–DC microgrid system, the AC and DC buses are connected through
interlinking, bi-directional converters. However, this interlinking creates a stability
issue and requires control algorithms to maintain the power quality. Microgrids,
AC DG
system unit-3 WECS
Hydro- DG
unit-4
turbine
AC
loads
PCC
AC
loads
LVAC line LVAC line
LVDC line
Sensitive
Public
DC load
utility
DG DG
unit-1 unit-2
DC ESS PV array
DC loads
system
LVAC: Low voltage alternating current; ESS: Energy storage system;
Figure 15.3 The general structure of hybrid AC–DC microgrid with DG units and
mixed types of loads and generators
which are having different types of sources, and loads are the type of AC–DC
systems [9].
15.3.3 DC microgrid system

The traditional electric power system was designed to move the central station AC
power, via high-voltage AC transmission lines and lower voltage distribution lines
to households and businesses that use the power in incandescent lights, AC motors
and other AC equipment for heating and cooling. Meanwhile, the DC power system
has been used in industrial power distribution systems, telecommunication infra-
structures and point-to-point transmissions over long distances or via sea cables and
for interconnecting AC grids with different frequencies. Power electronics devices
dominate today’s consumer equipment and tomorrow’s DG units [10]. These
devices (such as computers, fluorescent lights, variable speed drives, households,
businesses, industrial appliances and equipment) need DC power for their opera-
tion. However, all these DC devices require conversion of the available AC power
into DC for use, and the majority of these conversion stages typically use ineffi-
cient rectifiers. Moreover, the power from DC-based DG units must be converted
into AC to tie with the existing AC electric network, only later to be converted to
DC for many end users. These DC–AC–DC power conversion stages result in
substantial energy losses. Using the positive experiences in the high voltage direct
current (HVDC) operation and the advance in power electronics technology,
interests to use effective solutions have increased. Figure 15.4 shows typical DC
DC-1 DG
system unit-3 WECS
Hydro- DG
unit-4
turbine
DC
loads
PCC
DC
loads
LVDC line LVDC line
LVDC line LVDC line

Sensitive
DC
DC load
supply
DG DG
unit-1 unit-2
DC ESS PV array
DC-2 loads
system
LVDC: Low voltage direct current; ESS: Energy storage system;
Figure 15.4 General structure of DC microgrid with DG units and mixed load
and generator types
microgrid system interconnected with the main systems at PCC which can be a
medium voltage AC network from the conventional power plants or an HVDC
transmission line connecting an offshore wind farm.
15.4 Topologies for DC microgrid for residential applications
There are several topologies used by the DC microgrid. However, the unipolar or
bipolar type structure is typically used to configure the DC microgrid.
15.4.1 Unipolar LVDC system

The unipolar system consists of a two-winding transformer and a line converter,
connected as shown in Figure 15.5. In a unipolar DC system, the line and loads are
connected via two conductors, a neutral and other a positive polarity DC voltage.
The unipolar system has a one voltage level for energy transmitted. All the
customers are connected to this one voltage level [10]. The unipolar system has
the small power transfer capacity compared to bipolar type system with the same
voltage level (i.e. 12 and 12 V) and does not have a broad range of choices of DC
voltage level.
15.4.2 Bipolar LVDC system

The bipolar system is a combination of a three-winding transformer, and two line
converters, connected as shown in Figure 15.6, i.e. the bipolar system is a combi-
nation of two unipolar systems connected in series. The connection alternatives
may be between a positive pole and neutral, between a negative pole and neutral,
between positive and negative poles. The bipolar system provides a broad range of
DC voltage levels compared to the unipolar system [11]. In the bipolar type, the
system requires more components, and the system may result in an unbalanced
situation when the loads are not identically connected.
DC/AC
converter 1
AC/DC Consumer 1
main
PCC converter
Consumer 2
DC/AC
Public converter 2
utility Consumer 3
DC/AC
converter 3
Figure 15.5 Unipolar LVDC distribution system used in a cluster of residential

buildings
DC/AC
converter 1
AC/DC Consumer 1
main
converter
PCC
Consumer 2
DC/AC
Public converter 2
utility
Consumer 3
DC/AC
converter 3
Figure 15.6 Bipolar LVDC distribution system used in a cluster of residential

buildings
15.5 Mathematical analysis of AC vs DC microgrid system

In this section, mathematical descriptions have been derived to calculate voltage,
current and power losses during an AC as well as DC distribution system supply.
15.5.1 Total daily load

The total daily load (TDL) has been calculated with respect to the different DC
voltage level supply as well as the different AC voltage levels and inverter effi-
ciencies. When the DC load is connected to the system, the DC load (g) in amperes
is given as
k
g¼ (15.1)
VDC
where k is the load rating in kilowatt; VDC is the DC system voltage; and TDL,
LDC, can be obtained by multiplication of the DC load with the number of operating
hours (l) of the load per day in ampere hours, and it can be expressed as:
LDC ¼ g c (15.2)
If there is a variable DC load connected to the system, the calculation of total daily
load is as follows:
X
W
LDC ¼ gj cj (15.3)
j¼1
where, g1, g2, . . . , gW and c1, c2, . . . , cW are variable DC loads in amperes and
different time instants at which DC loads are switched ‘ON’ in a day.
Table 15.1 Illustrative values of total daily load on DC and AC systems
System voltage Power rating Inverter Total daily

(kW) efficiency (%) load (A h)
24 V DC 2.4 – 2,400.0
48 V DC 2.4 – 1,200.0
120 V AC 2.4 92 2,608.7
220 V AC 2.4 92 2,608.7
220 V AC 2.4 95 2,526.3
If the AC load is connected to the system, then the AC voltage must be con-
verted to DC voltage and the inverter efficiency is also considered. The DC load of
the AC system can be expressed as:
k
g¼ (15.4)
VDC hinv
Table 15.1 shows the characteristic values of a DC system based on the type of load
(AC or DC load). If the DC load 2.4 kW is supplied at 24 and 48 V, the TDL is
2,400 and 1,200 A h, respectively. In the case of 2.4 kW AC load at 120 and 220 V
supplied by the same efficiency inverter (92%), then TDL is 2,609 A h remains the
same but higher than the TDL as obtained in the case of 2.4 kW DC load at 24 V. If
a 2.4 kW AC load supplied at 220 V by the 95% efficiency inverter, the obtained is
TDL 2,526 A h. The above calculation shows that TDL depends upon the DC
system voltage and inverter efficiency. If the inverter efficiency is high, the losses
in DC to AC conversion are less, and the system TDL will be reduced. As the
system voltage is high, the system TDL will also decrease, i.e. system losses may
decrease. On the other hand, if there is a lower inverter efficiency or DC supply
voltage, the TDL will be higher. It means the battery bank (BB) will discharge at
the fastest rate, which may decrease the battery lifetime and efficiency.
15.5.2 Voltage, current and power losses in DC supply

In India, the ACDS for residential buildings is single phase 230 V (RMS). The
equivalent DC voltage (325 V) applied to the same load can reduce the current
ratings. For example, 230 and 110 V AC supply the current rating can be reduced to
30% and 70%, respectively [6,7]. The DC system voltage stress equivalent to the
single-phase AC system can be calculated as
pffiffiffi
VDC ¼ 2VAC (15.5)
The DC system has less potential stress compared to the AC system for the same
voltage. For example, if a system is designed for 230 V AC, it can bear 325 V DC
without any rapture in insulation. The voltage level helps to reduce the gap between
two conductors of the distribution line. The less potential stress and weight of
conductor reduce the size of the tower and insulator. This decreases the cost of the
system and makes the system more economical. The power transfer in a DC system
can be expressed as
P¼V I (15.6)
The current in a DC system can be expressed as
P
I¼ (15.7)
V
where P is the transferred power, V and I are the system voltage and current,
respectively. As shown in (15.7), if DC voltage is applied instead of AC voltage
then the insulation can bear higher voltage stress, which allows applying higher DC
voltage for the same system. The current is inversely proportional to the voltage.
It means as the system voltage increase, their current will decrease. Therefore, the
system copper losses (p) will also decrease.
p ¼ I 2R (15.8)
where p represents power losses in the system and R is the resistance of the feeder
cable.
The AC and DC system comparison in terms of current and power losses can
be found in Table 15.2. The current and power losses in the 325 V DC system
(equivalent to 230 V AC system) are approximately one-third and a half respec-
tively as compared to the 230 V AC system (single-phase AC voltage level in
India). Additionally, it has around 2/3 times and 7/8 times less current flow and
power losses, respectively, as compared to the 110-V AC system (single-phase AC
voltage level in the United States).
Table 15.3 shows the energy savings, which can be obtained by switching from
AC technologies to the most energy efficient DC-internal technologies. It is seen
from Table 15.3 that the total energy saving using DC technology in the residential
loads is varying between 30% and 71%. Some rectifier losses have already existed
in a few cases like a refrigerator, cloth washer and fans, in AC system.
The integration of DC sources to conventional AC system necessitates the
introduction of DC–AC converter at the generation end, thereby adding conversion
losses and complexity [11]. In last two decays, the continued development of DC
technologies to produce an energy efficient DC appliance is a cause of significant
decrement in the building load but insists on introducing AC–DC converter and
Table 15.2 Comparative analysis of AC and DC system based on current rating

and power losses
DC voltage AC voltage Reduction in current Reduction power

rating (%) losses (%)
325 230 29.2 50.1
325 110 66.2 88.5
Table 15.3 Possible energy saving using most efficient DC internal technology-
based appliances
Appliances Efficient DC compatible replacement Total energy

technology savings (%)
Fan Run by brushless DC motor in place of single- 47
phase AC induction motors
Room air conditioners Variable-speed compressor 35
Lighting-incandescent 14 LPW incandescent goes to CFL (electronic 73
ballast) at 52 LPW
Lighting-reflector 15 LPW goes to CFL (electronic ballast) at 71
52 LPW
Lighting-touchier Assuming 80%, incandescent at 14 LPW goes to 69
CFL at 52 LPW, and 20% CFL stays the same
Electric water heaters Heat pump 50
Refrigerators Assuming 85% standard-size at 587 kW h AEU 53
with savings of 51% and 15% compact at
331 kW h AEU with savings of 75%
Clothes washers Brushless DC permanent magnet (BDCPM) 30
variable speed
Ceiling fans BDCPM variable speed system 30
Table 15.4 Description of DC appliances and their AC–DC converter in India
Appliance name Voltage Current Power AC–DC converter

rating (V) rating (A) rating (W) efficiency (%)
LED bulb 12 0.6 7 79
CFL bulb 12 1.0 12 79
Electric geyser 96 10.5 1,000 89
Sandwich maker 24 23.0 550 87
Water purifier 24 0.5 11 79
Refrigerator 24 3.0 72 87
Coffee maker 12 11.0 135 87
Washing machine 24 3.0 70 86
Water pump 24 14.9 350 87
Vacuum cleaner 12 8.0 95 87
Air conditioner 24 33.3 800 88
Hybrid car 96 32.0 3,000 90
Cell phone 12 0.3 4 78
Ceiling fan 12 1.7 20 83
Hair drier 24 15.0 425 87
TV 12 2.5 30 83
Computer 12 14 170 87
increase the conversion loss and complexity of the system [12]. The details of the
voltage and power ratings of the DC-appliance and their AC–DC converters effi-
ciency to connect these appliances to the conventional AC system can be found in
Table 15.4. The AC–DC converter efficiencies vary from 78% to 90% according to
Table 15.4. It can be noted that higher the converter power rating is, the higher
is the AC–DC efficiency, as the highest efficiency 90%, which is in the case
of a hybrid car with a converter power of rating 3,000 W. The cell phone
converter of 4-W power rating has the lowest efficiency of 78%, as mentioned
in Table 15.4.
15.6 Comparison between AC and DC residential buildings

A grid-connected residential building consists of DC appliances for AC and DC
distribution system is shown in Figures 15.7 and 15.8, respectively. The building is
supplied by the PU and PV in both cases. The energy storage (BB and EV) is used
to store the energy to supply the future load in case the PV and PU are not acting.
Additionally, the ES is also responsible for storing the surplus power produced by
the PV and supply the surplus load to balance the power in the system in case of
PU acts as outage source. The residential building consists of six rooms. The
electrical specifications of the loads are mentioned in Table 15.4. The voltage
ratings of the appliances are 12, 24 and 96 V for the low, medium and high power
rating loads. As the building appliances have a wide range of voltage levels, the
bipolar type of distribution system is used in order to optimise the conversation
stage in the building.
Positive (+) R1 : Study room

Solar R2 : Entertainment room
Neutral ( )
panels R3 : Bed room
R4 : Kitchen
R1 R3 Charging pad for R5 : Garage
Computer portable appliances R6 : Laundry, control room
SST : Solid-state transformer
DC–AC =
converter ~
Garage
R2 R4 R5
Consumer TV
portal
Car
SST Refrigerator
Main R6 Washing
AC–DC–AC machine
Transformer switch converter
11 kV / 440 V AC board
~
=
Battery
bank
Figure 15.7 Conceptual layout of AC residential building

R1 = Study room
Positive (+) R2 = Entertainment room
Solar
Neutral ( ) R3 = Bed room
panels
Negative (–) R4 = Kitchen
R5 = Garage
R1 R3 Charging pad for
R6 = Laundry, control room
Computer portable appliances
SB = Switch board
SB SST: Solid-state transformer
Garage
R2 R4 R5
Consumer TV
portal SB Car
SST Refrigerator
Main R6 Washing
AC–DC converter switch machine
230 VAC/24 V DC board SB
Battery
bank
Figure 15.8 Conceptual layout of DC residential building
15.7 AC residential buildings

In this case, each line has a single voltage level of 230 V AC. The DC compatible
loads are more efficient than the AC compatible load [11]. Therefore, it is also
assumed that each appliance is DC compatible, which helps to reduce the building
load compared to the AC compatible load.
Moreover, each appliance has its own internal AC–DC converter to connect to
the AC line, which is the cause of additional power losses. The specifications of the
appliances and their converter efficiencies are given in Table 15.4. The total power
consumption in buildings with ACDS (PACB) is the combination of power con-
sumed by appliances (PA) and power loss (pC) in the converters. The power equa-
tion can be expressed as
PACB ¼ PA þ pC (15.9)
The power consumption by the appliances can be expressed as
X
j¼na
PA ¼ Paj (15.10)
j¼1
where Paj is the power consumed by the jth appliance.

The total power loss (pC) in the converters is the addition of power loss in the
internal converters of the appliances (pac) and the power loss in the source con-
verter (psc) and can be expressed as
X
j¼na X
j¼ns
pC ¼ pacj þ pscj (15.11)
j¼1 j¼1
where na is the number of appliances, and ns number of source converters.
15.8 DC residential buildings

In this case, it is also assumed that each appliance is DC compatible which helps to
reduce the building load compared to the AC compatible load. The main DC bus
has 24 V voltage level. Moreover, one boost DC–DC converter to increase the
voltage level from 24 to 96 V and supply EV and electric geyser. A DC–DC buck
converter is used to tie 24–12 V DC bus. The appliances of 12 V ratings, such as
CFL, LED, computer, TV, etc. are directly connected to the 12 V DC bus, while for
the remaining 24 V rating, appliances are also connected to the 12 V DC bus
between the conductors of positive and negative polarity as shown in Figure 15.8.
The total power consumption in the buildings for DC distribution system
(DCDS) (PDCB) is the addition of power consumed by appliances (PA) and power
losses in DC–DC and AC–DC converters (pc). The power expression is as given
below [8]:
X
j¼na X
j¼nc
PDCB ¼ paj þ pcj (15.12)
j¼1 j¼1
where nc ¼ 3, the number of converters in buildings with DCDS.
15.9 Automation architecture for smart DC residential

buildings
The building automation system is where centralised control and monitoring of the
building services are done. The purpose of the automation system is to maintain the
building environment more efficiently to reduce the building’s environmental
impact and energy costs. There may be different types of architecture for the
building automation. A general architecture having all type of complex activities
and facilities has been shown in Figure 15.9, and it can be divided into the fol-
lowing levels [13].
15.9.1 Field level

The control and detection of the devices consist of this level. These devices may be
sensors, actuators, light or smoke detectors, valves, switches and other intelligent
sensors.
Central control
SCADA Web server station
Management
level
Primary network
Gate way
Secondary network
RTU RTU RTU RTU RTU
Automation level
RTU: Remote
terminal unit Sensor Actuator Field level
Figure 15.9 The basic network architecture of a building automation system
15.9.2 Field network

It is the connection network between the field level and automation level. The main
purpose of this network is to connect the field level devices to an RTU (remote
terminal unit), in the automation level. These connections may be of four types:
1. Hardwired
2. Bus system
3. Powerline
4. Wireless
15.9.3 Automation level

This level has different advanced controllers to control and regulate the sensors,
actuators and other types of field-level devices. Usually, the digital-based micro-
processor is used to freely programme them with different control and control logic
like proportional control, integral control, differential control and any other logic
control as well as a combination of logic controls, etc. [14]. The controllers include
the RTU.
15.9.4 Primary and secondary network

Most building automation networks consist of a primary and secondary bus which
connects high-level controllers with lower level controllers, input/output devices
and user-interface devices. The ASHRAE’s open protocol BACnet or the open
protocols LonTalk specify how most such devices interoperate. Modern systems
use the simple network management protocol (SNMP) to track events, building on
decades of history with SNMP-based protocols in the computer networking world.
The primary network is the management network, and it is the backbone of the
system. The primary network connects the automation level and the management
level in the building automation system. A primary network can either be separated
or shared with the conventional local area network in a building.
Secondary networks connect the automation level and the primary network
as a sub-network to the primary network. The purpose of this level is to connect
the device to the automated level, but these devices are working with a different
protocol compared to the devices connected to primary network devices directly.
15.9.5 Management level

It is the level that has the capability to manage activities and monitor the building
automation system. The interaction method may be personal or with the internet.
Some examples of these devices are databases that log activity, web servers,
operator’s panels, central control station and servers that translate messages into
different protocols.
15.10 Advantages, challenges and barriers of smart

DC residential buildings
Some factors will always be there to make the candidature of DC systems

more strong then the AC system such as [15] (i) DC systems have renewable
energy sources, like PV panels and FC, and ESSs, as batteries, (ii) around 50% less
energy consumption of the complete load that appears in the AC system in some
operating point, (iii) the scope of the integration of new sources like electric vehicle
will increase the use of DC devices, (iv) zero skin effect makes the DC systems
more efficient then the AC systems, (v) use of DC devices, sources and storage
to interconnect and distribute the energy between them avoid the DC–AC and
AC–DC conversions stages and reduces the losses.
A detailed discussion and analysis of LVDC are required to see the potential of
DC technology. However, some advantages of it already are being discussed. There
are various advantages including the system efficiency improvement and energy
saving in the residential building. In [16,17], a study of the US-based region was
carried out with different topologies and at various locations of the country [16].
Different topologies have been considered with or without some ESS. The results
of this study show that the efficiency is significantly improved with the storage
device. Annual 5% energy savings are estimated without storage system, and 14%
energy saving is observed with ESS. This difference in energy saving is because of
the consequence of PV generation and consumption time according to the resi-
dential load. The consumption peaks in the afternoon and evening while the peak
PV generation is in the afternoon. Because of this reason, storage system is required
to store the excess energy that is available during the peak PV generation and
utilised during high peak demand. There are some more studies that are showing to
achieve little more energy saving up to 25% energy saving [18,19]. The building
loads, which are affected by environmental conditions such as cooling or heating,
need to be considered for such type of studies. Another condition of comparison is
that the AC or DC comparable loads have to consider for the different AC or DC
distribution systems that have to be compared [20]. One important consideration is
also that energy saving should not be considered which is achieved with extremely
efficient DC loads in the place of AC loads.
As several advantages and features of DC distribution system are discussed,

it also faces critical challenges and some barriers when going to implement it into
residential buildings. There are some challenges and obstacles to the DC distribu-
tion network:
● There is a lack of proper standardisation and codes for the DC system. How-
ever, continuous work is going on in this area. The leads, in these directions,
are taken by some organisations like IEEE, the International Electrotechnical
Commission, Emerge Alliance, the European Telecommunications Standards
Institute, and others, which are already actively developing the necessary
regulation and standards. However, much work is required in this direction.
● Protection schemes for the use of DC systems have to be developed. New
protection devices have to be designed for the safe use of DC systems
[4,21,22].
● Currently, there are fewer industries and devices available those will work well
for the DC distribution systems. There are some rare products available when
going to analyse the DC systems for residential buildings; there is a lack of
products which can be used on DC voltage directly. However, there are many
DC devices that can be used directly on the DC voltage as having internal
conversion unit [4,22].
In respect to energy efficiency and the fulfilment of demand, DC systems taking
ahead with ACDSs. However, for residential use, more time is needed in order to
use it in residential buildings safely with proper standards. Lack of standards,
regulation methods and protection schemes are the main challenges that need to be
discussed.
15.11 Comparison AC and DC residential buildings:

an illustrative example
To investigate the performance of electrical distribution system, an LVDC dis-

tribution network for a residential building supplied by PV including a BB and the
PU has been simulated in MATLAB. The 24 V BB is configured with the series,
and parallel combination of 16 cells of 150 A h. The 1.44 and 2.28 kW with a 24 V
rated voltage PV is considered for the DCDS with DC compatible appliances (DC
residential building), and ACDS with DC compatible appliances (AC residential
building), respectively. The PU is tied to the consumer portal via AC–DC converter
and a step-down transformer (solid-state transformer) to 24 V DC bus and 230 V
AC for AC and DC residential building, respectively. The distribution lines of the
power system are considered as lossless. The power consumption in the AC and DC
residential building is shown in Figure 15.10.
In the case of DC residential building, the building demand curve consists of
power consumption in the DC compatible appliances of the building and power loss
in PU inverter and it is represented by the blue line in Figure 15.10. In the case of
AC residential building, the building demand curve consists of power consumption
2,000
DC residential building
AC residential building
1,500
Power (W)
1,000
500
0
0.00 4.00 8.00 12.00 16.00 20.00 23.59
Time (h)
Figure 15.10 Demand for AC and DC residential building
2,000
PV: DC residential building
PV: AC residential building
1,500 PU: DC residential building
PU: AC residential building
Power (W)
1,000
500
0
0.00 4.00 8.00 12.00 16.00 20.00 23.59
Time (h)
Figure 15.11 Power consumed from PV and PU in AC and DC residential

building
in the DC compatible appliances of the building, power loss in the internal con-
verter of appliances, power loss in the BB and PV converter as represented in
Figure 15.10 by the red line. The building demand always remains higher for AC
residential building compared to the DC residential building.
The PU and PV power curves for AC and DC residential building are shown in
Figure 15.11. The PU and PV power curve remains always lower for the DC resi-
dential building compared to the AC residential building. Different operation
modes are discussed (Mode I–Mode III).
Mode I (PU as a source): The PU supplies the building load during the
0.00–03.00 and 21.30–23.59 h time interval. In the case of AC residential building,
all the building loads are DC compatible and connected via AC–DC internal
converters. The converters loss is the combination of power loss in the internal
converters of the switch ‘ON’ appliances, and power equation can be expressed as
Xao
j¼n

Ppu ðtÞ ¼ paj ðtÞ þ pacj ðtÞ (15.13)
j¼1
where Ppu(t) is the PU power at the time instant t, Paj(t) is the power consumption
in the jth appliance, nao is the number of switched ‘ON’ appliances in the building
and pacj(t) is the power loss in the internal converter of the jth appliance.
Moreover, in the DC residential building, all the building loads are DC com-
patible and directly (without any converter) connected to the 12 V bipolar DC
distribution system. The load is connected to the PU via AC–DC converter. So the
power loss in the converter is the power loss in the PU converter. The power
equation can be expressed as
Xao
j¼n X
j¼ns
Ppu ðtÞ ¼ paj ðtÞ þ pscj ðtÞ (15.14)
j¼1 j¼1
where pscj(t) is the power loss in the jth source converter, ns is the number of power
source converters.
Mode II (BB as a source): The BB supplies the building load during the
03:01–07:21 and 19:51–20.00 h time interval as represented in Figure 15.12. In the
AC residential building, all the building loads are DC compatible and connected via
AC–DC internal converters. The converters losses are a combination of power loss
in the internal converters of a switch ‘ON’ appliances and the BB inverter. The
power equation can be expressed as
X
j¼na
Xj¼ns
Pbb ðtÞ ¼ Padj ðtÞ þ padcj ðtÞ þ pscj ðtÞ (15.15)
j¼1 j¼1
1,500
1,000 AC residential building
500
Power (W)
–500
–1,000
–1,500
0.00 4.00 8.00 12.00 16.00 20.00 23.59
Time (h)
Figure 15.12 Battery bank power in AC and DC residential building

In the case of DC residential building, all the building loads are DC compa-
tible. The appliances and DC power source (i.e. BB) are directly (without any
converter) connected to the DC bus. Therefore, the power loss in the sources con-
verters and internal converter of the appliances remain zero in this mode as
represented by a green line in Figures 15.13 and 15.14, respectively. The power
equation can be expressed as:
X
j¼na
Pbb ðtÞ ¼ Paj ðtÞ (15.16)
j¼1
Mode III (BB and PV): The PV generates the power during the 07.22–19.50 h
time interval. The BB balances the power by supplying the surplus load and
absorbing the surplus PV power. In the case of AC residential building, all the
400
300
Power (W)
200
100
0
0.00 4.00 8.00 12.00 16.00 20.00 23.59
Time (h)
Figure 15.13 Conversion losses in AC and DC residential building
200
150
Power (W)
100
50
0
0.00 4.00 8.00 12.00 16.00 20.00 23.59
Time (h)
Figure 15.14 Conversion power losses in internal converters loads in AC and

Table 15.5 Description of energy consumption in the building
System Demand Conversion Total energy PV energy PU energy

(kW h) loss (kW h) consumption (kW h) (kW h)
(kW h)
AC residential 12.47 3.34 15.81 11.38 4.43
building
DC residential 11.12 0.44 11.56 7.2 4.36
building
building loads are DC compatible and connected via AC–DC internal converters.
The converter losses are the combination of power losses in the internal converters
of a switch ‘ON’ appliances including with PV and BB inverter. The power
X
j¼na
Xj¼ns
Ppv ðtÞ Pbb ðtÞ ¼ Padj ðtÞ þ padcj ðtÞ þ pscj ðtÞ (15.17)
j¼1 j¼1
In the case of DC residential building, all the building loads are DC compatible.
The DC appliances and DC power source (i.e. BB and PV) are directly connected to
the DC bus without any converter. Therefore, the power loss in the source con-
verters and internal converter of appliances remains zero in this mode as repre-
sented by the green line in Figures 15.13 and 15.14, respectively. The power
X
j¼na
Ppv ðtÞ Pbb ðtÞ ¼ Paj ðtÞ (15.18)
j¼1
The energy consumption, conversion losses, including energy supplied by PV and

PU can be found in Table 15.5. The energy demand and conversion loss for the AC
residential building are higher than the DC residential building. The conversion
losses in the AC residential building is approximately 7.5 times higher than the
conversion losses in the DC residential building.
15.12 Conclusions
In this chapter, the concept of DC microgrid for the residential buildings has been
discussed. Comparison of AC vs DC system and DC microgrids architecture has
been discussed. The distribution topologies discussed in this chapter are very
helpful to understand the most efficient way of interconnection. The data of energy
dissipated in DC appliances, and the cable cost data representing are used to show
the correlation of different parameters associated with the losses. This chapter
demonstrates different configurations for both the ACDS and DC distribution
system. A power system control strategy based approach is used for the voltage
standardisation. This approach enables the development of energy efficient econ-

omy and flexible LVDC system and as well as medium-voltage standardisation.
A comparative analysis of AC and DC residential building shows the superiority
DC residential building, in respective energy saving. Simulation results also show
that the power consumed in the DC residential building is less than the power
consumed in the AC residential building. Because the converter stages and con-
version losses are much less in the DC residential building compared to the AC
residential building.
References
[1] F. Gonzalez-Longatt, B. S. Rajpurohit, and S. N. Singh, ‘‘Optimal structure

of a smart DC micro-grid for a cluster of zero net energy buildings,’’ in Proc.
IEEE International Energy Conference (ENERGYCON), Leuven, Belgium,
Apr. 4–6, 2016.
[2] F. Gonzalez-Longatt, B. S. Rajpurohit, and S. N. Singh, ‘‘Smart multi-
terminal DC micro-grids for autonomous zero-net energy buildings: implicit
concepts,’’ in Proc. IEEE Innovative Smart Grid Technologies-Asia (ISGT
ASIA), Bangkok, Thailand, Nov. 04–06, 2015, pp. 1–6.
[3] R. K. Chauhan, B. S. Rajpurohit, S. N. Singh, and F. M. Gonzalez-Longatt,
‘‘DC grid interconnection for conversion losses and cost optimization,’’ in
Renewable Energy Integration: Challenges and Solutions, J. Hossain and
A. Mahmud, Eds. Singapore: Springer Singapore, 2014, pp. 327–345.
[4] L. Mackay, L. Ramirez-Elizondo, and P. Bauer, ‘‘DC ready devices – is
redimensioning of the rectification components necessary?,’’ in Proc. 16th
International Conference on Mechatronics – Mechatronika, Dec. 03–05,
2014, pp. 1–5.
[5] F. Katiraei, R. Iravani, N. Hatziargyriou, and A. Dimeas, ‘‘Microgrids
management,’’ IEEE Power and Energy Magazine, vol. 6, no. 3, pp. 54–65,
May 2008.
[6] X. Fang, S. Misra, G. Xue, and D. Yang, ‘‘Smart grid – the new and
improved power grid: a survey,’’ IEEE Communications Surveys & Tutor-
ials, vol. 14, no. 4, pp. 944–980, Dec. 2011.
[7] J. Stamp, ‘‘The SPIDERS project – smart power infrastructure demonstra-
tion for energy reliability and security at US military facilities,’’ in Proc.
2012 IEEE PES Innovative Smart Grid Technologies (ISGT), Jan. 16–20,
2012.
[8] R. K. Chauhan, and B. S. Rajpurohit, ‘‘DC distribution system for building,’’
in Proc. IEEE, 18th National Power System Conference, IIT Guwahati,
India, Dec. 18–20, 2014, pp. 1–6.
[9] D. Nisbet and H. Thiesen, ‘‘Review of the initial phases of the LHC power
converter commissioning,’’ in Proc. 11th European Particle Accelerator
Conference, Jun. 2008.
[10] R. K. Chauhan, B. S. Rajpurohit and N. M. Pindoriya, ‘‘DC power dis-

tribution system for rural applications,’’ in Presented at the 8th National
Conference on Indian Energy Sector, Ahmedabad, India, Oct. 11–12, 2012,
pp. 108–112.
[11] R. K. Chauhan, B. S. Rajpurohit, R. E. Hebner, S. N. Singh, and
F. Gonzalez-Longatt, ‘‘Voltage standardization of DC distribution system
for residential buildings,’’ Journal of Clean Energy Technologies, vol. 4,
no. 3, pp. 167–172, 2016.
[12] I. Z. D. M. Larruskain, O. Abarrategui, and Z. Aginako, ‘‘Conversion of ac
distribution lines into dc lines to upgrade transmission capacity,’’ Electric
Power Systems Research, vol. 81, no. 7, pp. 1341–1348, Jul. 2011.
[13] R. O. J. Kensby, ‘‘Building automation systems design,’’ Master Thesis,
Division of Building Services Engineering, Chalmers University of Tech-
nology, Göteborg, Sweden, 2012.
[14] R. Kumar, K. Saini, and M. L. Dewal, ‘‘Intelligent SCADA system,’’
International Journal on Power System Optimization and Control, vol. 2,
no. 2, pp. 143–149, 2010.
[15] M. Saeedifard, M. Graovac, R. F. Dias, and R. Iravani, ‘‘DC power systems:
challenges and opportunities,’’ in Proc. of IEEE PES General Meeting,
Providence, RI, USA, Jul. 25–29, 2010, pp. 1–7.
[16] K. G. V. Vossos, and H. Shen, ‘‘Energy savings from direct-DC in U.S.
residential buildings,’’ Energy and Buildings, vol. 68, no. A, pp. 223–231,
Jan. 2014.
[17] E. Vossos, ‘‘Optimizing energy savings from direct-DC in U.S. residential
buildings,’’ Master’s Thesis, The Faculty of the Department of Environ-
mental Studies San Jose State University, San José, USA, 2011.
[18] R. R. N. P. Savage, and S. P. Jamieson, ‘‘DC microgrids: benefits and
barriers,’’ in From Silos to Systems: Issues in Clean Energy and Climate
Change, pp. 51–66, 2010.
[19] V. V. Karina Garbesi, and H. Shen, ‘‘Catalog of DC Appliances and Power
Systems,’’ Lawrence Berkeley National Laboratory (Berkeley Lab) 2011,
vol. LBNL-5364E.
[20] E. Rodriguez-Diaz, M. Savaghebi, J. C. Vasquez, and J. M. Guerrero,
‘‘An overview of low voltage DC distribution systems for residential
applications,’’ in Proc. IEEE 5th International Conference on Consumer
Electronics-Berlin (ICCE-Berlin), 2015, pp. 318–322.
[21] D. Salomonsson, L. Soder, and A. Sannino, ‘‘Protection of low-voltage
DC microgrids,’’ IEEE Transactions on Power Delivery, vol. 24, no. 3,
pp. 1045–1053, Apr. 2009.
[22] G. Makarabbi, V. Gavade, R. Panguloori, and P. Mishra, ‘‘Compatibility and
performance study of home appliances in a DC home distribution system,’’
in Proc. of IEEE International Conference on Power Electronics, Drives and
Energy Systems (PEDES), Dec. 16–19, 2014, pp. 1–6.
Chapter 16
DC microgrids for photovoltaic powered systems
Rasool Heydari1, Saeed Peyghami1, Yongheng Yang1,
Tomislav Dragičević1, and Frede Blaabjerg1
16.1 Introduction
Climate changes induced by the use of fossil fuels and also the shortage of such
energy have been the major driving forces to develop environmental-friendly
renewable energy systems [1,2]. Many attempts have been made to advance
renewable energies, two of which are wind power and solar photovoltaic (PV)
power that are still the most favorable, as it is seen in Figure 16.1. Depending on
the power ratings, PV systems are mostly installed at a residential-scale, while wind
power systems are connected to medium-voltage (MV) or high-voltage (HV) power
grids. Hence, for microgrid applications, PV energy is more feasible also due to its
high scalability. The still declining price of PV panels is another reason for a
massive deployment of PV microgrid systems. Notably, in those applications,
power electronic converters (e.g., dc–dc converters and dc–ac inverters) are the
key, which converts the dc power from PV panels to another type of source (e.g., dc
at a different level or ac power with a fixed amplitude and a constant frequency).
Additionally, the high penetration rate of dc loads including energy storage
(ES) systems (e.g., batteries and electric vehicles) makes dc microgrids a promising
solution for energy integration [3,4]. For one thing, in the context of a large-scale
utilization of dc loads, the dc microgrid architecture eliminates at least one con-
version stage (i.e., ac–dc rectification), and thus, the entire system efficiency
maybe is improved. For another thing, it simplifies the system control, and the
frequency stability is addressed to some extent, which, however, is a bigger chal-
lenge in ac power systems. Yet, despite the above compelling reasons to deploy
PV microgrids, issues are also encountered [4]. For instance, the PV power is
fluctuating, possibly leading to dc bus voltage variations, being harmful to sensitive
equipment. Nevertheless, employing dc microgrids has been increasingly
acknowledged in recent years, and a great deal of the focus has been put on relia-
bility, efficiency, and cost-effectiveness, i.e., reducing the cost of PV energy.
In practice, dc microgrids are utilized in various sectors, ranging from small-scale
1
466.5 GW
Hydro
Marine energy
109.7 GW
Wind energy
Solar energy
Bioenergy
1,243 GW 295.7 GW Geothermal
Figure 16.1 Global renewable power capacity composition in 2016 (more than
2,000 GW in total) [2], where hydro power also includes pumped
storage and mixed plants; solar energy includes solar photovoltaic
(290.8 GW) and concentrated solar power (4.7 GW)
residential to medium-scale industrial applications including power generation

and/or distribution systems, space stations, maritime application, data centers, oil
drilling platforms, remote regions (the energy accessibility is poor), and residential
smart homes.
Many aspects should be considered in the design phase of dc microgrids.
Efficiency, reliability, and stability are the major concerns as aforementioned. In
respect to the system efficiency, the system architecture or layout and the power
converter technologies should be considered. This chapter first overviews the dc
microgrid architecture, where the primary energy sources are PV panels. Then, the
focus is on power electronic converter technologies for PV dc microgrid systems,
where an industrial case is also given. Regarding the control dc microgrids for PV
collection plants, it is discussed in this chapter as well. A design example is also
provided.
16.2 Architecture of dc microgrids

In the following, some outstanding applications and architecture schemes of dc
microgrids are introduced.
16.2.1 MVDC and LVDC microgrid for dc distribution systems

Recently, due to the fast pace of the commercialization of distributed energy
resources, a wide range of power distributed generation (DG) and ESs have been
employed in power systems. Since it is known that a microgrid is dynamically
separated from the main grid, the future of power systems relies on isolated ac and
dc microgrids. Compared to the ac grids, dc microgrids bring various advantages,
as mentioned in the previous section. For instance, it is easy to interface dc loads
and renewable energy resources, in particular, PV panels. And there are no issues
DC microgrids for photovoltaic powered systems 391
HVDC
ac Utility system Fuel cell PV panels Wind farms
ac dc dc dc ac
dc dc dc dc dc
LVDC–MVDC
dc dc dc dc dc dc
dc dc dc dc dc ac
Sensitive ac Loads
Energy loads
dc Industrial dc Data Electric
facility centers vehicle storage
Figure 16.2 General LVDC–MVDC unipolar architecture layout
like frequency stability, reactive power compensation, and harmonics. More

importantly, high efficiency, power quality, and reliability make the dc microgrid
more and more popular [5–7].
A majority of renewable energy resources such as PV arrays and fuel cell
modules as well as battery and ESs are naturally dc. Furthermore, by utilizing
advanced power electronic devices, domestic and industrial dc loads are increasing
in recent years. Therefore, utilizing dc microgrids to integrate dc loads and dc-
renewable energy resources increases the efficiency and reliability of the distribu-
tion systems. Figure 16.2 shows general layout of an LVDC–MVDC distribution
system, in which renewable energy resources (e.g., wind turbines, solar panels, and
fuel cells) and dc loads are connected to a common dc-bus through dc–dc or ac–dc
converters. The dc microgrid is employed in the distribution system to increase the
controllability of DGs, which are typically installed in the vicinity of loads. It thus
leads to increased reliability, reduced transmission losses, and consequently
decreased operation and planning costs.
A typical bipolar LVDC microgrid structure is shown in Figure 16.3 [8]. This
topology introduces better reliability in comparison to the unipolar one. In this
topology, loads, sources, and storage can be connected to the different lines.
A typical voltage for unipolar system can be 170, 380, and 750 V based on the
required application.
Storage (battery) PV panels
dc dc
dc dc
+170 V 3-phase 200 V
ac Utility ac balancer dc
Voltage
0V
dc ac
–170 V
Single-phase
dc 100 V
dc dc
ac
dc 12-48 V
dc
dc
Fuel cell
Figure 16.3 A typical bipolar LVDC distribution systems [8]
LVDC 70 V Distribution
dc Power
dc and
control
protection
PV Storage Essential,
panels (battery) semi-essential,
nonessential
loads
Figure 16.4 Power system structure in a space station
16.2.2 LVDC microgrid for space applications

Generally, a space station comprises various devices designed to address air craft
missions. In such applications, the required power in early satellites is in the order
of several watts, while, in today’s space stations, it is rated at several kilowatts.
Furthermore, the estimated power demand in some strategic defense space crafts is
expected to be at the order of 100-MW level and it is still increasing. The basic
structure of a space station power system is demonstrated in Figure 16.4, where
the primary energy resources are mainly solar panels, nuclear reactors, and
radioisotopes.
Solar PV panels, batteries, and dc–dc converters have been the major infra-
structure in space power systems since the beginning. PV arrays supply the loads
and charge the batteries. If the generated power by the PV arrays is not sufficient,
batteries are then enabled to power the loads. Furthermore, dc–dc converters are
employed to protect the space station grid and regulate the voltage. These com-
ponents have formed an on-board dc microgrid in the space station. Since high
reliability and flexibility are necessary for missions in space stations, reliable dc
microgrid with high robustness in both hardware and software is required and a
promising solution [9,10].
The loads in space crafts can be categorized into three groups: essential, semi-
essential, and nonessential loads. dc Microgrids in space applications should solely
supply the on-board loads. Therefore, nonessential loads are typically adopted to
be sheded during critical situations to regulate the voltage. It should be noted that
the dc microgrid employs special protection systems to maintain the stability in
space stations.
16.2.3 MVDC microgrid for marine applications

Shipboard microgrids are another type of reliable microgrids, which have no access
to the main grid. The shipboard power system and an islanded microgrid have a
reasonable analogy in terms of supplying loads with local generation units [11].
The control structure, volume, and weight of shipboard are some major concerns in
the design of shipboard microgrids, and there are important issues to deal with
vitally imposing pulse loads. Some of the loads are intermittent, operating at a
time-scales that can be milliseconds or even less, and the load power can be various
from a few kWs to MWs, which may reach 90% of the installed power capacity in a
short period [12]. Furthermore, the load priority can vary, depending on the mis-
sions of the ship or operating conditions [13]. For instance, a load may be crucial
in a cruising mission, while negligible in another case. Thus, the prioritization of
loads and dc microgrid control structures, at runtime, are important in shipboard
microgrids. Figure 16.5 illustrates an industrial shipboard dc microgrid structure for
PV panels PV panels
dc
dc
dc
ac Link ac Link dc
Generator Generator
ac dc
dc dc
dc ac
ac ac
MVDC 5–38 kV dc Link
dc dc dc dc
ac ac ac ac
Motor Motor
Motor Motor
Figure 16.5 dc Microgrid for marine power applications

marine applications [14]. With such structures, not only the operation cost is
reduced but also the maintenance of engines is lowered due to more efficient
operation. Additionally, dynamic response enhancement and maneuverability are
advantages of dc microgrids for marine applications.
16.2.4 LVDC microgrid for data centers

Figure 16.6 shows a typical data center structure consisting of a dc bus, loads,
batteries, and power electronic converters. In order to supply sensitive loads in data
centers, a local generation unit is implemented. The best solution for data centers to
integrate local generation units and reliable power is to employ dc microgrid
technologies. In such cases, dc microgrids can operate in the islanded mode and be
connected to the grid simultaneously. The main challenge of data centers is to
supply the loads with a high power quality and availability. Practically, data center
loads are sensitive to voltage and current changes, and hence, a fast, accurate, and
reliable control structure is the first requirement for data centers [15]. Furthermore,
ESs (batteries) and spinning reserve are also important for reliable operation of dc
microgrids in data centers. On the other hand, the required cooling system leads to
high power consumption in data centers. Nevertheless, when utilizing LVDC
ac Utility
ac
Storage dc
(battery)
PV panel
Charge bus
Discharge bus
Storage LVDC 380 V
(battery)
PDU
MNS/RPP MNS/RPP
dc/dc dc/dc
PSU PSU
Load Load
Figure 16.6 LVDC microgrid architecture for data center. PDU, power
distribution unit; RPP, remote power panel; PSU, power supply unit;
MNS, switchgear
microgrids, ES devices, renewable energy resources, and dc loads can be integrated

to form a relatively high reliable power system [16].
16.2.5 LVDC microgrid for homes

Residential dc microgrids enable a distributed power system in a flexible manner
for a smart home. In future smart homes, dc loads and electrical vehicles will be
massively adopted and coordinated to achieve an optimal residential operation. As
it can be seen in Figure 16.7, dc loads are the main load of future smart homes,
which are connected to the dc bus via power electronic devices. In the past decade,
approximately 80% of the residential power demand—from cell phones and LED
lightening to air conditioning systems—are processed through power electronic
converters and thereby dc-powered. Such a dc microgrid increases the overall
reliability and decreases power losses in green homes by eliminating a conversion
stage from dc to ac [16,17]. Furthermore, the integration of renewable energy
resources and ESs into LVDC-microgrids makes it possible for zero-net-energy
buildings.
ac Utility Electric Fuel cell PV panels Wind turbine

vehicle
ac dc dc dc ac
dc dc dc dc dc
LVDC 380 V
dc dc dc dc dc
ac dc dc dc dc
ac Loads
TV,
LED light
Washer, computer
Stove dryer
Figure 16.7 A residential dc microgrid in a future smart home, where the arrow
indicates the power flow direction
A residential dc microgrid typically has two voltage levels, i.e., low voltage at
24 and 380 V. The 24-V system powers LED lighting and small appliances, while
high-power heating equipment is supplied by the 380 V. In such a residential dc
microgrid, the power is fed from devices and controllable converters, which can
provide a wide range of voltages as well as current limitations (protections).
Therefore, electromechanical protection devices can be eliminated and the system
safety can be maintained.
16.2.6 MVDC microgrid for oil-drilling applications

Off-shore oil and gas drilling systems are heavy-industrial applications. Drilling,
station keeping, propulsion, and pumping products are the main applications of
drilling rigs. In such a heavy energy-demand, the power generation and conversion
are major barriers to the optimal operation of oil drilling stations. On the other
hand, prior-art investigations envision the potential deployment of large wind farms
along the shore, where drilling stations are installed. That is, the MV-dc-microgrid
architecture with a high penetration of wind turbines can serve as a link to harvest
energy for off-shore oil-drilling rigs. Figure 16.8 shows a dc microgrid for oil-
drilling applications with collocated wind farms and PV panels [18]. It is noticed in
Figure 16.8 that the main grid can be connected, and in that sense, it can be con-
sidered as an onshore system [18,19].
ac Utility PV panels Wind turbine PV panels Wind turbine
ac dc ac dc ac
dc dc dc dc dc
MVDC Common dc-bus
Bidirectional dc–dc
converter
dc dc dc
ac ac ac
M M M
Figure 16.8 MVDC structure utilized in an offshore oil-drilling application

integrated with wind farms
16.3 dc Microgrids for photovoltaic power plants

The output power of PV panels is dc, and hence, the operation of PV power plants
in dc microgrids may introduce high efficiency and reliability. As the output
power of PV panels is intermittent and varies according to weather changes, uti-
lizing a battery storage will increase the system availability. Depending on the
combination of PV panels and batteries, two types of dc microgrids are developed:
(1) hybrid PV-battery system and (2) distributed PV panel and batteries. In the
first structure, as shown in Figure 16.9, a battery storage unit is locally connected
to the PV source, and then both are connected to the microgrid. In this case, the
entire unit including PV panels and batteries is a unified dispatchable unit. If the
maximum power of PV panels (typically achieved by a maximum power point
tracking algorithm) is higher than the load demand, the excessive power will
charge the battery. Furthermore, once the load power exceeds the PV maximum
power, the battery is deployed to supply the load. Notably, a power management
system is required for properly sharing power among the clusters of the hybrid
PV-battery unit.
On the other hand, in the distributed PV and battery, every PV unit and battery
storage are connected to the microgrid through a power converter, as shown in
Figure 16.10. In this case, the maximum power from PV plants is used to charge the
battery and supply the load. Hence, they can be modeled as current sources.
Here, the dc-link voltage will be formed by the batteries. Furthermore, a super-
visory control is required to monitor the state-of-charge (SoC) level of batteries and
also to manage the output power of PV panels. Once the microgrid demand
goes below the maximum power of PV units, the maximum power tracking control
of PV panels should be adjusted (e.g., reduced power control) to ensure the
demand–supply balance in the microgrids. The control system of PV panels in the
PV panels Battery PV panels Battery
dc dc dc dc
dc dc dc dc
Local dc-bus
dc dc
dc dc
Common dc-bus
Loads
Figure 16.9 dc Microgrid structure for hybrid PV-battery units

PV panels PV panels Battery Battery
dc dc dc dc
dc dc dc dc
Loads
Figure 16.10 dc Microgrid structure for distributed PV and battery units
two PV-integrated microgrids will be explained after the modeling of PV systems

in this chapter.
16.4 PV system modeling

With the high penetration of renewable energy resources, academic and industrial
interest in utilizing PV and solar energy has been increased. Due to the inter-
mittency, low efficiency, and high manufacturing cost of PV panels, there are
challenges when integrating PV energy generation units into dc microgrids.
Detailed modeling, analysis, and simulation of PV systems are the prerequisite to
design an efficient PV microgrid and properly address the aforementioned chal-
lenges. In the practical model of PV modules, different types of partial shading, and
the effects of temperature, irradiance, and series resistance have to be scrutinized.
In the following, a theoretical model is presented for PV panels to be used for
modeling and control.
16.4.1 Ideal model of a PV cell

The PV cells modeling involves P–V (output power–output voltage) and I–V
(output current–output voltage) characteristic curves. Most ideal PV cells have
been modeled as a p–n junction with a Shockley-based equivalent circuit. Fig-
ure 16.11 shows the basic PV model, which comprises of a diode and an ideal
current source.
The mathematical description of the I–V characteristic of the PV model is
I ¼ IPh ID ; (16.1)
where I is the output current of the PV cell, Iph is the photo-generated current by the
sunlight that is proportional to the solar irradiance level, and ID is the Shockley
current modeling the diode behavior as

ID ¼ I0 eqVD =akT 1
qðV þIÞ=akT ; (16.2)
I ¼ IPh II0 e D 1
Iph D V
ID
Iph ID I
V V V
Figure 16.11 Ideal model of a PV cell. Iph, photo-generated current; I, PV cell

current; ID, diode current; V, PV voltage
I
Rs I
Rs = 0
Iph D V Rs > 0
ID
(a) (b) V
Figure 16.12 Single-diode PV model with series resistance (Rs represents the
series resistance): (a) electrical circuit and (b) V–I characteristic
I is a reverse bias current of the diode, k is the Boltzmann’s constant, q is the charge
of an electron (¼ 1:602 1019 C), and T refers to the absolute p–n junction
temperature in Kelvin, and a is the diode ideality factor typically within 1–2 [20].
Nevertheless, this simplified model cannot demonstrate the real behavior of a PV
cell. Thus, a single-diode model with series resistance has to be introduced.
16.4.2 Single diode with series resistance model

By utilizing series resistance as shown in Figure 16.12(a), the current through base
and emitter of solar cell is modeled. Furthermore, the series resistance can model
the interconnection losses between silicon and metal contacts, as well as top and
rear metal contacts [21,22]. The effect of series Rs on the output I–V characteristic
is shown in Figure 16.12(b).
Therefore, the I–V characteristic in this model can be described as:
h i
I ¼ IPV I0 eqðVD þIRs Þ=akT 1 ; (16.4)
where Rs refers to the series resistance. In order to model the behavior of a PV cell
in high-temperature conditions and manufacturing defects, an additional shunt
resistor is also applied to the model [23].
16.4.3 Single-diode with shunt resistance model

Figure 16.13 shows the single-diode model with shunt resistance. The shunt resis-
tance explains the manufacturing defects to some extent. The effect of the shunt
resistance is more significant at a weak irradiance level due to the decreased photo-
generated current. The electrical model of the PV cell with series and shunt resis-
tance is shown in Figure 16.13(a). The effect of the shunt resistance on the output
current is illustrated in Figure 16.13(b).
Therefore, the output current of the PV cell shown in Figure 16.13 can be
obtained as
h i V þ IR
I ¼ IPV I01 eqðVD þIRs Þ=akT 1
s
; (16.5)
Rsh
where Rsh refers to the shunt resistance. However, this model cannot show the
accurate behavior of PV cells at low solar irradiance. Therefore, the double-diode
model was presented in the literature to model the PV cell characteristics at dif-
ferent operation conditions, e.g., weak solar irradiance.
16.4.4 Double-diode model

Equation (16.5) shows a single-diode model with the ideality factor a. However,
the ideality factor depends on the voltage across the device. Thus, at a HV, a ¼ 1
and at a low voltage, a ¼ 2. By adding a parallel diode to the model in
Figure 16.13, a relatively accurate PV cell model can be achieved, as shown
in Figure 16.14. In order to include the effect of diode voltages in the model, two
diodes with a ¼ 1 and a ¼ 2 are considered.
The comprehensive model shown in Figure 16.14 can be described by
h i h i V þ IR
I ¼ IPV I01 eqðVD þIRs Þ=a1 kT 1 I02 eqðVD þIRs Þ=a2 kT 1
s
; (16.6)
Rsh
I
Rs I
Iph D Rsh V
Rsh
ID IRsh Rsh finite
(a) (b) V
Figure 16.13 Single-diode model with shunt resistance (Rsh represents the shunt
resistance): (a) electrical circuit and (b) V–I characteristic
Rs I
Iph D1 D2 Rsh V
ID1 ID2 IR
sh
Figure 16.14 Double-diode model for PV cells
where a ¼ 1 and a ¼ 2. Hence, the diode (D1) in Figure 16.14 models the effect of
the HV and the diode (D2) models the effect of the low voltage. A multiple-diode
model is also presented in the literature which increases the model complexity.
As the power of a single PV cell is very low, a great deal of PV cells should be
connected in series and parallel to form a PV module/panel up to several hundred
watts. In a similar way, the PV modules/panels can be connected in series and
parallel to further achieve high-power and HV PV arrays. The number of series
and parallel panels in each PV array is dependent on the P–V characteristics of each
panel, the input voltage, and the rated current of the interfacing converter.
Furthermore, as the output power of PV panels varies following the solar
irradiance and ambient temperature, an interfacing converter is required to control
the output current of the PV panels to harvest the maximum power. This is called a
maximum power pointing tracking (MPPT) control. In the next section, different
dc–dc converters for PV applications are presented.
16.5 Power converter technologies
Generally, dc–dc converters can be implemented between PV panels and the dc-bus
or a dc–ac converter in order to control the output voltage/current of the PV panel
for power optimization. Thus, different power converter topologies for isolated and
nonisolated converters and various control strategies have been presented [24–26].
In the following, transformerless and transformer-based (typically, high-frequency
transformers) topologies are explained. Since this chapter focuses on dc microgrids,
some dc–dc converters are introduced in this section. Furthermore, some com-
mercial inverter topologies will be introduced, where the first conversion stage can
be used in dc microgrids as the interlinking between PV panels and the dc-bus.
16.5.1 Transformerless topologies

Transformerless topologies offer a low-cost and high-efficient solution for different
PV structures. In order to perceive the foundation of transformerless technologies, a
simple single-phase boost converter will first be presented. The simple but the most
common dc–dc converter for PV applications is the conventional boost converter as
shown in Figure 16.15. It can be seen that there are a parasitic capacitance (Cpv ),
a filter inductor (Lb ), and a switch (S) representing a common mode behavior of the
conventional boost converter. In this topology, the output voltage is higher than the
IPV Lb
PV strings/modules
Cpv S D
Cdc Vdc
o
Cp
Figure 16.15 Conventional boost converter for PV applications
S Db
Lb Df
Cdc1
IPV
PV strings/modules
Vdc
Cpv Cdc2
o
Cp
Figure 16.16 Parallel input, series output bipolar dc–dc converter for PV
applications
input voltage. The input voltage depends on the PV characteristics and can be
determined by an MPPT control system to obtain the maximum power. Further-
more, the output voltage is almost a constant value, which can be controlled by the
microgrid.
In order to increase the efficiency of the boost converter, a dc–dc converter
with a series output and a parallel input can be used. This structure can be seen in
Figure 16.16. It is worth to note that the voltage step-up of this boost converter
structure is also enhanced [27].
Boost converters can be used as an interface between PV panels and dc
microgrids and/or between PV panels and inverters. Furthermore, a boost converter
is employed in industrial PV inverters as the first stage in the entire energy con-
version process to control the PV panels to operate at MPPT.
For example, Figure 16.17 shows a commercial topology of a boost converter,
introduced in SMA Sunny Boy 5000 TL.
In order to increase the voltage level and standardization of the output PV
voltage, safety, transformer-based topologies, which isolate the PV side-circuit
from the grid side, have been presented.
PV string 1 Vdc
IPV Lb Db
Cpv S Cb
IPV Lb Db
PV string 2
IGrid
S
Cpv S Cb
LGrid
Grid S
IPV Lb Db
PV string 3
Cpv S Cb
Figure 16.17 SMA Sunny Boy 5000 TL dc–dc converters
IPV
S S
PV string
Cpv
Vdc
HFT
S S
Figure 16.18 Typical transformer-based for dc converters for PV applications.

HFT, high frequency transformer
16.5.2 dc Converters with high-frequency transformers

Figure 16.18 shows a basic isolation transformer-based converter for a PV string.
By utilizing a high-frequency transformer, the galvanic isolation from the
dc-bus is provided. The active full-bridge converter controls the PV string to
operate at MPPT, and the voltage level can be increased by the transformer.
Vdc
IPV Cb
D1 D2
S1 S2
PV string 1
Cdc
Cpv
S4 HFT
S3
D3 D 4
IPV Cb D1 D2
S1 S2 S1 IGrid S3
PV string 2
Cdc LGrid
Cpv S4
S2 Grid
S4 HFT
S3 D3 D 4
Cb
IPV D1 D 2
S1 S2
PV string 3
Cdc
Cpv
HFT
S3 S4
D3 D 4
Figure 16.19 A 4.5-kW inverter system from Powerlynx
The full-bridge diode-based converter rectifies the output voltage of the trans-
former to a dc voltage.
Figure 16.19 shows a commercial high-frequency transformer-based dc–dc
converter designed by Powerlynx. In this application, dc–dc full-bridge transformer-
based converters are employed as an interface converter between PV modules/strings
and the dc–ac inverter.
Another transformer-based topology, i.e., the push–pull converter, is shown in
Figure 16.20, which is one of the most economic dc–dc converter configurations.
Minimum number of switching devices, the galvanic isolation, and the simple
design are the main reasons making this topology more popular.
16.5.3 Next generation PV converters

Due to the electric codes, material isolation, and safety issues, the maximum dc
input voltage of PV converters is limited to 600 V in North America and 1,000 V in
D1 D3 D1 D3
Cdc Vdc Cdc Vdc

Lb Lb
PV string
PV string
HFT HFT
S1 S2 D4 D2 S1 S2 D4 D2
(a) (b)
Figure 16.20 Single-phase, push–pull topology converter: (a) with hard switching
operation and (b) with clamped snubber circuit
Europe [28–30]. This voltage determines the maximum number of PV modules in

each string. Compared to 600-V PV systems, a PV system with the maximum dc
voltage of 1,000 V will decrease the required components of the balance-of-system
(BOS) (e.g., combiner boxes, dc wiring, and converters) and installation efforts,
which thus contributes to reduced overall system cost and increased efficiency.
In today’s market, commercial PV converters for utility-scale applications are
mostly based on 600 and 1,000-V technologies. However, in order to further reduce
the cost of PV power plants and increase the system efficiency, the industry is
moving toward a higher voltage level of 1,500 V, where the BOS cost can be
reduced [30,31]. As aforementioned, employing the 1,500-V design lowers wiring
power losses in ac and dc sides, transformers, and inverters by 1.5%–2%. Another
advantage of these systems is the design cost reduction due to less BOS compo-
nents for 1,500-V systems. For example, in the case of 155-V PV modules, the
number of modules in a string will be expanded from around 22 in 1,000-V systems
to 32 in 1,500-V systems. Therefore, for the same power rating, the number of
strings, combiner boxes, and dc-side cables will be decreased. Consequently, the
less volume of equipment due to a higher power density will reduce the costs of
installation, transportation, and maintenance.
Besides the above advantages of 1,500-V systems, there are many new chal-
lenges to be addressed including PV modules and related components (market
availability at the 1,500-V level), raw materials, certifications, regulations, and
safety. In the case of 1,500-V dc voltage, the main challenges for PV modules are
electric safety and potential-induced degradation, which may weight down the
entire benefits of 1,500-V. Furthermore, increasing the dc voltage from 1,000 to
1,500 V brings new challenges from the insulation point of view. Hence, con-
nectors, circuit breakers, protection devices, and switching power supplies require
withstanding higher voltage. Finally, a comprehensive electric code update or
revision is needed to certificate all the components employed in 1,500-V dc sys-
tems. Nevertheless, many attempts by the industry have been made to dissolve the
above barriers so that HV (e.g., 1,500-V) PV systems are expected to improve the
entire system economics [28]. Figure 16.21 exemplifies the general configuration
of 1,500-V PV systems.
PV modules
ac Utility
1,500 V
dc
ac
Transformer
Combiner box
Figure 16.21 PV power systems for utility-scale applications with a high dc

voltage (i.e., 1,500 V), where the grounding of PV modules is not
shown. In this case, each PV string can reach to a voltage of
1,500 V and multiple strings can be adopted
16.6 Control of dc microgrids for PV collection plants
Microgrid brings an infrastructure to integrate renewable resources into the grid in

order to increase the power system reliability and availability [32]. Hence, both
grid-connected and islanded operation should be viable to reach the aforementioned
aims [7]. As previously mentioned, PV units can be merged with battery units and
then both can be connected to the dc grid through power converters. In this case, the
entire unit can be seen as a dispatchable unit. Therefore, droop-based controllers
can be employed for each unit. Here, the PV unit operates at MPPT and the demand
can be supplied by the PV units and batteries. If the demand is lower than the PV
power, the excessive power will charge the batteries; however, for the condition,
the demand is higher than the PV power, and the battery will support the remaining
load demand.
Furthermore, in most cases, PV units and batteries are distributed through the
dc grid and individually connected to the grid by its own converters. In this con-
dition, the PV units are not dispatchable, and hence, they cannot fully support
the grid. It is obvious that in this case, the PV units cannot operate in the MPPT
mode once the generation power is higher than the microgrid demand. Therefore,
the microgrid power management system has to control the output power of PV
units. There are some supervisory controls to monitor and control the power bal-
ance in the microgrids with MPPT-based units like PV panels in order to change the
operation mode from MPPT to a constant power mode [33]. Furthermore, droop-
based autonomous approaches have been proposed for MPPT-based units. Some
power sharing approaches for PV-based dc microgrids have been presented
[32,34,35].
For a typical dc microgrid with PV and battery units, a typical droop char-
acteristics can be considered as shown in Figure 16.22. Following Figure 16.22(a),
PV units can operate in two modes, i.e., the MPPT mode in which all the PV
modules generate an MPPT power, and the droop-controlled mode in which the PV
Voltage Voltage
SoC range
SoC 100%
SoC 0%
Rated current
Rated current
Rated current
MPPT range
SoC range
(a) 0 Irated (b) Icharge 0 Idischarge
Figure 16.22 Droop characteristics for dc microgrid: (a) nondispatchable units

and (b) storage converters
units are operating under the MPPT power and share the load based on their droop
gains [33]. Furthermore, the maximum supplied power by the PV units are limited
by the MPPT power as shown in Figure 16.22(a). By decreasing the microgrid
demand, the dc voltage will be increased, and hence, the PV controller enters into
the droop-controlled mode.
The battery units also should be controlled in terms of power sharing and
energy level, i.e., SoC level. An autonomous droop characteristic for distributed
batteries in the microgrid can be shown as Figure 16.22(b), where the maximum
charging and discharging power can be set taking into account the SoC level of
batteries. Hence, by adapting the droop characteristics in the local controller, the
energy level of each battery can be controlled. Furthermore, the power sharing,
both charging and discharging powers, among the batteries can be controlled by
properly setting the droop gains shown in Figure 16.22(b).
Since the scope of this chapter focuses on the PV units in microgrids, possible
control structures of PV converters are shown in Figure 16.23. For simplicity,
a dc–dc boost converter is considered as shown in Figure 16.23(a). For the control
structures shown in Figure 16.23(b) and (c), a supervisory control changes the PV
operation mode following the system generation-load balance constraints. How-
ever, in the control structures shown in Figure 16.23(d) and (e), the operation mode
of the PV units will be determined by the local parameters, i.e., the MPPT power
and the droop-induced power. These approaches are suitable for autonomous droop
techniques.
16.7 Simulation of a droop-controlled PV microgrid

In this section, a simulation is performed in order to demonstrate the droop-based
autonomous control of dc sources in a dc microgrid with two PV units and two
PV converter
Ioutput
Iin
Voutput IL Vin
MPPT
(a)
VMPPT IMPPT
VMPPT –
PI PI
– IMPPT
Vin IL
PI
V* V* PI –
PI PI –
– – IL
IL Supervisory Voutput
Voutput Ioutput Supervisory
control Droop Rd
Rd Ioutput Droop control
(c)
(b)
Vinput
VMPPT –
IMPPT PI
MIN
PI
V* PI PI V* PI –
– – –
0 IL
Voutput IL Voutput
Droop Rd Ioutput Rd Ioutput
Droop
(d) (e)
Figure 16.23 Adjustable droop control for MPPT-based units (e.g., PV):
(a) dc–dc boost converter for PV, (b), (c) mode transition strategy,
and (d), (e) seamless strategy
batteries as shown in Figure 16.24. The parameters of the simulated dc grid are
given in Table 16.1. The droop characteristics of the units are set as shown in
Figure 16.23(a) and (b), where the droop gains of batteries are 2 and 1 W.
Furthermore, the MPPT currents of PV units are set to 2.25 and 2.75 A, and the
droop gains of PV units are 10 W. The batteries are considered to be fully charged
in the beginning. Simulation results are shown in Figure 16.25. The microgrid load
is 3.2 kW and the PV units are operating in MPPT mode with (2.25 þ 2.75)
400 ¼ 2 kW. The remaining load power is supplied by batteries, where the output
current of the first battery is two times that of the other one following their droop
gains as shown in Figure 16.25(b). Therefore, the batteries are operating in the
droop mode and forming the dc-link voltage, and the maximum power of the PV
units is injected into the microgrid. Once the load is decreased to 1.6 kW, the dc-
link voltage is increased as shown in Figure 16.25. Following Figure 16.22(b), by
increasing the dc voltage, the PV units should operate in the droop-controlled mode
and decrease the output current to balance the demand of the system. The load
sharing among the dc sources is shown in Figure 16.25(b). The PV units are
working in the droop-controlled mode and supplying the equal current following
their droop gains. Furthermore, the battery currents are zero, since they are con-
sidered as fully charged. In this mode, the dc voltage is formed by the PV units.
Vdc PV converter 1 Ii,PV1
Ldc
Vi,PV1
Vo,PV1
PV 1
Cdc Cdc
MPPT
–
PI PI IMPPT
–
MIN
V*
1/R1
PV converter 2 Ii,PV2
Ldc
Vi,PV2
Vo,PV2
PV 2
Cdc Cdc
MPPT
–
PI PI IMPPT
–
MIN
V*
1/R2
Battery converter 1
Io,BT1 Ldc
Battery 1
Vo,BT1
Ii,BT1 Cdc
Cdc
Ii,BT1 Battery Rd1

Rd1 Ich
– mana-
PI PI Ich, Idch
– gement
Load
Idch
VPCC V*
Battery converter 2
Io,BT2 Ldc
Battery 2
Vo,BT 2
Ii,BT2 Cdc
Cdc
Ii,BT2 Battery Rd2

Rd2 Ich
– mana-
PI PI Ich, Idch
– Idch gement
V*
Figure 16.24 A typical dc microgrid with two dispatchable units, i.e., battery
and two nondispatchable PV units
Table 16.1 Parameters of the power system controllers
Parameter Value
PV converter droop gains (W) 10, 10
Battery converter droop gains (W) 2, 1
Load power (kW) 1.6, 1.6
Operating MPPT current of PVs (A) 2.25, 2.75
Rated power of PV converter 2 kW
Rated power of battery converter 2 kW
dc Inductor of converters 2 mH
dc Capacitor of converters 500 mF
410
dc link voltage (V)
405
Battery controlled PV controlled
400
395 3.2 kw 1.6 kW
390
(a) 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2
Time (s)
4
MPPT mode Constant power mode
3
Output current (A)
1
PV 1 PV 2
0
Batt. 1 Batt. 2
–1
(b) 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2
Time (s)
Figure 16.25 Load sharing among dispatchable and nondispatchable units:

(a) dc link voltage and (b) output currents
16.8 Summary
dc Microgrids introduce an infrastructure for integrating dc-based energy resources,
storages and loads with high reliability, availability, and efficiency. Meanwhile, PV
systems are one of the most popular renewable resources due to its modularity as
well as lower costs. This chapter discussed different applications and structures of
dc microgrids, where PV units can be utilized as an energy resource. These appli-

cations can be divided into dc distribution systems, marine applications, data
centers, oil-drilling platforms, off-grid applications, dc homes, etc. Different PV
integration possibilities into dc microgrids were presented as well. Following
the application of dc microgrids, hybrid PV-battery units or distributed PV and
battery units can be employed. For example, the first structure is suitable for
clusters of dc homes, where each home has its own PV and battery units. For an
industrial application, the second structure, i.e., distributed PV and battery units,
is more viable.
Furthermore, in this chapter, the electrical model of PV systems was explained
and different dc–dc converter topologies for PV applications were presented.
Moreover, various control systems of PV plants in dc microgrid were introduced
as well as power sharing concepts among paralleled PV units and battery storages
was provided by simulations on a typical dc microgrid employing the autonomous
droop control approach.
Optimal design and control of PV power plants in dc microgrids taking into
account the reliability and availability constraints are required. Since the micro-
grids may operate in the islanded mode, the critical loads need to be supplied with
an acceptable level of availability. Therefore, this concept should be taken
into consideration in future research of PV-based dc microgrids.
References
[1] Blaabjerg, F., Yang, Y., Yang, D., Wang, X.: ‘‘Distributed power-generation
systems and protection’’ Proc. IEEE, 2017, 105, (7), pp. 1311–1331.
[2] International Renewable Energy Agency: ‘‘Renewable Energy Capacity
Statistics 2017,’’ 2017.
[3] Youichi, I., Yang, Z., Hirofmi, A.: ‘‘DC Microgrid Based Distribution
Power Generation System’’, in ‘‘Power Electronics and Motion Control
Conference, 2004’’ (IPEMC, 2004), the 4th International, 2004, 3, pp.
1740–1745. IEEE.
[4] Kumar, D., Zare, F., Ghosh, A.: ‘‘DC microgrid technology: system archi-
tectures, AC grid interfaces, grounding schemes, power quality, commu-
nication networks, applications, and standardizations aspects’’ IEEE Access,
2017, 5, pp. 12230–12256.
[5] Diaz, N.L., Dragicevic, T., Vasquez, J.C., Guerrero, J.M.: ‘‘Intelligent dis-
tributed generation and storage units for DC microgrids: a new concept on
cooperative control without communications beyond droop control’’ IEEE
Trans. Smart Grid, 2014, 5, (5), pp. 2476–2485.
[6] Elsayed, A.T., Mohamed, A.A., Mohammed, O.A.: ‘‘DC microgrids and
distribution systems: an overview’’ Electr. Power Syst. Res., 2015, 119,
pp. 407–417.
[7] Dragicevic, T., Lu, X., Vasquez, J.C., Guerrero, J.M.: ‘‘DC microgrids—
part II: a review of power architectures, applications, and standardization
issues’’ IEEE Trans. Power Electron., 2016, 31, (5), pp. 3528–3549.
[8] Kakigano, H., Miura, Y., Ise, T.: ‘‘Low-voltage bipolar-type DC microgrid
for super high quality distribution’’ IEEE Trans. Power Electron., 2010,
25, (12), pp. 3066–3075.
[9] McNelis, A.M., Beach, R., Soeder, J.F., et al.: ‘‘Simulation and Control Lab
Development for Power and Energy Management for NASA Manned Deep
Space Missions,’’ in ‘‘12th International Energy Conversion Engineering
Conference’’ (American Institute of Aeronautics and Astronautics, 2014).
[10] Dever, T., Trase, L., Soeder, J.F., et al.: ‘‘Application of Autonomous
Spacecraft Power Control Technology to Terrestrial Microgrids,’’ in ‘‘12th
International Energy Conversion Engineering Conference’’ (American
Institute of Aeronautics and Astronautics, 2014).
[11] Khooban, M.-H., Dragicevic, T., Blaabjerg, F., Delimar, M.: ‘‘Shipboard
microgrids: a novel approach to load frequency control’’ IEEE Trans.
Sustain. Energy, 2017, 9, (2), pp. 843–852.
[12] Hebner, R.E., Uriarte, F.M., Kwaskinski, A., Gattozzi, A.L., Estes, H.B.,
Anwar, A.: ‘‘Technical cross-fertilization between terrestrial microgrids and
ship power systems’’ J. Mod. Power Syst. Clean Energy, 2016, 4, (2),
pp. 161–179.
[13] Im, W.-S., Wang, C., Tan, L., Liu, W., Liu, L.: ‘‘Cooperative controls for
pulsed power load accommodation in a shipboard power system’’ IEEE
Trans. Power Syst., 2016, 31, (6), pp. 5181–5189.
[14] Hansen, J.F., Lindtjørn, J.O., Vanska, K.: ‘‘Onboard DC Grid for enhanced
DP operation in ships’’ MTS Dyn. Position. Conf., Houston, 2011.
[15] Jiang, T., Yu, L., Cao, Y.: ‘‘Green Energy and Technology Energy Man-
agement of Internet Data Centers in Smart Grid.’’ Berlin Heidelberg:
Springer, 2016.
[16] Yu, L., Jiang, T., Zou, Y.: ‘‘Real-time energy management for cloud data
centers in smart microgrids’’ IEEE Access, 2016, 4, pp. 941–950.
[17] Rodriguez-Diaz, E., Vasquez, J.C., Guerrero, J.M.: ‘‘Intelligent DC homes
in future sustainable energy systems: when efficiency and intelligence work
together’’ IEEE Consum. Electron. Mag., 2016, 5, (1), pp. 74–80.
[18] Reed, G.F., Grainger, B.M., Sparacino, A.R., Zhi-Hong M.: ‘‘Ship to grid:
medium-voltage DC concepts in theory and practice’’ IEEE Power Energy
Mag., 2012, 10, (6), pp. 70–79.
[19] Kounev, V., Tipper, D., Grainger, B.M., Reed, G.: ‘‘Analysis of an Offshore
Medium Voltage DC Microgrid Environment; Part II: Communication
Network Architecture,’’ in ‘‘2014 IEEE PES T&D Conference and Exposi-
tion’’ (IEEE, 2014), pp. 1–5.
[20] Patel, H., Agarwal, V.: ‘‘MATLAB-based modeling to study the effects of
partial shading on PV array characteristics’’ IEEE Trans. Energy Convers.,
2008, 23, (1), pp. 302–310.
[21] Mette, A., Pysch, D., Emanuel, G., Erath, D., Preu, R., Glunz, S.W.: ‘‘Series
resistance characterization of industrial silicon solar cells with screen-
printed contacts using hotmelt paste’’ Prog. Photovoltaics Res. Appl., 2007,
15, (6), pp. 493–505.
[22] Pysch, D., Mette, A., Glunz, S.W.: ‘‘A review and comparison of different
methods to determine the series resistance of solar cells’’ Sol. Energy Mater.
Sol. Cells, 2007, 91, (18), pp. 1698–1706.
[23] Salam, Z., Ishaque, K., Taheri, H.: ‘‘An Improved Two-Diode Photovoltaic
(PV) Model for PV System,’’ in ‘‘2010 Joint International Conference on
Power Electronics, Drives and Energy Systems & 2010 Power India’’ (IEEE,
2010), pp. 1–5.
[24] Liu, K.-H., Lee, F.C.Y.: ‘‘Zero-voltage switching technique in DC/DC
converters’’ IEEE Trans. Power Electron., 1990, 5, (3), pp. 293–304.
[25] Lee, K.-H., Chung, E., Han, Y., Ha, J.-I.: ‘‘A family of high-frequency
single-switch DC–DC converters with low switch voltage stress based
on impedance networks’’ IEEE Trans. Power Electron., 2017, 32, (4),
pp. 2913–2924.
[26] Rehman, M.M.U., Zhang, F., Zane, R., Maksimovic, D.: ‘‘Control of Bidir-
ectional DC/DC Converters in Reconfigurable, Modular Battery Systems,’’
in ‘‘2017 IEEE Applied Power Electronics Conference and Exposition
(APEC)’’ (IEEE, 2017), pp. 1277–1283.
[27] Yang, Y., Blaabjerg, F.: ‘‘Overview of single-phase grid-connected
photovoltaic systems’’ Electr. Power Components Syst., 2015, 43, (12),
pp. 1352–1363.
[28] S. Moskowitz: ‘‘The Next Opportunity for Utility PV Cost Reductions:
1,500 Volts DC, Greentech Media,’’ 2018. https://www.greentechmedia.com/
articles/read/the-next-opportunity-for-utility-pv-cost-reductions-1500-volts-
dc, accessed February 2018.
[29] Serban, E., Ordonez, M., Pondiche, C.: ‘‘DC-bus voltage range extension in
1500 V photovoltaic inverters’’ IEEE J. Emerg. Sel. Top. Power Electron.,
2015, 3, (4), pp. 901–917.
[30] Inzunza, R., Okuyama, R., Tanaka, T., Kinoshita, M.: ‘‘Development of a
1500 Vdc Photovoltaic Inverter for Utility-Scale PV Power Plants,’’ in
‘‘2015 IEEE 2nd International Future Energy Electronics Conference
(IFEEC)’’ (IEEE, 2015), 2015, pp. 1–4.
[31] C. Crowell: ‘‘1,500-Volt Systems to Trend in 2017: Here’s What You Need
to Know – Yaskawa – Solectria Solar,’’ 2018. https://solectria.com/com-
pany/in-the-news/1-500-volt-systems-to-trend-in-2017-heres-what-you-need-
to-know/, accessed February 2018.
[32] Boroyevich, D., Cvetkovic, I., Burgos, R., Dong, D.: ‘‘Intergrid: a future
electronic energy network?’’ IEEE J. Emerg. Sel. Top. Power Electron.,
2013, 1, (3), pp. 127–138.
[33] Dragicevic, T., Guerrero, J.M., Vasquez, J.C., Skrlec, D.: ‘‘Supervisory
control of an adaptive-droop regulated DC microgrid with battery manage-
ment capability’’ IEEE Trans. Power Electron., 2014, 29, (2), pp. 695–706.
[34] Gu, Y., Xiang, X., Li, W., He, X.: ‘‘Mode-adaptive decentralized control for
renewable DC microgrid with enhanced reliability and flexibility’’ IEEE
Trans. Power Electron., 2014, 29, (9), pp. 5072–5080.
[35] Peyghami, S., Mokhtari, H., Blaabjerg, F.: ‘‘Decentralized load sharing in an
LVDC microgrid with an adaptive droop approach based on a superimposed
frequency’’ IEEE J. Emerg. Sel. Top. Power Electron., 2017, 5, (3),
pp. 1205–1215.
Chapter 17
Demonstration sites of dc microgrids
Aditya Shekhar1, Laura Ramı́rez-Elizondo1,
Seyedmahdi Izadkhast1 and Pavol Bauer1
17.1 Introduction
In recent years, several demonstration projects using dc microgrids have been
implemented across the world due to some distinct advantages of dc over ac sys-
tems. The purpose of such initiatives is to validate the theoretically predicted
benefits of dc distribution in practical scenarios. This chapter presents a non-
comprehensive overview of existing demonstrations and pilot projects for a wide
range of applications such as off-grid microgrids, transportation electrification,
datacenters, residential and industrial purposes. For each application, key aspects
such as architecture, components, control, protection and socioeconomic impacts
are highlighted. A short discussion is offered on trends in voltage levels, capacity
and topology, progressing toward possible standardization approaches based on
recognized best practice.
17.2 Off-grid dc microgrids

Off-grid applications can refer to remote locations with limited access to the con-
ventional ac grid. The grid may be required to be operated in stand-alone mode for
long hours during the day, e.g., from 12 to 24 h. For instance, in many states of
India, 70%–90% households fall under this category [1]. In such applications, lack
of market inertia of the ac grid may aid to the proliferation of dc nanogrids. The
energy needs of such areas are scarcely met due to economic constraints that make
load shedding a preferred option [1]. Consequently, off-grid solar dc homes similar
to one shown in Figure 17.1 are the emerging solution for rural electrification [2].
17.2.1 Architecture
A typical architecture for off-grid dc solar home systems (SHS) is shown in
Figure 17.2. The structure is radial with a 48-V common dc bus interconnecting the
1
Department of Electrical Sustainable Energy, DC Systems, Energy Conversion and Storage (DCE&S)
Group, Delft University of Technology, The Netherlands.
125 kW PV
5 W LED
bulb
Rooftop PV
Central
controller
with 48 V
dc bus
Figure 17.1 Illustration of dc powered offgrid Tribal homes at Kundithal village,

The Nilgiris, Tamil Nadu, India (courtesy: [2])
Solar panel
N: Normal line
E: Emergency line
48 V dc N
IRU E Home 1
48 V dc N
230 V ac IRU E Home 2
48 V dc N
IRU E Home 3
48 V dc N
IRU E Home 4
Grid
Bluetooth
low-energy
interface
Battery
Cloud
Figure 17.2 Inverterless-500 dc powered solar home system architecture [1]
distributed sources, storage and loads. To guarantee power supply to critical loads,
connections are bifurcated into normal line (N) and the emergency line (E). Each
home has an inverter-less remote unit (IRU). The option of remote monitoring
using, for example, a Bluetooth interface is also available.
Demonstration sites of dc microgrids 417
17.2.2 Components
The minimum needs for a low-income household may at least consist of loads such
as a few fans, tube-lights, bulbs and mobile phone chargers. With available
opportunities, some households showed preference for entertainment loads such as
television and commercial loads such as tailoring [2–4], amounting to a demand
of 100–500 W. With use of dc, 2–2.5 times reduction in battery and solar module
size at half the cost and efficiency enhancement at upward of 15% have been
reported [1,2]. The use of dc-ready appliances is encouraged, claiming efficiency
gains of almost 50% with such practice [2], such as 1 18 W LED tube-light,
1 32 W brush-less dc fan, 1 5 W LED bulb and a mobile phone charger. The
125-Wp solar photovoltaic (PV) generation has a trade-off between cost and
availability. Storage capacity caters for load demands when neither PV nor grid is
available, designed to supply power for a downtime of 1–2 days [2]. Some ac loads
can be directly supplied from the grid or through the dc bus, even though, the use of
an additional inverter is discouraged.
17.2.3 Control and operational safety

Considering the small scale of application, centralized control is employed in the
main control board for one or several households. Heuristic algorithms imple-
mented by the IRU minimize grid usage and maximize PV utilization. They have to
also ensure adequate electricity supply to critical loads during significant down-
times. The state of charge (SoC) of the batteries is monitored and if no power is
generated, the storage supplies power to all loads until a certain threshold. If the
threshold is reached, then the non-critical loads are shed. dc Voltage of 48 V is
usually chosen for low capacity nanogrids, as occurrence of sustained series arcing
is limited [5].
17.2.4 Socioeconomic impact

Case-studies of solar house systems for rural electrification across the world shared
positive experiences from the end users, showing enhancement in the standard of
living with fulfillment of basic energy needs [1–4]. Of the six cases of SHS studied
in [4], only four had payback below 3 years and an internal rate of return above
39%. Demonstrating financial viability with larger capacity and/or encouragement
of rural businesses, it was highlighted that sustainability of dc based SHS hinges on
greater support from various government and nongovernment organizations, local
sources of funds and a premium on social benefits of rural electrification.
17.2.5 Discussion
A summary of various dc offgrid applications worldwide with respect to voltage
levels, capacity, topology, and control is listed in Table 17.1. In most cases, the
demonstrations were dominated by low capacity (below 500 W) and low voltage
levels (48 V or below, in line with future IEEE P2030.10 Standard for rural and
remote dc), signifying that this technology is developed to meet the basic energy
Table 17.1 Demonstration sites for offgrid or near offgrid dc microgrids
Demo-site description Voltage Capacity Topology Control

level
1 SHSa, Bangladesh [3,4] 12 V 100 W
2 SHSa, Africa [6] 20–100 W
3 SHSa, Kundithal, 48 V 20 homes, Central
Tamil Nadu, India [2] 100–500 W each
4 SHSa, Irrukam Island [2] 48 V 26 homes All radial Central
Andhra Pradesh, India 100–500 W each
5 SHSa, Bhomji ka Gaon [1] 48 V 125 Wp, 1 kW h Central
6 School, Tirmal, 48 V 3 kW, 50 kW h
Orissa, India [7]
7 PECb, Portugal [8] 24 V 1.8 kWp Central
a
SHS: solar house systems.
b
PEC: polygeneration energy container.
needs of socioeconomically backward areas. The topology choice is usually radial

with PV panels, small storage and dc-ready loads attached to a common dc bus,
generally with no interconnection between individual nanogrids. Thus, end-user
need based system design with entity level centralized control scheme was preferred.
17.3 Transportation electrification
Electrification of the transportation sector is driven by the desire to shift toward

clean energy sources and reduced emissions. Potential size and weight reduction
play a pivotal role in dc-based transportation electrification. To give a numerical
example on the size, The University of Trieste in Italy conceived a 2 MVA,
22,500 rpm, 12 phase permanent magnet prototype generator [9], connected to 3 kV
dc shipboard power system with a controlled active ac–dc converter. The weight of
generator was reduced to less than 50% of its predecessor. The adoption of dc
technologies in transportation can be rapid, since these systems are stand-alone.
This section provides an overview of shipboard, traction and electric vehicle (EV)
charging dc power systems.
17.3.1 Shipboard dc microgrids

Electric propulsion has been considered in naval destroyers (e.g., UK’s Type 45
destroyer) as it may provide opportunities to improve the power density using
advancements such as emerging permanent magnet and super-conducting technol-
ogies [10]. The interest in all electric ships (AES), fueled by the need for efficient
space utilization, has given rise to the concept of a dc-based integrated power system
(IPS) [11]. Typically, the shipboard power systems have a high capacity in the order
of a few megawatts. The voltage levels are between 1 and 35 kV. The recommended
practices for reliable integration of shipboard power components in a medium
voltage dc (MVDC) system are discussed by the IEEE Standards Association [12].
Figure 17.3 An illustration of a dc-based all electric ship (adapted from [14])
17.3.1.1 Architecture
Keeping in mind the redundancy requirements, two buses, the port and starboard
bus run through the length of the ship. An illustration of such a dc microgrid island
on ship is shown in Figure 17.3. Different shipboard dc electrical distribution
system architectures with trade-offs between reliability, complexity and efficiency
are discussed in [11]. An interesting compact design concept of ‘‘Power Corridor’’
toward the IPS keeping in mind the space reservations for AES is proposed in [13].
17.3.1.2 Components
As part of The Electric Ship Research and Development consortium, the University
of Texas at Austin center for electromechanics (UT-CEM) assembled a 2-kV
MVDC microgrid testbed as shown in Figure 17.4 [15].
The isolated naval microgrid is envisioned to operate at 80–100 MW with
different steady state and pulsed power loads. The following components are
highlighted: (a) gas turbine driven 3 MW high speed generator, (b) full bridge
passive diode rectifier, (c) bidirectional, active, IGBT controlled converter,
(d) 1.25 MW Toshiba model variable frequency drive capable of operating up to
300 Hz modified from its commercial version to operate as a ‘‘dc-ready’’ device
able to directly interface the load with the common dc bus, (e) Kahn model
hydraulic dynamometer capable of absorbing 4.77 MW at 18,000 rpm, that can be
controlled to mimic transient conditions in dc distribution systems.
17.3.1.3 Power quality, control and protection

System level studies on transients associated with microgrid islanding and recon-
nection were planned in [15], focusing on ancillary services with different control
strategies for better power quality. In [9], the bus connected choppers integrating
the prototype high speed permanent magnet generator are responsible for the output
480 V
(c)
utility (b)
(a) line (d)
ALPS load bank

Field controllable 1.3 MW @ 2 kVdc
engine-driven Transformer ALPS
ac generator 480 V/1,400 V converter/ ALPS
active rectifier converter/chopper
ALPS
inductors
Field controllable 300 A 1,500 V
engine-driven ALPS ALPS
ac generator transformer passive rectifier
Toshiba Dyno
480 V 1.25 MW VFD
ALPS (e)
utility Kahn
motor/generator
line dynamometer
Field controllable Oldenburg

engine-driven Oldenburg Oldenburg chopper and
ac generator transformer controlled rectifier load
Low voltage equipment

680 Vdc maximum
ALPS auxiliary chopper
and load
Figure 17.4 Testbed at UT-CEM for MVDC shipboard microgrid (courtesy: [15])
voltage control. The machine reacted to the power electronics and resulted in dis-
tortion in phase current, leading to flux pulsations and eddy current losses, an
important challenge to consider during design.
For stable ship operation, it is important to maintain power availability during
contingencies. It is identified that due to interconnectivity, the reconfiguration
capability maybe a potential advantage of MVDC IPS in AES [11]. While
acknowledging significant efficiency gains and lesser problems with harmonic
distortion, The ‘‘ABB’’ group of industries noted that the challenge in the design of
the marine on-board dc system was related to protection and selective localization
of faults. Their approach was a coordinated use of fuses and isolation switches
along with protective power electronic devices. Further, the controlled thyristor
rectifier connected to the generator played an added role of protection device. It was
claimed that dc fault currents can be controlled within 10–20 ms, providing a sig-
nificant reduction in fault energy as compared to the conventional ac system [16].
17.3.1.4 Economical operation

The interest in onboard dc led to the commercial design of Dina Star by ABB. This
eliminated the ac switchboards and need for bulky transformers, as well as reduced
the power conversion stages. It was claimed that the use of 1 kV dc with up to
20 MW demand improves the vessel’s energy efficiency by 20% and the weight
of electrical system by 30% [16]. Based on the operational experience with Dina
Star, ABB claimed that 27% efficiency gains could be achieved for 10% loading.
It was concluded that between 9% and 14% efficiency gain is achievable with
variable load.
17.3.2 Railway traction with dc supply

High starting torque, ability to share torque, good natural response to load varia-
tions and rapid speed control have historically made series dc motors an attractive
choice for short distance traction applications [17]. For example, the share of
1.5/3 kV dc was about 37% in France in a survey conducted as recently as 2003
[18], with about 66% of railway lines running at 1.5 kV dc in Japan [19], almost
10,000 km including 1,650 dc substations with 2,700 rectifiers span the length of
the country as of 2013 [20]. Similarly, most of the 20,000-km Poland’s railway
tracks are electrified by 3 kV dc traction supply [21]. The split between preference
for short distance dc traction and long distance ac traction should be kept in mind.
This section describes several dc traction projects and discusses the typical char-
acteristics of such systems.
A typical dc traction system with rail return is shown in Figure 17.5. Considering
that catenaries of more than one conductors are usually not preferred [18], mono-
polar dc maybe advantageous as it simplifies the overhead catenary system for
collecting traction current and reduces costs.
Substation A Substation B Substation C

ac bus bar
(10 kV)
Rectifier
transformer
Rectifier
dc busbar
(750 V/1,500 V)
dc feeder
cable
Upward catenary
Downward catenary
M Locomotive
M
Upward rail
Downward rail
Figure 17.5 dc Traction system architecture (courtesy: [22]

With standard voltage levels of 750 V for urban metros and trams and 1.5–
3 kV for interurban and regional system [17], spacing between 1 and 10 MW
substations is selected between 3 and 20 km depending on the operating voltages
(higher the voltage, greater the allowable spacing). The system is sectionalized
using tie breakers to ensure isolation of faults to a limited zone of influence [23]. In
general, the zone of influence is about 8 km for a 1.5-kV dc substation with 4 kA
current [24].
17.3.2.2 Components
The key components of each dc substation are rectifiers, filters, transformers and
dc switchgear. With 4 12-car trains nearby (2 leaving and 2 approaching), each
drawing 4 kA, the demand could be as high as 25 MVA [25]. However, the nature
of the load is such that it is not expected to burden the substation for a long time
continuously. In order to minimize the related system costs, such substations
maybe designed to operate with overload capacity for a short while, typically
150%, 300% and 450% for 1 h, 1 min and 10 s, respectively, for a 3 MW, 750 V
supply [24]. The use of energy storage and tie-lines is employed in dc traction on
the Yamatoji line in Japan [19]. Apart from providing ancillary services such as
regenerative energy absorption, supplying power deficit and compensating
catenary voltage drop, at least 3% decrease in energy from dc substation feeders
was reported in the project.
17.3.2.3 Protection
Protection of 3 kV dc traction systems from lightening surges, and particularly, the
vulnerability of overhead catenary conductor to lightning strike is discussed in
[21,26]. For rail-return traction supply, issues associated with rail to earth potential
have been cautioned in [27]. Generally, a combination of ac and dc switchgear is
needed to isolate faults in traction supply systems [28]. High-speed dc circuit
breakers are employed to clear the faults within 15–30 ms. An intertripping
mechanism is incorporated while ensuring a coordinated discriminatory operation
between ac and dc side circuit breakers [29]. Further, a reverse current tripping is
incorporated in the dc switchgear to isolate faulty rectifiers from the common dc
bus. In [23], it is highlighted how dynamic nature of traction loads could mimic the
sharp rise in fault currents when train is accelerating. Therefore, the time setting of
the di/dt tripping relay should be carefully chosen based on the traction system
characteristics.
17.3.3 Discussion
Some examples of transportation electrification using dc microgrids are summar-
ized in Table 17.2. It is observed that the trend is toward high installed capacity of
operation and correspondingly medium voltage levels are chosen. The dc shipboard
IPS is controlled in a decentralized way to crucially increase the reliability, while
using radial or distributed topologies. Traction systems typically use hierarchical
control scheme in a sectioned, monopolar with rail return topology.
Table 17.2 Demonstration sites for transportation electrification using dc
Description Voltage Capacity Topology Control Protection

level
Shipboard dc microgrids
1 Dina Star (ABB) 1 kV 20 MW Distributed, SSCB with
multidrive fusea
2 Austin, USA [15] 2 kV 2–10 MW rdl.b dc bus Dctr.c Series arc
3 Italy [9] 3 kV 2 MVA rdl.b dc bus SSCBa
Traction systems [17–30]
4 BE, ES, IT 3 kV 1.5–10 MW, Sectioned, ac, dc
20 km rail return switchgear
5 FR, NL 1.5 kV 1.5–10 MW, Sectioned, ac, dc
10 km rail return switchgear
6 United Kingdom 750 V 3 MW, 3 km Sectioned, ac, dc
rail return switchgear
7 Atlanta, USA 750 V 3 MW Sectioned, ac, dc s
rail return witchgear
8 Singapore 1.5 kV 3 MW, 5 km Sectioned, ac, dc
rail return switchgear
9 CN 750 V, Few MW Sectioned, Hrr.d ac, dc
1.5 kV 2.4 km rail return switchgear
11 Yamanote, JP 1.5 kV Few MW
Electric vehicles [31–33]
12 Hawaii, USA 240 V 1–5 MW Meshed Dctr.c
13 Italy 790 V 20 kV A Bipolar Hrr.d
a
Solid state circuit breaker.
b
Radial.
c
Decentralized.
d
Hierarchical.
Innovator Elon Musk envisioned a synergistic, large-scale evolution of rooftop

PVs, storage and EVs [34]. Signs of potential in these technologies can be observed
from the launch of Tesla power wall 2.0 and the acquisition of SolarCity by
Tesla [35]. The role of dc in interconnecting PV generation, EV charging and
storage can be significant [36].
17.4 Datacenters
Currently, many datacenters around the globe are tending to employ dc technology
due to its high reliability and efficiency. To highlight the importance of efficiency,
in 2006 1.5% of total US electricity consumption was from datacenters [37]. This
section provides an overview of dc datacenters at Lawrence Berkeley National
Laboratory (LBNL) and then reviews various dc datacenters around the world.
380 V
480 V Transformer UPS Server 480 V UPS DC Bus Server
ac grid ac grid
ac dc ac dc dc ac dc dc
dc ac dc dc dc dc dc dc
+ dc + dc
–
dc 12 V dc
(a) (b)
Figure 17.6 Datacenter distribution system: (a) ac based and (b) dc based [38]
17.4.1 Architecture
The schematics of ac and dc datacenters studied by LBNL are shown in
Figure 17.6. The dc datacenter servers are connected to a 380 V dc bus, where from
the main ac grid is routed via uninterruptible power supply (UPS). It is indicated
that one dc/ac conversion step in UPS and one ac/dc conversion step in server
power supply is reduced for a dc based datacenter.
17.4.2 Components
The need for standardization of various dc-ready commercially available compo-
nents such as servers, connectors, power cords and in-rack distribution strips was
identified in [38]. The project compared the performance of ac and dc datacenters
with the same level of achievable functionality and claimed that the results
demonstrated an efficiency improvement by 7%–8% with dc. Out of this, it was
reported that the share of efficiency gain was 4% for UPS, 2% for transformer and
2% for server power supply. It was anticipated that with component optimization,
further improvements are possible.
17.4.3 Protection and grounding system

The LBNL datacenter grounding system was designed based on the IEEE Standard
1100-2005, describing the recommended practices for powering and grounding
electronic equipment [39]. According to ANSI T1.311, the highest dc voltage level
is 160 V, while LBNL uses 380 V dc. Highlighting that suitable dc breakers are
required for proper coordinated protection, it is deemed necessary to further
investigate the over-current protection devices and disconnect devices for the dc
system.
17.4.4 Discussion
Table 17.3 lists the information of some dc-based datacenters from around the
world. The system level operating voltage of datacenters is standardizing at about
380 Vdc. Even the technical report of INTEL labs describing a 400 V datacenter
affirms its support to the standardization around 380 V [40]. Regarding the size,
there is a clear trend toward high power dc datacenters, as Intel Labs have reached
Table 17.3 Demonstration sites for data centers using dc distribution
Description Voltage Capacity Topology Control Protection

level
1 LBNL, USA [38] 380 V 15–40 kW
2 Intel Labs 400 V 5.5 MW Bipolar dc-Rated IEC
USA [40,41] 309 Connectors
3 Nedo, Japan [42] 380 V 500 kW Adaptive
4 San Diego, USA [37] 380 V 2.8 MW Bipolar
5 Switzerland [37] 380 V 1 MW Bipolar
6 Rokkasho, Japan [43] 380 V 13.6 kW Bipolar
up to 5.5 MW. The key challenges highlighted by the LBNL study include (a)
economics of scale may shift in favor of dc if knowledge on its benefits are prop-
erly disseminated. (b) Industry leadership in moving toward development of stan-
dardized practices is desirable for widespread market adoption on dc-based
datacenters. Demonstration sites for datacenters using dc distribution
17.5 Residential and commercial dc buildings
The importance of dc microgrids in realizing a flexible, smart and efficient energy

network, termed ‘‘Enernet’’ is envisioned in [44]. As shown in Figure 17.7, the
need for large number of ac/dc conversions can be eliminated resulting in lower
costs and higher efficiency in dc buildings.
17.5.1 dc House
Many household appliances such as televisions, refrigerators, computers, LED
lighting, EVs as well as sources like solar PV and storage elements can be inter-
connected to form a dc nanogrid. A typical grid connected dc residential house is
shown in Figure 17.8.
17.5.1.2 Components
An example of a demo house from EPARC, Taiwan [45], is shown in Figure 17.9.
The focus of this demonstration was to adapt commonly used ac residential appli-
ances indicated in the figure to operate with dc input. Various components were
connected radially to a common dc bus of 380 20 V with about 5–10 kW
capacity.
The products were converted from ac source with power factor correction by
rerouting the protection and circuitry. It was highlighted that several appliance
manufacturers such SAMPO, TECO, Eulife, Jamicon and Fego supported this
modification of their products. It is encouraged to mark such products as ‘‘dc-
ready’’ to reinforce end-user confidence in adopting dc technologies [46]. Opera-
tional data collected showed a efficiency enhancement of 8.5% compared to ac.
Traditional Proposed
PV PV
ac dc dc "DC Ready"
Loads loads
dc dc dc
ac dc dc
dc dc dc
ac dc dc
dc ac Heat pump ac Heat pump
ac dc dc
Fuel cell Fuel cell
dc dc dc
ac dc – + dc – +
dc dc dc
Drive Drive
Figure 17.7 dc Buildings compared to traditional ones
Battery PV
ac grid
= =
MPPT
= =
Bidirectional
dc grid
System ~ voltage
monitoring regulator
=
and control
380 ± 20 V
+
–
Refrigerator TV Computer dc lamp dc fan EV
Figure 17.8 Grid connected residential dc microgrid (based on [45])
17.5.1.3 Control
Heuristic controls try to minimize the energy exchange with the main grid using
storage elements while maximizing the use of distributed green energy resources in
the system [45]. The grid connected bidirectional inverter performs the energy
balance and voltage regulation of the dc bus. The role of this inverter can be
diversified, for example, to draw or store energy in the flywheel or operating it as a
high power rating charger/discharger for dc UPS [45].
Figure 17.9 Grid connected residential microgrid (courtesy: [45])
17.5.1.4 Protection
It was reported that a special dc plug was developed by Fujitsu to eliminate arcing.
Issues with series arcing are more relevant in residential dc nanogrids due to a
relatively high operating voltage where sustained arcs can be expected [5]. Con-
sidering the wear and tear issues as well as added costs of such components, a novel
selective series arc extinction scheme is developed in [47] with experimental proof of
concept. Challenges such as localized short circuit protection and standardization of
safety codes pertaining to different system components are important considerations.
17.5.2 Building interiors

The layout of the dc office building at the Fraunhofer Institute in Erlangen, Germany
is shown in Figure 17.10 [48]. This system has two bipolar dc buses. The main bus is
controlled to 38020 V and interconnects various office loads, the ac grid and the
PV electrical source in a radial structure at 15 kW capacity. An alternate PV to grid
bus maintains a higher dc voltage at 800 V. The chosen voltage levels are consistent
with the ETSI EN 300 132-3-1 standard [49] for dc voltage range 260–400 V,
existing ac infrastructure and semiconductor technology at adequate efficiency.
17.5.2.2 Components
The front-end ac/dc conversion step in individual devices can be removed, com-
prising filter, rectifier, power factor correction and link capacitor. A central rectifier
Single PV-panels
250 W micro MPP dc-lighting
PV-strings tracker (heliox)
String 1
String 2
dc Distribution cabinet dc+ M
24 V nanogrid
dc+ M dc+ M
dc 15 kW solar dc dc
dc MPPT (ENP) dc dc
Solar dc dc
inverters ac ac Central dc bus
L N L N 15 kW 10.5 kW 10 kW
dc dc 3 kW dc dc
central electronic micro-CHP unit bidirectional EV
rectifier ac load ac dccharger dc
(MTT)
(ENP) (heliox)
L1 dc+ M dc+ M
L2
L3
ac mains N
Figure 17.10 Layout of dc office building demonstrator (courtesy: [48])
with common dc bus shared among dc-ready devices can offer size, cost and energy
savings [48]. This design concept can further benefit from emerging fast switching
power electronic devices based on materials such as silicon carbide (SiC) and
gallium nitride (GaN). The microgrid components illustrated in Figure 17.10 were
developed by partners Emerson Network Power, Philips, Heliox, MTT and/or were
retrofitted for 380 V, dc operation, from commercially available ac devices [48].
17.5.2.3 Control
Central energy management system (EMS) can optimize the operation of dc office
building. The central rectifier, for example in [48], can regulate the output dc bus
voltage. Therein, a star point programmable logic controller is employed to monitor
the system parameters and govern the power flow in each subsystem. Information
such as weather forecast and user behavior can be utilize by the EMS to fulfill
higher level control objectives such as building CO2 reduction. Specifically in [50],
it was shown that about 10% reduction in CO2 emissions is achieved just by con-
verting the system from ac to dc, while above 16% are achieved if dc EMS is used.
Heuristic control was used to minimize grid energy exchange, maximize the
available green energy resources and optimize the use of energy storage.
17.5.2.4 Protection
In [48], dc switch technology from ABB is employed to interrupt fault currents.
Fast responding power electronic based solutions may also be considered.
TN-S grounding scheme is used in the test bed, consistent with the ETSI EN 301
605 standard for grounding systems in 400 V dc datacenters [51]. Special attention
must be paid to arc detection, for instance, based on the methods presented in
[47,52].
17.5.2.5 Economics
The study at Xiamen University showed that the installation cost of 150 kW dc
microgrid to power a building is about $2.2/W [53], having a payback of 9 years
and 5.5 years with and without incentives, respectively.
17.5.3 dc Distribution grids

The impediment to the grid level realization of dc distribution is the market inertia
of ac system and the lack of standardization. This subsection looks into projects
specifically aiming to make a bottom-up approach to transition from ac to dc [54].
Three nanogrids, each with 2–5 kW capacity and 51.2 V internal dc bus, were
interconnected by 380 V external dc bus as shown in Figure 17.11 [54]. The open
energy system (OES) was a multilayered, recursively scalable, bottom-up approach
as shown in Figure 17.12 and was realized for a cluster of 19 dc nanogrids [55].
17.5.3.2 Components
A full scale prototype of the idea was realized at the Okinawa Institute of Science
and Technology, Japan as shown in Figure 17.13.
ac grid
Windmill
PV PV
dc sub dc sub dc sub

grid grid grid
Control Control Control

dc/dc dc/dc dc/dc
Communication
Breaker
dc power bus
Control Battery
Battery
dc/dc Battery
Battery
Optional community storage
Figure 17.11 Schematic of the open energy architecture (courtesy: [54])

Power trade among OES

OES OES networks
Local power
trade OES
Figure 17.12 Scalability of the OES concept (courtesy: [54])
Center for
open energy system research
Primary test @ pump room
OES platform @faculty houses
Figure 17.13 Full scale demo prototype of the OES concept (courtesy: [54])
Each nanogrid was based on the labscale prototype, consisting of 3.6 kW h,

45–57.6 V lithium-ion batteries, 2 kW PV charger, 2.5 kW ac/dc converter, 3.5 kW
dc/ac converter and an internal controller, connected to the external dc power bus
using a 2.5-kW bidirectional dc/dc converter. dc Breakers of 10 A are employed.
Additional PV panels of 5, 2.5 and 2.5 kWp are installed at the entrance of the
university. Power exchange was remotely controlled and visualized, showcasing
the technical feasibility and safety of the OES-based setup.
17.5.3.3 Control
The peer-to-peer (P2P) multiagent control for OES dc microgrids is described in
[56]. Highlighting that the standard hierarchical control in IEEE 1547 was limited
to large generation, top-down approach, it was argued that a grid independent
fully decentralized design was necessary for resilient, scalable and self-sufficient
operation. It was anticipated that the role of Information Communication
Technologies for active, multilayered and collaborative P2P control in OES would
be significant.
Ensuring that the failure of any single entity would not result in complete loss
of network services, the 19 autonomous dc nanogrids at Okinawa, Japan were
operated as community-wide, interconnected P2P controlled microgrid [56]. It was
shown that the self-sufficiency ratio, representing minimization of external non-
renewable energy, needs improved by more than 4% when decentralized P2P dc
exchange was allowed as compared to a stand-alone nanogrid. The solar operation
ratio, signifying reduction in maximizing solar generation usage and minimizing
curtailment, improved by almost 10% with dc interconnections as compared to
stand-alone mode. A centralized control approach provided a further improvement
of 4% and 2%, respectively. While it was claimed that an improvement in resilience
could not be shown experimentally, its conceptual discussion was presented [56].
17.5.4 Discussion
A selection of demonstration projects of residential and commercial buildings,
transitioning toward campus dc and universal dc grids is presented in Table 17.4.
Demonstration sites for dc microgrids in residential, commercial buildings and
campus applications
The dc operating voltage was usually selected at 380 V with capacity as high
as 150 kW. Most small-scale projects used radial topology with centralized control.
A scalable, multilayered approach with multiagent control was suggested in [54].
There is potential for a universal dc distribution with meshed, multilayered and
bipolar topology employing distributed control. One important issue discussed in
many initiatives was the protection aspect. Specifically, the localization and
extinction of arcing was considered vital. Impact and reliability of dc switch based
protection is another important domain. Perhaps, a clear demonstration of the
interplay between reliability and availability, supported by a selective and recon-
figurable protection system, will strengthen the acceptability of a meshed universal
dc grid.
Table 17.4 Demonstration sites for dc microgrids in residential, commercial

buildings and campus applications
Description Voltage Capacity (kW) Topology Control Protection

level (V)
1 Taiwan [45] 38020 5–10 Monopolar, Centralized, Special plug
radial Heuristic
2 Japan [57] 420 20 Hybrid ac–dc
3 Bosch [58] 380 15 Radial Centralized
4 Japan [59] 38020 15–100 Centralized
5 Germany [48] 38020 15 Bipolar Centralized dc
380 Switches
6 China [53] 38020 150 Radial Centralized
7 Japan [54] 350–380 2–5 per house Multilayered Multiagent
References
[1] Jhunjhunwala A, Lolla A, Kaur P. Solar-dc microgrid for Indian homes:
a transforming power scenario. IEEE Electrification Magazine. 2016
Jun;4(2):10–19.
[2] Kaur P, Jain S, Jhunjhunwala A. Solar-DC deployment experience in off-grid
and near off-grid homes: Economics, technology and policy analysis. In: IEEE
First International Conference on DC Microgrids; 2015. p. 26–31.
[3] Chakrabarty S, Islam T. Financial viability and eco-efficiency of the solar
home systems (SHS) in Bangladesh. Energy. 2011; 36(8): 4821–4827.
{PRES} 2010.
[4] Mondal MAH. Economic viability of solar home systems: case study of
Bangladesh. Renewable Energy. 2010;35(6):1125–1129.
[5] Liu Z, Shekhar A, Ramı́rez-Elizondo L, et al. Characterization of series arcs
in LVdc microgrids. In: IEEE Second International Conference on DC
Microgrids (ICDCM); 2017.
[6] IRENA. Solar PV in Africa: Costs and Markets. International Renewable
Energy Agency (IRENA). Available at: https://www.irena.org/-/media/
Files/IRENA/Agency/Publication/2016/IRENA_Solar_PV_Costs_Africa_
2016.pdf
[7] Shenai K, Jhunjhunwala A, Kaur P. Electrifying India: using solar dc
microgrids. IEEE Power Electronics Magazine. 2016 Dec;3(4):42–48.
[8] Paleta R, Pina A, Santos Silva CA. Polygeneration energy container:
designing and testing energy services for remote developing communities.
IEEE Transactions on Sustainable Energy. 2014 Oct;5(4):1348–1355.
[9] Sulligoi G, Tessarolo A, Benucci V, et al. Shipboard power generation:
design and development of a medium-voltage dc generation system. IEEE
Industry Applications Magazine. 2013 Jul;19(4):47–55.
[10] Jin Z, Sulligoi G, Cuzner R, et al. Design and control of hybrid power and
propulsion systems for smart ships: a review of developments. Applied
Energy. 2017;194:30–54.
[11] Shekhar A, Ramı́rez-Elizondo L, Bauer P. Next-generation shipboard DC
power system: introduction smart grid and dc microgrid technologies into
maritime electrical networks. IEEE Electrification Magazine. 2016 Jun;
4(2):45–57.
[12] IEEE Recommended Practice for 1 kV to 35 kV Medium-Voltage DC Power
Systems on Ships. IEEE Std 1709-2010. 2010 Nov; p. 1–54. doi: 10.1109/
IEEESTD.2010.5623440.
[13] Shekhar A, Ramı́rez-Elizondo L, Bauer P, et al. DC microgrid islands
on ships. In: IEEE Second International Conference on DC Microgrids
(ICDCM); 2017.
[14] Herbst JD, Gattozzi AL, Ouroua A, et al. Flexible test bed for MVDC and
HFAC electric ship power system architectures for Navy ships. In: 2011
IEEE Electric Ship Technologies Symposium; 2011. p. 66–71.
[15] Chryssostomidis C, Cooke CM. Space reservation for shipboard electric

power distribution systems. In: 2015 IEEE Electric Ship Technologies
Symposium (ESTS); 2015. p. 187–192.
[16] Hansen JF, Lindtjorn JO, Myklebust TA, et al. Onboard dc grid for enhanced
DP operation in ships. In: Marine Technology Society Dynamic Positioning
Conference; 2011.
[17] Hill RJ. Electric railway traction. I. Electric traction and DC traction motor
drives. Power Engineering Journal. 1994 Feb;8(1):47–56.
[18] Steimel A. Under Europe’s incompatible catenary voltages a review of
multi-system traction technology. In: 2012 Electrical Systems for Aircraft,
Railway and Ship Propulsion; 2012. p. 1–8.
[19] Suzuki T, Hayashiya H, Yamanoi T, et al. Application examples of energy
saving measures in Japanese DC feeding system. In: 2014 International
Power Electronics Conference (IPEC-Hiroshima 2014 – ECCE ASIA);
2014. p. 1062–1067.
[20] Uzuka T. Faster than a speeding bullet: an overview of Japanese high-speed
rail technology and electrification. IEEE Electrification Magazine. 2013
Sep;1(1):11–20.
[21] Zielenkiewicz M, Maksimowicz T, Burak-Romanowski R. Coordination of
surge arresters in DC 3 kV railway traction system-field tests. In: 2016 33rd
International Conference on Lightning Protection (ICLP); 2016. p. 1–10.
[22] Song XM, He JH, Yip T, et al. Research into fault location methods for DC
railway traction system. In: 12th IET International Conference on Devel-
opments in Power System Protection (DPSP 2014); 2014. p. 1–6.
[23] Thong M, Cheong A, Wijaya H. Overview of DC traction protection scheme
for Singapore rapid transit system. In: 2005 International Power Engineering
Conference; 2005. p. 1–587.
[24] Van Eck RA. The separately excited DC traction motor applied to DC and
single-phase AC rapid transit systems and electrified railroads, Part II—
Application. IEEE Transactions on Industry and General Applications. 1971
Sep;IGA-7(5):650–657.
[25] Sane S, Sharma S, Prasad SK. Harmonic analysis for AC and DC supply in
traction sub station of Mumbai. In: IEEE International Conference on
Electrical, Computer and Communication Technologies; 2015. p. 1–5.
[26] Hermoso B, Montanya J, Garnacho F, et al. Catalonia (Spain) railway
traction DC system project to limit lightning transient overvoltage. In:
2010 30th International Conference on Lightning Protection (ICLP); 2010.
p. 1–5.
[27] Ku BY, Hsu T. Computation and validation of rail-to-earth potential for
diode-grounded DC traction system at Taipei Rapid Transit System. In:
ASME/IEEE Joint Rail Conference. Proceedings of the; 2004. p. 41–46.
[28] Nikouee S, Ledbetter T. DC traction power supply and distribution system
for MARTA’s Armour Yard Rail Services Facility. In: Proceedings of the
2006 IEEE/ASME Joint Rail Conference; 2006. p. 241–252.
[29] Goh EJ, Chu KN, Ng NK. 1500 V DC traction system for the North East
Line. In: 2004 International Conference on Power System Technology,
2004. PowerCon 2004. vol. 2; 2004. p. 1904–1909.
[30] Kumagai K, Fujita T, Nakahira M, et al. Study on train operation energy
between commuter train and traction substations in a Japanese urban Railway.
In: IEEE Electrical Power and Energy Conference (EPEC); 2015. p. 50–55.
[31] Black, Veatch. Analysis of Smart Grid Technologies. Hawai’i Natural
Energy Institute; 2013.
[32] Roose LR. Hawaii Smart Technology Demonstration for a 100% RE Future
[Presentation]. www.nedo.go.jp: Smart Community Summit 2016; Jun 2016
[cited July, 2017].
[33] Capasso C, Veneri O. Experimental study of a DC charging station for full
electric and plug in hybrid vehicles. Applied Energy. 2015;152:131 – 142.
[34] Powerwall 2 and Solar Roof Launch [Video]. www.tesla.com; Oct 2016
[cited July, 2017].
[35] Tesla and SolarCity [Web News]. www.tesla.com; Nov 2016 [cited July,
2017].
[36] Exploring Future Power Management [Technical News]. www.delta.tudelft.nl;
Apr 2016 [cited July, 2017].
[37] AlLee G, Tschudi W. Edison Redux: 380 Vdc brings reliability and effi-
ciency to sustainable data centers. IEEE Power and Energy Magazine. 2012
Nov;10(6):50–59.
[38] Ton M, Fortenbery B, Tschudi W. DC Power for Improved Data Center
Efficiency. Berkeley, CA: Lawrence Berkeley National Laboratory; 2008
[39] IEEE Recommended Practice for Powering and Grounding Electronic
Equipment - Redline. IEEE Std 1100-2005 (Revision of IEEE Std 1100-
1999) – Redline. 2006 May; p. 1–703.
[40] Aldridge T, Pratt A, Kumar P, et al. Evaluating 400V direct-current for data
centers: a case study comparing 400 Vdc with 480–208 Vac power dis-
tribution for energy efficiency and other benefits. Intel Labs.
[41] Pratt A, Kumar P, Aldridge TV. Evaluation of 400V DC distribution in telco
and data centers to improve energy efficiency. In: INTELEC 07 – 29th
International Telecommunications Energy Conference; 2007. p. 32–39.
[42] Yajima H, Babasaki T, Usui K, et al. Energy-saving and efficient use of
renewable energy by introducing an 380 VDC power-supply system in data
centers. In: IEEE International Telecommunications Energy Conference;
2015. p. 1–4.
[43] Kaga M, Noritake M, Hirose K, et al. Verification of container data center
using 380 V dc power distribution system. In: 2012 International Conference
on Renewable Energy Research and Applications (ICRERA); 2012. p. 1–5.
[44] Patterson BT. DC, come home: DC microgrids and the birth of the ‘‘Enernet’’.
IEEE Power and Energy Magazine. 2012 Nov;10(6):60–69.
[45] Wu TF, Chen YK, Yu GR, et al. Design and development of dc-distributed
system with grid connection for residential applications. In: 8th International
Conference on Power Electronics – ECCE Asia; 2011. p. 235–241.
[46] Mackay L, van der Blij NH, Ramirez-Elizondo L, et al. Toward the universal
DC distribution system. Electric Power Components and Systems. 2017;45
(10):1032–1042.
[47] Shekhar A, Ramirez-Elizondo L, Bandyopadhyay S, et al. Detection of
series arcs using load side voltage drop for protection of low voltage DC
systems. IEEE Transactions on Smart Grid. doi: 10.1109/TSG.2017.2707438.
[48] Wunder B, Ott L, Szpek M, et al. Energy efficient DC-grids for commercial
buildings. In: 2014 IEEE 36th International Telecommunications Energy
Conference (INTELEC); 2014. p. 1–8.
[49] Environmental Engineering (EE); Power supply interface at the input to
telecommunications and datacom (ICT) equipment; Part 3: Operated by
rectified current source, alternating current source or direct current source up
to 400 V; [ETSI EN 300 132-3-1]. European Standards; 2012.
[50] Noritake M, Yuasa K, Takeda T, et al. The demonstration of the CO2 reduc-
tion effect of 400V class DC micro grid for offices. In: 2015 IEEE Interna-
tional Telecommunications Energy Conference (INTELEC); 2015. p. 1–6.
[51] Environmental Engineering (EE); Earthing and bonding of 400 VDC data and
telecom (ICT) equipment [ETSI EN 301 605]. European Standards; 2013.
[52] Strobl C. Arc fault detection in DC microgrids. In: 2015 IEEE First Inter-
national Conference on DC Microgrids (ICDCM); 2015. p. 181–186.
[53] Zhang F, Meng C, Yang Y, et al. Advantages and challenges of DC micro-
grid for commercial building a case study from Xiamen university DC
microgrid. In: IEEE First International Conference on DC Microgrids
(ICDCM); 2015. p. 355–358.
[54] Werth A, Kitamura N, Tanaka K. Conceptual study for open energy systems:
distributed energy network using interconnected DC nanogrids. IEEE
Transactions on Smart Grid. 2015 Jul;6(4):1621–1630.
[55] Werth A, Tokoro M, Tanaka K. Bottom-up and recursive interconnection for
multi-layer DC microgrids. In: 2016 IEEE 16th International Conference on
Environment and Electrical Engineering (EEEIC); 2016. p. 1–6.
[56] Werth A, André A, Kawamoto A, et al. Peer-to-peer control system for DC
microgrids. IEEE Transactions on Smart Grid. 2018 Jul;9(4):3667–3675.
[57] Hirose K, Reilly JT, Irie H. The Sendai microgrid operational experience in
the aftermath of the Tohoku earthquake: a case study. New Energy and
Industrial Technology Development Organization (NEDO); 2013.
[58] Fregosi D, Ravula S, Brhlik D, et al. A comparative study of DC and AC
microgrids in commercial buildings across different climates and operating
profiles. In: IEEE First International Conference on DC Microgrids; 2015.
p. 159–164.
[59] Noritake M, Yuasa K, Takeda T, et al. The demonstration of the CO2
reduction effect of 400V class DC micro grid for offices. In: 2015 IEEE
International Telecommunications Energy Conference (INTELEC); 2015.
p. 1–6.
Index
AC–DC converter control 258, 299 energy storage system (ESS) 274
balancing leg control 261 power distribution 269–73
direct power control (DPC) 260–1 power generation 268–9
single-phase control 258–9 power quality requirements in
symmetrical component aircrafts 274–6
control-based methods 261–2 power utilization 273
voltage-oriented control 259–60 stability analysis 285–8
AC generators 297–8 starter/generator control 276–8
AC microgrid system 1, 20, 368–70 Aircraft Electrical Generation with
AC power systems 1, 153, 301 Active Rectification
AC residential buildings 378–9 Technology (AEGART)
battery bank power in 384 project 276, 278
conceptual layout of 377 aircraft power systems 84–5
conversion losses in 385 all-electric ship (AES) 295
conversion power losses in internal architecture of dc microgrids 390
converters loads in 385 LVDC microgrid
and DC residential buildings, for data centers 394–5
comparison between 377, 382–6 for homes 395–6
active-bridge converter 236–8 for space applications 392–3
active current-sharing techniques 169 MVDC and LVDC microgrid for
active front end (AFE) rectifier dc distribution systems 390–1
276, 280 MVDC microgrid
active power filters in HV battery for marine applications 393–4
chargers 328–30 for oil-drilling applications 396
AC vs DC microgrid system, autonomous load control 164, 175
mathematical analysis of 373 architecture 165–6
total daily load (TDL) 373–4 control levels 164–5
voltage, current and power losses in strategy for controlling the load to
DC supply 374–7 be an energy asset 166–8
adjustable speed drives (ASD) 353 auto-transformer rectifier unit
Aggregators 199 (ATRU) 270
aircraft DC microgrids 267 auxiliary loads 323–4
control strategies in aircraft DC auxiliary power module (APM) 323
microgrids 279 auxiliary power unit (APU) 269
primary control 280–2 average current sharing (ACS)
secondary control 283–5 technique 15–17
back-up power 351–2 cascade control 257–8

balance-of-system (BOS) 405 centralized coordination 101–2
balancing leg control 261 centralized P/EMS 92–3
balancing topologies 253 CHAdeMO charging protocol 190
bidirectional buck-boost topologies CHAdeMO interface 190
254–5 circuit breakers (CBs) 64
coupled inductor current circular-chain-control (CCCs) 36
redistributor 255–6 closed-loop controller 172
three-level DC–DC current conductive HV battery chargers 325–8
redistributors 256–7 connectors 356
Basic Insulation Level tests 206 consensus algorithms 35–6
Battery bank power in AC and DC consensus global average voltage
residential building 384 estimator 35
battery-based storage 154 Constant Current (CC) mode 193
Battery Management System (BMS) 193 constant frequency (CF) aircraft
bipolar configuration 68–9 system 269
bipolar LVDC system 372 constant power loads (CPLs) 43–5,
bipolar-type DC microgrids 245 58–9, 164, 166, 170, 285
balancing topologies 253 Constant Voltage (CV) mode 193
bidirectional buck-boost contactor relay (CR) 315
topologies 254–5 contingency analysis 178
coupled inductor current continuous conduction mode (CCM)
redistributor 255–6 227, 325–6
three-level DC–DC current controller area network (CAN bus) 17
redistributors 256–7 control strategies in aircraft DC
bipolar-type DC distribution microgrids 279
systems 247–8 primary control 280–2
control schemes 257 secondary control 283–5
AC–DC converter control 258–62 converter-based protection scheme,
cascade control 257–8 time sequence for 313
topologies and operational converter topologies 194
aspects 249 in electrified vehicles 325
distribution converter topologies active power filters in HV battery
249–53 chargers 328–30
boost and buck-boost converters 57 conductive HV battery chargers
bridgeless totem-pole PFC (BTPPFC) 325–8
326–7 LV auxiliary power modules
building automation system, basic 330–1
network architecture of 380 cooling system 301, 353–4
building interiors 427 coordinated protection of DC
architecture 427 microgrids 63
components 427–8 coordinated protection techniques 71
control 428 AC side protection 72–3
economics 429 applications 80–5
protection 428–9 DC side protection 73–9
Index 439
faults in dc power systems 65 incumbent cost and familiarity

fault current analysis 66–8 advantages 361
faults in various bus overly optimistic claims 360
configurations 68–71 protection and DC circuit
fault types and behavior 65–6 breakers 360–1
coupled inductor current redistributor LVDC microgrid for 394–5
255–6 power quality 358
critical conduction mode (CRCM) protection 357–8
325–6 reliability 349
current control/grid-following mode back-up power 351–2
107–8 fault tolerance 350–1
current control mode (CCM) 105–7 stability 358
current-mode droop control 8 DC/AC converters 299–301
current regulator 36–8 DC auxiliary loads in electrified
vehicles 323–5
data centers 83–4, 343, 423–5 DC bus, structure of 296–7
development of DC power DC bus bars, design considerations of
distribution in 345–6 333–6
efficiency 346–9 dc bus signaling (DBS) 19, 97
existing high voltage DC data DC circuit breakers (CBs) 301
centers 358 dc converters with high-frequency
Duke Energy Data Center, transformers 403–4
Charlotte, NC, USA 359 DC/DC converters 77, 197, 299
Green Datacenter, Zurich, dc–dc regulators 164
Switzerland 359 dc distribution grids 429
NTT Cloud, Tokyo, Japan 359 architecture 429
Texas Advanced Computing components 429–30
Center, Austin, TX, USA control 430–1
359–60 dc Fast charger power converter
installation 354 topologies 194–8
connectors 356 dc fast charging systems and
grounding 355 standards 190–2
isolation 354–5 dc house 425
total cost of ownership (TCO) architecture 425
356–7 components 425
wiring 355–6 control 426
integration with other DC sources protection 426
and loads 352 DC-link capacitors, selections of
cooling 353–4 332–3
lighting 354 DC load systems 44
renewable and distributed energy DC microgrid system 371–2
sources 353 DC residential buildings 379
key obstacles to widespread and AC residential buildings,
adoption of 360 comparison between 377, 382–6
emergence of rack-level UPS 360 battery bank power in 384
conceptual layout of 378 distributed approaches 24, 34

conversion losses in 385 fully communicated control 34–5
conversion power losses in internal sparse communicated (consensus-
converters loads in 385 based) control 35–6
decentralized coordination 102–3 sparse communicated control using
decentralized power sharing current information 36
approaches 23–4 current regulator 36–8
frequency droop control 31–4 voltage regulator 38–40
mode-adaptive (autonomous) droop distributed control 13, 95, 103–4, 165
control 24–8 distributed coordination 103–5
nonlinear droop control 28–31 distributed energy resources (DERs) 368
Delta Products Corporation 359 distributed energy storage devices
demand-side management 180 (DESD) 121, 142
demonstration sites of dc distributed generators 70, 246
microgrids 415 distributed LCs (DCLs) 97, 103
datacenters 423 distributed P/EMS 93–4
architecture 424 distributed renewable energy resources
components 424 (DRER) 119, 121, 142
discussion 424–5 distributed secondary controller 35–6
protection and grounding distribution converter 247, 249, 257
system 424 three-level neutral point clamped
off-grid dc microgrids 415 converter 251–3
architecture 415 two-level voltage source converter
components 417 250–1
control and operational distribution generation 367
safety 417 distribution system operator (DSO) 19
discussion 417–18 double-diode model 400–1
socioeconomic impact 417 droop control 2–6, 8–9, 28, 101, 280
residential and commercial dc for stability and information
buildings 425 communication 168
building interiors 427–9 constant power load and its
dc distribution grids 429–31 deleterious effect on dc
dc house 425–6 systems 169–70
discussion 431 dynamic stabilization 172–5
transportation electrification 418 steady state stabilization 170–2
discussion 422–3 droop-controlled PV microgrid,
railway traction with dc supply simulation of 407–10
421–2 droop gain 10
shipboard dc microgrids dual active bridge (DAB) converter
418–21 197–8, 229, 299, 330–1, 337
diesel engines (DEs) 296 analysis of the DAB using the PSM
digital ACS (DACS) 17 231–3
direct power control (DPC) 260–1 current stresses on the DAB 233–5
discontinuous conduction mode efficiency calculation 235
(DCM) 227, 325 modulation strategy 230–1
Index 441
dual half-bridge (DHB) converter 198 DC auxiliary loads in 323–5

dual-voltage charging systems, power electronic system in 321–3
topological reconfigurations practical design considerations 331
in 339–40 DC bus bars, design
Duke Energy Data Center, Charlotte, considerations of 333–6
NC, USA 359 DC-link capacitors, selections of
dv/dt-based dynamic load 332–3
interruptions 180–1 switching devices, selections of
DYMOLA 172 331–2
dynamic characteristics of dc topological reconfigurations 337
microgrids 43–5 in dual-voltage charging systems
dynamic consensus protocol 35 339–40
dynamic power sharing 10–11 in HV charging and propulsion
systems 338–9
efficiency 346–9 electromagnetic interference (EMI)
electrical (galvanic) isolation 354–5 emissions 220
electrical power system (EPS) electronic transformer 215
267, 269 energy management system (EMS)
electric vehicle (EV) 160–1 3, 428
electric vehicle (EV) charging energy management systems for dc
infrastructure and dc microgrids 91
microgrids 189 coordinated control 101
dc fast charger power converter centralized coordination 101–2
topologies 194–8 decentralized coordination
dc fast charging systems and 102–3
standards 190–2 distributed coordination 103–5
EV and EVSE markets and trends current control/grid-following
189–90 mode 107–8
future trends 208–9 illustrative example 108–10
microgrid topologies 199 interactive power/energy
medium-voltage dc fast chargers management strategies 92
203–7 centralized P/EMS 92–3
state-of-the-art dc fast charger distributed P/EMS 93–4
installation 200–3 hybrid P/EMS 94–5
state-of-the-art EV dc fast chargers local control functionalities 99–101
192–3 passive power/energy management
electric vehicle supply equipment strategies 95–7
(EVSE) 189 voltage control/grid-forming mode
electrified vehicles 321 105–7
converter topologies in 325 energy storage system (ESS) 96, 274,
active power filters in HV battery 298, 389–90
chargers 328–30 shipboard MVDC power systems
conductive HV battery chargers 302–3
325–8 EU Low Voltage Directive (LVD
LV auxiliary power modules 330–1 2014/35/EU) 246
facility-level battery system 347 hierarchical control for DC MGs 2–3

fault current analysis 66–8 high frequency transformer (HFT) 220
faults in bus configurations 68 dc converters with 403–4
multiterminal configurations 69–71 high-voltage (HV) battery 321
point-to-point (PtP) configuration high-voltage (HV) charging and
68–9 propulsion systems
faults in dc power systems 65 topological reconfigurations in
fault current analysis 66–8 338–9
faults in various bus configurations 68 high-voltage DC (HVDC) transmission
multiterminal configurations systems 63, 80, 245, 371
69–71 Hikari 359
point-to-point (PtP) hybrid AC–DC microgrid system 370
configuration 68–9 hybrid CBs 301
fault types and behavior 65–6 hybrid P/EMS 94–5
fault tolerance 350–1 hysteresis 175
fault types and behavior 65–6
flux-oriented control (FOC) 259 Idaho National Laboratory (INL) 200
FREDM vision with SST as an IEC 62196-3 Standard 190–1
intelligent energy router 208 impedance-based method to assess
frequency droop control 31–4 system stability 287
‘frequency-wild’ systems 269 instantaneous constant power loads,
fuel cells (FC) 353 operation of rectifiers with
fully communicated control 34–5 58–60
Future Renewable Electric Energy insulated-gate bipolar transistors
Delivery and Management 208 (IGBTs) 194–5, 250–1, 301, 331
integrated-gate commutated thyristors
gallium nitride (GaN) power (IGCTs) 301
MOSFETs 194 integrated power system (IPS) 418
gas turbines (GTs) 296 integrated propulsion system (IPS) 295
GB/T 190 intelligent energy management (IEM)
gen-sets 301–2 algorithm 121, 123
Google Little Box Challenge 347 interactive power/energy management
Green Datacenter, Zurich, strategies 92
Switzerland 359 centralized P/EMS 92–3
grid-scale storage 154 distributed P/EMS 93–4
grid-tied converter 258 hybrid P/EMS 94–5
ground-fault detection 314–15 International Electrotechnical
ground-fault location 315–17 Commission (IEC) 190
ground-fault protection devices inverter-less remote unit (IRU) 416
(GFPDs) 82–3
grounding 247, 355 Junction Barrier Schottky (JBS)
diodes 195
heating, ventilation, and air
conditioning (HVAC) ladder configuration 71
157–8, 160 large signal stability analysis 285, 311
Index 443
Lawrence Berkeley National low-frequency-voltage insulation

Laboratory (LBNL) 423–5 tests 206
light detection and ranging (LiDAR) 325 low-voltage (LV) auxiliary power
lighting load in data centers 354 modules 330–1
linear analysis tools 287 low-voltage (LV) battery 321
line-commutated converters (LCCs) 64 low voltage direct current (LVDC)
line-to-ground fault 65–7, 69, 74–5, active networks 262–3
81–3, 312–13, 315 distribution system 63, 367
line-to-line fault 65–6, 69, 82–3, 312 microgrid
liquefied natural gas (LNG) 298 for data centers 394–5
load as a controllable energy asset in dc for dc distribution systems 390–1
microgrids 153 for homes 395–6
autonomous load control 164 for space applications 392–3
architecture 165–6 Lyapunov-based method 285
control 164–5
strategy for controlling the load to marine applications, MVDC
be an energy asset 166–8 microgrid for 393–4
benefit of load control 155–8 market-based incentive programs 154
droop control for stability and MATLAB Power System Toolbox 176
information communication 168 maximum power point tracking
constant power load and its (MPPT) 26, 96, 122, 401
deleterious effect on dc mechanical CBs 301
systems 169–70 medium voltage dc (MVDC) 295, 418
dynamic stabilization 172–5 fast chargers 203–7
steady state stabilization 170–2 link 222
dv/dt-based dynamic load microgrid
interruptions 180–1 for dc distribution systems 390–1
load modulation 159–61 for marine applications 393–4
load prioritization and scheduling for oil-drilling applications 396
181–2 medium voltage EV fast chargers 204
local area power and energy system mesh configuration 71
154–5 microgrids 153, 367
time-scale of energy classification of 368
requirements 161 AC microgrid system 368–70
long-term transients 163–4 DC microgrid system 371–2
short-term transients 161–3 hybrid AC–DC microgrid
voltage-based load interruption 175 system 370
contingency analysis 178 topologies for EV charging 199
power flow analysis 176–7 medium-voltage dc fast chargers
search algorithm 178–80 203–7
load-shed programs 154 state-of-the-art dc fast charger
local area power and energy system installation 200–3
(LAPES) 153–5, 165 MIL-STD-407F 275
long-term transients 163–4 mode-adaptive (autonomous) droop
low frequency transformer (LFT) 215 control 24–8
modular multilevel converter (MMC) photovoltaic (PV) system modeling 398

topology 299–300 double-diode model 400–1
molded-case CBs (MCCBs) 64 ideal model of a PV cell 398–9
monopolar configuration 68 single diode with series resistance
more electric aircraft (MEA) concept model 399–400
267–8, 272 single-diode with shunt resistance
multi-agent system (MAS) 94 model 400
multiple-active-bridge (MAB) 229 photovoltaic generators 367
multiterminal configurations photovoltaic panels 63
69–71 photovoltaic park 82–3
multiterminal DC (MTDC) photovoltaic powered systems, dc
systems 63 microgrids for 389, 397–8
architecture of dc microgrids 390
nondominated sorting genetic LVDC microgrid for data centers
algorithm (NSGA-II) 156 394–5
nonlinear droop control 28–31 LVDC microgrid for homes 395–6
NTT Cloud, Tokyo, Japan 359 LVDC microgrid for space
Nyquist plot 311 applications 392–3
Nyquist stability criterion 288 MVDC and LVDC microgrid for
dc distribution systems 390–1
off-grid dc microgrids 415 MVDC microgrid for marine
architecture 415 applications 393–4
components 417 MVDC microgrid for oil-drilling
control and operational safety 417 applications 396
discussion 417–18 control of 406–7
socioeconomic impact 417 droop-controlled PV microgrid,
off-shore oil and gas drilling simulation of 407–10
systems 396 power converter technologies 401
oil-drilling applications, MVDC dc converters with high-frequency
microgrid for 396 transformers 403–4
open energy system (OES) 429 next generation PV converters
overcurrent protection devices 404–6
(OCPDs) 82 transformerless topologies 401–3
ownership, total cost of 356–7 PV system modeling 398
double-diode model 400–1
Pareto Frontier technique 156, 183 ideal model of a PV cell 398–9
partially modular approach 220 single diode with series resistance
passive power/energy management model 399–400
strategies 95–7 single-diode with shunt resistance
permanent magnet machine model 400
(PMM) 276 photovoltaics (PVs) 353
permanent magnet SG (PMSG) 297 photovoltaic solar systems 96
phase locked loop (PLL) 142 plug in hybrids (PHEV) 160
phase shift 198 point of common coupling (PCC)
phase shift modulation (PSM) 230 154, 368
Index 445
point-of-load (POL) control 153, 166 tertiary control algorithm 134

point-of-load (POL) converter change SST DC terminal output
43–5, 176 voltage 134–5
point-to-point (PtP) configuration shifting battery droop curve 135
68–9 power quality 358
portional-integral-derivative (PID) in aircrafts 274–6
controllers 100 power sharing approaches 23
power converter 298 power sharing errors 6–8
AC/DC converters 299 power source converter interface
DC/AC converters 299–301 (PSCI) 46–58
DC circuit breakers (CBs) 301 primary control 2–3
DC/DC converters 299 basics of droop control 3–6
technologies droop strategies 8–10
dc converters with high-frequency dynamic power sharing 10–11
transformers 403–4 interfaces to upper levels 12–13
next generation PV converters power sharing errors 6–8
404–6 primary controller 51
transformerless topologies 401–3 prime movers 297
topologies 224 propulsion motors 295
basic module topologies 225–6 propulsion system 323
requirements 224–5 protection and DC circuit breakers
power electronics 249 360–1
power electronics interface (PEI) 165 pulse-width modulation (PWM) 250
power electronic system in electrified rectifier 194
vehicles 321–3 signals 326
power/energy management schemes 92
interactive 92–5 quadruple active bridge (QAB) 235–6
passive 95–7 Queen Elizabeth II 295
power factor correction (PFC) 194,
276, 348 rack-level UPS, emergence of 360
power flow algorithm 177 radial configuration 70–1
power line communication (PLC) 92 railway traction with dc supply 421
power line signaling (PLS) 18, 97, 103 architecture 421–2
power management control, shipboard components 422
MVDC power systems 303 protection 422
case study 305 reliability in data centers 349
load profile 306 back-up power 351–2
model and control of MVDC IPS fault tolerance 350–1
305–6 renewable and distributed energy
simulation results 306–10 sources 353
central control 304–5 renewable energy resource (RES) 367
local control 303–4 residential and commercial dc
power management strategy 131 buildings 425
primary control algorithm 131–2 building interiors 427–9
secondary control algorithm 133–4 dc distribution grids 429–31
dc house 425–6 secondary droop control 53

discussion 431 secondary networks 380–1
residential buildings, DC self-inductance 335
microgrid in 367 series resistance model, single diode
AC and DC residential buildings, with 399–400
comparison between 377 series resonant converter (SRC)
AC residential buildings 378–9 226, 239
comparison AC and DC residential current stress 228
buildings 382–6 efficiency expectation 229
conceptualisation 368 series-stacked power architecture 348
DC residential buildings 379 shifting battery droop curve 135
mathematical analysis of AC vs DC shipboard dc microgrids 418
microgrid system 373 architecture 418
total daily load (TDL) 373–4 components 418
voltage, current and power losses economical operation 421
in DC supply 374–7 power quality, control and
microgrids, classification of 368 protection 419–20
AC microgrid system 368–70 shipboard MVDC power systems 295
DC microgrid system 371–2 components of 297
hybrid AC–DC microgrid AC generators 297–8
system 370 energy storage systems 298
smart DC residential buildings power converters 298–301
advantages, challenges and ship loads 298
barriers of 381–2 configuration of 296
smart DC residential buildings, energy storage systems 302–3
automation architecture for 379 faults and protection 312
automation level 380 architecture of protection
field level 379 schemes 312–13
field network 380 ground-fault detection 314–15
management level 381 ground-fault location 315–17
primary and secondary network gen-sets 301–2
380–1 power management control 303
topologies for DC microgrid 372 central control 304–5
bipolar LVDC system 372 load profile (case study) 306
unipolar LVDC system 372 local control 303–4
residential smart EMS (RSEMS) 108 model and control of MVDC IPS
ring configuration 70 (case study) 305–6
simulation results (case study)
secondary control 2, 13 306–10
centralized approach 13–15 prime movers 297
distributed approach 15 ring bus structure of 297
dedicated communication channels, stability analysis 310–12
communication through 15–18 structure of DC bus 296–7
power lines, communication shipboard power system 80–1
through 18 hierarchical control structure of 303
Index 447
ship loads 298 architecture and benefits 119–21

short-term transients 161–3 centralized power
shunt resistance model, single-diode management of 121
with 400 islanding mode 128
silicon carbide (SiC) 194 islanding mode transition 128
SiC MOSFETs 195, 229 passive grid interaction (case
silicon-controlled rectifier (SCR) 299 study) 124–6
simple network management protocol power management strategy 121–4
(SNMP) 380 transition from passive grid
single-phase AC–DC converter interaction mode to active grid
control 258–9 interaction mode 126–8
small signal stability 287 power management strategy 131
smart DC residential buildings primary control algorithm 131–2
advantages, challenges and barriers secondary control algorithm 133–4
of 381–2 tertiary control algorithm 134–5
automation architecture for 379 small-scale DC microgrid 136–42
automation level 380 comparison of two tertiary control
field level 379 methods (case study) 139–42
field network 380 first tertiary control
management level 381 (case study) 139
primary and secondary network primary control (case study) 136–7
380–1 secondary control (case study)
smart transformer (ST) architecture 137–9
classification 217–18, 223–4 second tertiary control (case
dc Link voltage availability 221–2 study) 139
modularity 219–20 as solid-state synchronous machine
modularity level 220–1 (SSSM) 142
power conversion stages 218–19 case study 147–50
Society of Automotive Engineers concept of 142
(SAE) 190 frequency regulation 144–5
solid-state CBs (SSCBs) 301, 312 power up/down reserve support
solid-state synchronous machine 145–7
(SSSM) 119, 142 voltage regulation 147
case study 147 solid-state transformers (SST) 203,
islanding and reconnection 150 206, 208–9, 215
load change 148 active-bridge converter 236–8
power up/down reserve 148–50 concept 215
source change 148 dc–dc stage 224
concept of 142 basic module topologies 225–6
frequency regulation 144–5 requirements 224–5
power up/down reserve support dual active bridge (DAB)
145–7 converter 229
voltage regulation 147 analysis of the DAB using the
solid-state transformer (SST)-enabled PSM 231–3
DC microgrids 119 current stresses on the DAB 233–5
efficiency calculation 235 symmetrical component control-based

modulation strategy 230–1 methods 261–2
in electric distribution grid synchronous generator (SG) 3,
application 216–18 32, 297
series resonant converter
(SRC) 226 tertiary control 3, 18–20
current stress 228 Tesla Superchargers 193
efficiency expectation 229 Texas Advanced Computing Center,
smart transformer (ST) architecture Austin, TX, USA 359–60
classification 218, 223–4 three-level DC–DC current
dc Link voltage availability redistributors 256–7
221–2 three-level neutral point clamped
modularity 219–20 (3L-NPC) converter 249, 251–3
modularity level 220–1 total cost of ownership (TCO) 356–7
power conversion stages 218–19 total daily load (TDL) 373–4
in traction application 216 total harmonic distortion (THD) 206
space applications, LVDC microgrid totem pole 326
for 392–3 traction power systems 81–2
space vector modulation (SVM) 250 transformerless topologies 401–3
sparse communicated control 35–6 transformer rectifier units (TRUs) 270
using current information 36–40 transmission system operator
stability 358 (TSO) 19
stability analysis, in aircraft DC transportation electrification 418
microgrids 285–8 discussion 422–3
stability analysis and stabilization, of railway traction with dc supply 421
dc microgrids 43, 45–8 architecture 421–2
control strategies for components 422
stable operation 50–8 protection 422
dynamic characteristics 43–5 shipboard dc microgrids 418
operation of rectifiers with architecture 418
instantaneous constant power components 418
loads 58–60 economical operation 421
passive approaches for 48–50 power quality, control and
star configuration 71 protection 419–20
State of Charge (SOC) 193 trapezoidal modulation (TPM) 231
state-of-the-art dc fast charger triangular current mode (TCM) 231
installation 201 twice-line frequency energy
state-of-the-art EV dc fast chargers buffering 347
192–3 two-level voltage source converter
super-Sepic- and super-Zetatype (2L-VSC) 249–51
voltage balancers 254
sustaining time 163 under-voltage protection (UVP) 164
switching devices, uninterruptible power supply (UPS)
selections of 331–2 83, 347
switching surface 56 unipolar LVDC system 372
Index 449
University of Texas at Austin center voltage-based real and reactive power

for electromechanics control strategy 107
(UT-CEM) 419 voltage control/grid-forming mode
Utility Direct Fast Charger (UDFC) 203 105–7
voltage control mode (VCM) 105
variable-frequency (VF) generations 269 voltage droop control 2
variable-speed CF (VSCF) aircraft 269 voltage-oriented control (VOC)
variable speed variable-frequency 259–60
(VSVF) 269 voltage regulator 38–40
Vehicle Identification Number voltage-source converters (VSCs) 64
(VIN) 193
vehicle-to-grid (V2G) technology 199 wide bandgap (WBG) semiconductor
V–I droop control concept 4–5 power devices 203, 209
Vienna rectifier 196 wind turbines (WTs) 96
virtual resistance 3 wing ice protection system (WIPS)
voltage-based load interruption 175 272–3
contingency analysis 178 wiring 355–6
power flow analysis 176–7
search algorithm 178–80 zero-voltage switching (ZVS) 196, 326

1785613820

Uploaded by

Copyright:

Available Formats

1785613820

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1785613820

Uploaded by

Copyright:

Available Formats

What are some applications of smart transformers discussed in the document?

What are some applications of smart transformers discussed in the document?

What are some components of shipboard DC microgrids mentioned in the document?

What are some components of shipboard DC microgrids mentioned in the document?

IET ENERGY ENGINEERING 115

The Institution of Engineering and Technology

The Institution of Engineering and Technology

British Library Cataloguing in Publication Data

ISBN 978-1-78561-382-1 (hardback)

Typeset in India by MPS Limited

1 DC microgrid control principles – hierarchical control diagram 1

2 Distributed and decentralized control of dc microgrids 23

3 Stability analysis and stabilization of DC microgrids 43

4 Coordinated protection of DC microgrids 63

5 Energy management systems for dc microgrids 91

6 Control of solid-state transformer-enabled DC microgrids 119

6.3 Centralized power management of solid-state transformer-based

7 The load as a controllable energy asset in dc microgrids 153

7.7 dv/dt-Based dynamic load interruptions 180

8 Electric vehicle charging infrastructure and dc microgrids 189

9 Overview and design of solid-state transformers 215

9.9 Summary 238

10 Bipolar-type DC microgrids for high-quality power distribution 245

11 Aircraft DC microgrids 267

12 Shipboard MVDC microgrids 295

12.2.2 Prime movers 297

13 DC-based EVs and hybrid EVs 321

14 DC data centers 343

14.5.2 Cooling 353

15 DC microgrid in residential buildings 367

15.10 Advantages, challenges and barriers of smart DC residential

16 DC microgrids for photovoltaic powered systems 389

17 Demonstration sites of dc microgrids 415

17.3 Transportation electrification 418

The current revolution in power electronics has enabled direct-current (DC)

Local Local Local

Figure 1.1 Hierarchical control of an MG

1.2 The hierarchical control for DC MGs

● The tertiary control takes responsibility of power management between MG

1.3 Primary control

1.3.1 Basics of droop control

Figure 1.2 V–I droop characteristic

Figure 1.3 A simple MG with two DGs and a resistive load

Figure 1.4 Droop curves for DG1 and DG2

idc1 RD1 RD2 idc2

Figure 1.6 The equivalent circuit of two DC power supplies

Figure 1.7 An illustration of V–I droop with maximum allowed deviation

1.3.2 Power sharing errors

Vdc Large Small

Figure 1.8 Unequal load sharing due to load distribution in DC MG

Figure 1.9 A simplified DC MG with cable resistance

1.3.3 Droop strategies

I–V droop P–V droop

I–V droop idc

1.3.4 Dynamic power sharing

Figure 1.14 Composite of a DC MG with hybrid energy storages

Figure 1.16 Frequency-domain impedance comparison of converters [15]

1.3.5 Interfaces to upper levels

Figure 1.17 Interfaces of primary level to upper levels

Figure 1.18 Implications of changing interface variables to the droop curve