This articleDownloaded

has been accepted for http://iranpaper.ir

from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1

A Superconducting Magnetic Energy

Storage-Emulator/Battery Supported
Dynamic Voltage Restorer
A. M. Gee, F. Robinson, Member, IEEE and W. Yuan.

 are defined in IEEE Std. 1159-1995 [3] as a decrease to

Abstract— This study examines the use of superconducting between 0.1 and 0.9 p.u. (per unit) r.m.s voltage for durations
magnetic and battery hybrid energy storage to compensate grid of 0.5 cycles to 1 min. They occur more frequently than long-
voltage fluctuations. The superconducting magnetic energy
term under-voltages with significant costs to industry [4].
storage system (SMES) has been emulated by a high current
inductor to investigate a system employing both SMES and Long-term under-voltage events are defined as a measured
battery energy storage experimentally. The design of the voltage less than 0.8-0.9 p.u. r.m.s voltage, lasting longer than
laboratory prototype is described in detail, which consists of a one minute [3] and can lead to load shedding and potentially
series-connected three phase voltage source inverter used to to voltage collapse [5]. The study below presents a means by
regulate AC voltage, and two bidirectional DC/DC converters which both short-term and long-term voltage fluctuations can
used to control energy storage system charge and discharge. ‘DC
be mitigated at the load using short-term magnetic energy
bus level signaling’ and ‘voltage droop control’ have been used to
automatically control power from the magnetic energy storage storage and long-term battery energy storage.
system during short-duration, high power voltage sags, while the
battery is used to provide power during longer-term, low power II. LITERATURE REVIEW
under-voltages. Methods to mitigate long-term voltage disturbance, such as
Energy storage system hybridisation is shown to be
advantageous by reducing battery peak power demand compared load disconnection [6] or modification of loads for greater
with a battery-only system, and by improving long term voltage low-voltage ride-through capability may be impractical [7].
support capability compared with a SMES-only system. Alternatively, supply voltage can be stabilised by tap changing
Consequently, the SMES/battery hybrid DVR can support both transformers, uninterruptable power supplies (UPS), shunt-
short term high-power voltage sags and long term undervoltages connected compensators, or dynamic voltage restorer (DVR)
with significantly reduced superconducting material cost systems. Tap changing transformers have been shown to suffer
compared with a SMES-based system.
from a slow response time and can only output discrete
Index Terms-- Dynamic Voltage Restorer (DVR), Energy voltage levels [8]. UPS systems provide the complete voltage
Storage Integration, Sag, Superconducting Magnetic Energy waveform during a power failure and may prove costly and
Storage, Battery. unnecessary in the event of partial voltage sags. A DVR is a
series-connected device capable of voltage compensation with
I. INTRODUCTION fast response time by injecting a voltage in series with the
supply.
T HEimprovement of power quality is an important
objective for electrical utilities and industrial and
commercial consumers. Highly intermittent distributed
DVR systems can be self-supporting by using power from
the grid to mitigate disturbances [9]. Alternatively, DVR
generation, rapidly changing loads, and direct-off-line power systems can use energy storage to provide power during
electronic systems all contribute to reduced power quality compensation such as capacitors [10] for short-term storage or
causing equipment downtime, overload and failure leading to batteries [11] for longer-term storage. Nielsen and Blaabjerg
lost revenue [1]. [12] have shown that capacitor-supported DVR systems can
Voltage disturbance is a common problem and under- suffer from relatively poor performance for severe and long
voltage conditions have been seen to occur more frequently duration sags. A recent study has shown that an ultra-capacitor
than overvoltage conditions [2]. Short-term under-voltage sags based DVR [13] can be used to mitigate short-term voltage
sags lasting less than one minute. Wang and Venkataramanan
[14] have shown that flywheels are a viable short-term energy
Manuscript received 02/11/2015. This work was supported in by EPSRC storage technology for use with voltage restorer systems both
grant EP/K01496X/1.
Dr. F. Robinson (e-mail: F.V.P.Robinson@bath.ac.uk) and Dr. W. Yuan experimentally and by simulation. Kim et al. [15] have
(e-mail: W.Yuan@bath.ac.uk) are with the Electronic and Electrical described a 3 MJ/750 kVA SMES-based DVR system and
Engineering Department, University of Bath, Bath BA2 7AY, U.K. shown experimental results confirming that SMES is suitable
A. Gee (e-mail: ag15969@bristol.ac.uk) was with the Electronic and
Electrical Engineering Department, University of Bath, Bath BA2 7AY, U.K for the compensation of short-term voltage sags. Shi et al. [16]
and is now with Department of Electrical and Electronic Engineering, have used a system-level simulation to also show that SMES
University of Bristol, Bristol, BS8 1UB, U.K.

0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This articleDownloaded
has been accepted for http://iranpaper.ir
from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2

Figure 2. Vector diagram of DVR control strategies [1] Udvr1: ‘In

phase compensation’. Udvr2 ‘Pre-sag compensation’. Udvr3: ‘Energy
optimal control’. (b) DVR control system.

Figure 1. Hybrid energy storage DVR system configuration.

III. METHODOLOGY

energy storage is capable of compensating voltage sags lasting Fig. 1 shows the DVR system considered. The SMES has
100ms. been emulated by a 15mH, 100A inductor. During a voltage
Short-term voltage compensation alone may not be error a three-phase inverter is used to generate the
sufficient to protect a sensitive load as both long-term [5], [17] compensation voltage at the primary of the injection
and short-term [2-4] voltage stability has been shown to transformers (T1-T3) so that the load voltage remains close to
present a problem for many consumers. For this reason, this nominal. DC/DC converters are used to interface the battery
and SMES-emulator to the DC bus. An auxiliary supply (Aux.
study considers the use of SMES/battery hybrid energy
Supply) is used to support the DC bus during standby
storage to compensate long and short-term voltage
operation and charge the energy storage devices. The auxiliary
fluctuations. Woong et al. [18] have also considered a supply is disconnected and the energy storage devices provide
SMES/battery hybrid and shown it is viable for smoothing of the necessary power for the inverter to support the load during
renewable energy generator output power and can result in a voltage error.
reduced energy storage system capacity and prolonged battery
life. Li et al. [19] have shown that a SMES/battery energy A. DVR Control
storage system can improve battery lifetime in electric buses. The objective of the DVR control system is to minimise
Deng et.al. [20] have presented a SMES/battery hybrid system supply voltage variations at the load terminals. This is
for reducing peak grid power in an electric vehicle charging achieved by generating a compensating voltage at the series
station. Nie et al. have also presented a SMES/battery hybrid injection transformer terminals. The phasor diagram in Fig.
system and shown its feasibility in dealing with long term and 2(a) shows various DVR voltage control techniques. ‘In phase
short term charge/discharge events in a wave energy compensation’ causes the compensating voltage to be in phase
conversion system [21]. This study extends previous with the incoming supply voltage and has been shown to result
simulation-based SMES/battery hybrid system studies [18-21] in the lowest DVR power rating [22]. ‘Pre-sag compensation’
Figure
preserves theX1. Measured
phase results: (a)supply
of the incoming Supply
at voltages
the time (b)
a sag
by considering the hardware implementation of a Loadwhich
voltage
occurs canred
bephase (c) Load
beneficial voltage redloads
in protecting phasethat
(d) are
SMES/battery energy storage system. The design is shown to Load voltage blue phase.
sensitive to phase disturbances. ‘Energy optimal’ control is
be capable of interfacing SMES and battery energy storage
used to minimise DVR energy storage requirement by
systems and controlling their power sharing to support a three injecting a voltage in quadrature to the load current. ‘In phase
phase load, during both long-term and short-term voltage sags. compensation’ and ‘Pre-sag compensation’ have been
This has benefits in terms of improved long-term voltage considered in this study. The control scheme was
support capability and reduced costs compared with a purely implemented in the synchronous reference frame as shown in
SMES-based system. Additional benefits include reduced Fig. 2(b) by converting three phase AC quantities to
battery power rating requirement and an improvement in equivalent two phase quantities:
expected battery life compared with a battery-only system.
V sa 
Vs  2 1  1 / 2  1 / 2   
    3  
 V sb (1)
Vs  3 0 3
 2 2  V sc 

0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This articleDownloaded
has been accepted for http://iranpaper.ir
from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 3

Figure 3. Simulink implementation of PLL algorithm where ω0 is the

fundamental output frequency in rad/s.

Vsd   cos pll sin pll  Vs 

    (2)
Vsq   sin pll cos pll  Vs 

where Vsa,b,c, Vsa,b, Vsd,q are the supply voltages and θpll is the
estimated supply phase angle.
A phase-locked loop (PLL) was used to determine the phase
angle θpll of the incoming supply based on an algorithm which
is robust in the presence of harmonics, non-symmetry and
transients [23]. Fig. 3 shows a Simulink implementation of the
PLL algorithm. This algorithm minimises the sine of the phase
error term causing the control system to be non-linear. For this
reason, the PI controller gains were tuned empirically. The
PLL algorithm requires the cosine and sine of the incoming
supply angle as inputs which can be obtained geometrically
using the orthogonal reference frame voltages from (1) as:
Vs
cos( )  Figure 4. Simulated PLL Algorithm results: (a) Simulated voltage sag
(3)
Vs 2  Vs 2 with phase jump (b) Phase jump angle (c) Blue trace: supply phase
angle. Red trace: PLL output: ‘Pre-sag compensation’ with controller
gains: kp = 0.5, ki = 5, (d) Blue trace: supply phase angle. Red trace: PLL
Vs output: ‘In phase compensation’ with controller gains kp = 200, ki = 50.
sin( )  (4)
Vs 2  Vs 2 three phase reference frame:
The PLL controller can be tuned to preserve the phase of Vref  cos pll  sin pll  Vrefd 
    (7)
the incoming supply before a sag event with phase jump or, Vref   sin pll cos pll  Vrefq 
alternatively, the PLL can be made to track the phase of the
incoming supply during a sag with phase-jump. Consequently,
Vrefa  2 3 0 
by changing the PLL gains the system can be controlled to    Vref 
3
provide ‘in phase’ or ‘pre-sag’ compensation. Fig. 4 illustrates Vrefb   k  1 3 1 3   (8)
the results of tuning the PI controller in this way.   2
  Vref 
Vrefc   1 3  1 3 
To detect the presence of a voltage error, the following
Figurewas
inequality X1.used
Measured results: (a) Supply voltages (b)
[24], [25]: The injection voltages, Vref_a,b,c, were multiplied by a feed-
Load voltage red phase (c) Load voltage red phase (d) forward constant, k to compensate for losses within the power
V *d blue
Load voltage 2  V *q Vs,q 2  Vthreshold
 Vs,dphase. (5) stage.
Sine-wave pulse width modulation (SPWM) or space vector
where Vs,d,q is the measured load voltage and V*d,q is the
modulation (SVM) were considered for generating the inverter
desired nominal voltage in the synchronous reference frame.
output voltage. SVM is advantageous due to better utilisation
Inequality (5) was also used to trigger the disconnection of the
of the DC bus voltage and which allows deeper sag
DC bus auxilliary power supply (see Fig. 1).
compensation. However, SPWM allows the possibility to
The compensation voltage Vrefd,q is determined, based on the
mitigiate unbalanced faults so this technique was implemented
error between the desired nominal voltage and the supply
in this study. The inverter control was implemented using a
voltage:
Texas Instruments F28069 32-bit micro-controller by
Vrefd ,q  V *d ,q  Vd ,q (6) discretisation of the control and PLL algorithms. The inverter
system parameters are listed in the Appendix, Table AI.
The PWM phase reference voltages Vrefa,b,c were generated by
transforming the required compensation voltage to the rotating

0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This articleDownloaded
has been accepted for http://iranpaper.ir
from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 4

Figure 5. Energy storage systems active current droop control. Figure 6. SMES DC/DC converter.

B. Energy Storage top-Level Control I csmes  I smes (1  D2 ) (11)

The objective of the top-level energy storage control For active current droop control according to (9), the desired
strategy was to control the charge/discharge of each energy converter output current Icsmes is given by (12).
Vbus  Vnom_ smes k smes
storage device. A current vs. voltage active droop
characteristic was chosen as this strategy has been shown to I csmes  (12)
provide good stability and active power sharing [26]. The where ksmes is the gradient of the droop controller and Vnom_smes
converter reference currents are based solely on the level of is the nominal voltage of the droop controller.
the DC bus voltage which is advantageous as high-bandwidth The required duty ratio can be determined by substituting
communication between the three different converters is not Eq. (10) during charge or (11) during discharge into (12) and
necessary. The charge or discharge priority of an energy solving for duty ratio in real time. This allows the SMES-
storage device can be adjusted raising or lowering its nominal inductor to be charged or discharged by simply raising or
DC bus voltage using a technique known as “DC bus lowering the DC bus voltage relative to the nominal Vnom_smes.
signaling” [27]. The system was configured to prioritise the The controller was implemented using a 16-bit microcontroller
SMES-emulator to charge/discharge before the battery. By (PIC24HJ128GP502) with a switching frequency of 500Hz.
always prioritising the short term energy storage, battery This low switching frequency was chosen to allow for extreme
power cycling is reduced which can improve battery lifetime (>>90% and <<10%) duty ratio operation without causing
[19]. overly narrow gate pulses. When operating at 100A nominal
The current droop characteristic for each device is shown in current, conduction losses could be reduced significantly by
Fig. 5, and is made up of three regions of operation. When the the use of low on-state resistance MOSFETs as opposed to
DC voltage is above voltage level Vh(x) (where x refers to IGBTs in this system. This is expected not to be the case for
energy storage system 1 or 2) or below below Vl(x),, the larger systems, operating at significantly higher nominal
converter current is limited to Imax(x) or -Imax(x). In between Vh(x) currents and voltage ratings.
and Vl(x), current is controlled based on the linear current vs.
voltage relationship: D. Battery System
Ix  Vnom  Vbus  k x (9)
A bidirectional synchronous-buck converter rated at 40A
output current with hysteresis current control was used to
where kx is a droop coefficient (A/V) and Ix is the energy control battery current. This topology has been previously
storage converter reference current. reported for use with interfacing an ultra-capacitor energy
storage system to the DC bus in DVR application [13].
C. SMES-Emulator System
However, the proposed system differs from previous studies
To reduce costs and be able to evaluate a SMES-battery [13] in that a variable-frequency hysteresis current control has
hybrid DVR control platform, the SMES device was emulated been used. This is advantageous as it features cycle-by-cycle
by using a 15mH iron-core inductor in this study. The SMES current limiting, making it tolerant to short circuit faults. Also,
converter was based on the asymmetric H-bridge the proposed technique is shown below to be globally stable
configuration shown in Fig. 6. This converter was rated at up over the operating range whereas typical current mode control
to 220A continuous current using forced air cooling. During techniques described previously [13] require slope
charge, Q2 is held ON and Q1 is modulated whereas during compensation to ensure global stability [28]. Further
discharge Q1 is held off and Q2 is modulated. The advantages include good dynamic current-tracking capability,
relationship between DC bus current, inductor current and and robust performance despite variation and uncertainty in
duty ratio, is given by (12) and (13) for charge and discharge, operating conditions [29].
respectively [16]. Assuming lossless, ideal components, the inductor current
I csmes  I smes D1 (10) in Fig. 7 is given
Figure by (13). results: (a) Supply voltages (b)
X1. Measured
Figure X1. Measured results: (a) Supply voltages (b) Load voltage red phase (c) Load voltage red phase (d)
Load voltage red phase (c) Load voltage red phase (d) Load voltage blue phase.
Load voltage blue phase.
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This articleDownloaded
has been accepted for http://iranpaper.ir
from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 5

To achieve active current droop control, the inductor current

reference I* is set as follows:
I*  Vbus  Vnom_ bat kbat VVbus (19)
bat

E. Energy Storage System Sizing

DC bus signaling has been used to control the
charge/discharge of the SMES-emulator system during short-
term voltage voltage-sags whereas the battery is used to
mitigate longer-term voltage variations. Consequently, the
sizing of the short-term and long-term energy storage systems
can be treated separately. The inductive energy storage
requirement is sized to mitigate short-term voltage sags
whereas the battery system is sized to mitigate long-term
under-voltages. The specific sag-depth and duration can be
Figure 7. Battery DC/DC converter.
determined by measurement [31] or by simulation. Once the
load power, depth of sag and sag duration are known, the
SMES energy-storage requirement can be determined as has
dI L
L  Vbusu  Vbatu (13) been previously reported [16]. The technique below extends
dt this method [16] by including the inverter losses in the
where u is a gate drive signal and Vbat and Vbus are battery and analysis.
DC bus voltages. While the DVR system is supporting the load voltage at its
Sliding-mode control theory may be used to describe the nominal value, the real power provided by the DVR to the
current regulation strategy [30]. The objective in this case is to load can be written:
regulate the inductor IL, such that it tracks the command  
3  Vd _ nom  Vd _ sag I d _ load 
Psag 
reference current I*. A function σ = 0 can be defined as
follows [30]:

  
2  Vq _ nom  Vq _ sag I q _ load 
(20)

  IL  I * (14) Assuming a sinusoidal current waveform, the losses in each

IGBT and diode in the inverter, can be estimated based on
A switching control law which satisfies (15) such that σ and previously reported loss models [32], [33]. Conduction loss in
its time derivative have opposite signs ensures the system will each IGBT is given by:
converge to the state σ = 0 [30] and consequently IL = I*.
Pci 
  0 :   0 (15) 1 M   1 M  (21)
  Cos  Rce I cm 2    Cos Vce0 I cm
 8 3   2 8 
A switching control law which satisfies (15) is defined by
(16), where h is a small constant hysteresis band lying where M = modulation index, Rce = IGBT on-state resistance,
symmetrically about the set-point value. Vce0 = IGBT forward voltage drop. Icm = IGBT forward current
and φ = power factor angle.
u  1, if   h
u  (16) Conduction loss in each parallel diode is given by:
 u  0, if   h 
Pcd 
If Vbus is greater than Vbat, then condition (15) is met, so this 1 M   1 M  (22)
  Cos  R f I cm 2    Cos V f 0 I cm
becomes an operational requirement.  8 3   2 8 
For active current droop control according to (9), the
following relationship can be written: where Rf = diode on-state resistance, Vf0 = IGBT forward

Vbus  Vnom_ bat kbat

voltage drop. Icm = diode current. The equivalent on-state
I cbat  (17) resistance of the IGBT (Rce) and diode (Rf) were determined
by linear interpolation of datasheet values [34]. M, φ, and Icm
where Icbat is the battery DC/DC converter reference current
can be found from the dq-axis components of the load current
and kbat is the gradient of the battery converter droop and required compensation voltage.
characteristic and Vnom_bat is the nominal voltage of the battery Switching loss in the IGBT and parallel diode can be
converter droop controller. determined as follows [32].
The average current delivered to the DC bus can then be
estimated, based on an ideal conversion ratio as follows: Pswi 
1

E on ( I cm )  E off ( I cm )
V
Vdc
(23)
dc _ ref
V
I cbatt  I * bat (18)
Vbus

Figure X1. Measured results: (a) Supply voltages (b)

0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Load voltage red phase (c) Load voltage red phase (d)
This articleDownloaded
has been accepted for http://iranpaper.ir
from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 6

Figure 8. Experimental Set-up.

Figure 9. Laboratory Setup.

1 Vdc maintained at 160Vdc during standby operation using the

Pswd  E rr ( I cm ) (24)
 Vdc _ ref auxiliary supply.
The short-term energy storage system capacity was
where Vdc is the commutation voltage, Vdc_ref is the reference estimated to support a sag to 35% of nominal voltage lasting
voltage at which the switching loss energies are given, approximately 100ms using (26). The battery bank was sized
Eon(Icm), Eoff(Icm) and ERR(Icm) are IGBT turn-on, turn-off and based on available lead acid batteries. Energy storage device
diode reverse-recovery energies as a function of forward parameters are given in the Appendix, Tables AII and AIII.
current [34]. The total inverter losses can then be estimated as: The experimental set-up is shown in Figs. 8 and 9.
Pinv _ loss  6Pci  Pcd  Pswi  Pswd  (25)
IV. RESULTS
If a voltage disturbance has duration Tsag, then the capacity
Initially a three-phase voltage sag to 35% of nominal volt-
requirement of the energy storage systems connected to the
age, lasting 100ms was used to demonstrate the response of
DC bus can be found from (20) and (25) as follows.
the DVR and energy storage systems. From Fig. 10 it can be

E sag  Psag Pinv _ loss Tsag  (26) seen that the hybrid DVR system mitigates the voltage sag
effectively during the sag event. The battery is discharged
For a given long-term under-voltage, the required discharge momentarily at -1.45A at the end of the sag when the inductor
power provided by the DVR can be found from (20). The energy has been depleted.
DVR energy can then be determined using the procedure The SMES-emulator was then removed from the system and
shown in (21) – (26). the test was repeated. The system response to the same three-
F. Experimental Setup phase sag with only battery energy storage is shown in Fig. 11.
In order to verify the performance of the proposed DVR, a As the DC bus voltage falls below 140Vdc (the battery system
hardware prototype was constructed. The battery was a 48V, nominal voltage) the battery is discharged to support the DC
75Ah sealed lead acid type and the SMES-emulator was bus. The peak battery current is 21.13A in this case and the
15mH/100A iron-core inductor. The nominal AC bus voltage DVR system can be seen to effectively mitigate the voltage
was set to 120Vac using a step down transformer. The load sag. Fig. 12 shows an oscilloscope trace of the DVR injection
was configured as a 1.4kW star-connected resistor. voltage and supply voltage during a three phase sag.
A simple laboratory sag generator was developed using To demonstrate the system performance during a long-term
solid-state relays to simulate faults to ground as shown in Fig. under-voltage, the experimental system was configured for a
8. System voltage measurements using isolating amplifiers three-phase under-voltage dropping to 80% of nominal lasting
were logged together with the battery and SMES-inductor 2s. The results are shown in Fig. 13. Initially, the SMES
currents using hall-effect current probes. The lead-acid battery system can be seen to support the DC bus. As the voltage error
was maintained at a full state of charge at the beginning of lasts longer than the SMES-emulator can support the bus
each experiment. To prevent battery overcharge, the maximum voltage, at time = 0.4s the DC bus voltage falls to below
battery chargeX1.
current was limited 140Vdc and the battery energy storage system automatically
Figure Measured results:to(a)
a low valuevoltages
Supply of ~800mA
(b)
during discharges.
Figure X1. Measured results: (a) Supply voltages (b)
Load voltage red phase (c) Load voltage red phase was
stand-by operation. The DC bus voltage (d)
Load voltage blue phase. Load voltage red phase (c) Load voltage red phase (d)
Load voltage blue phase.

0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This articleDownloaded
has been accepted for http://iranpaper.ir
from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7

Figure 11. Battery System Experimental results: 0.1s Three phase sag to
35% of nominal voltage. (a) Supply voltages (b) Load voltages (c) DC
Link Voltage (d) Battery Current.

nominal was reduced to 1.45A from 21.13A compared with

the battery-only system. This reduction in battery peak current
can have significant benefits in terms of improved battery life,
Figure 10. Hybrid System Experimental results: 0.1s Three phase sag to
35% of nominal voltage. (a) Supply voltages (b) Load voltages (c) DC which is negatively affected by discharge current magnitude
Link Voltage (d) Battery Current (e) SMES-inductor current. [35].
From Fig. 13, it can be seen that the long-term under-
The battery then provides the necessary power to support voltage performance of the system has been improved by the
the load for the remainder of the under-voltage condition. At introduction of battery energy storage. One of the main
Time = 2.1s the under-voltage event has finished and the DC shortcomings limiting the usage of SMES systems is the high
bus auxiliary supply reconnects. The battery and SMES- cost of the superconducting material [36]. By integrating a
emulator energy storage systems are then re-charged. While 2s battery, the hybrid DVR can be of significantly reduced cost
may not be considered a long-term under-voltage, this compared with a SMES only DVR and still be capable of
duration was chosen in order to be able to see the interaction compensating a variety of short-term sags and long-term
between the energy storage systems clearly. The system under-voltages. In order to estimate the potential scale of the
response reaches a stable operating point after approximately cost reduction, the prototype system used in this study has
0.6s which could be maintained to support the load until the been considered. The SMES-inductor was emulated to give L=
battery bank is depleted. 15mH at 100A DC. The energy capacity of an equivalent
SMES can be calculated:
V. DISCUSSION
LI 2
By use of the SMES-emulator hybrid system, the peak E HESS _ SMES   75 ( J ) (27)
battery current response during a three-phase sag to 35% of 2

0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This articleDownloaded
has been accepted for http://iranpaper.ir
from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 8

Figure 12. Hybrid System Experimental results: 0.14s Three phase sag to
35% of nominal voltage. Cyan trace: DVR phase-a injection voltage
(100V/div). Yellow trace: Phase-a supply voltage (100V/div.). Injection
transformer boost ratio: 2:1.

TABLE I
SMES PARAMETERS
Inner SC SC
Energy Turns
Case Coils Radius Length Cost
(J) /coil
(mm) (m) ($40/m)
Hybrid
75 1 95 40 30 $1200
System
SMES-
638 6 30 55 95 $3800
(2s)
SMES-
(2 38251 6 200 140 200 $19000
minute)

Without battery storage, the SMES needs to be able to

compensate long-term undervoltages. The nominal supply dq- Figure 13. Hybrid System Experimental results: Long-term three phase
axis voltage has no quadrature component, so the direct axis undervoltage (a) RMS supply phase-voltage. (b) RMS load phase-
voltage Vd_nom is the peak voltage in the experimental system, voltage (c) DC Bus Voltage (d) Battery Current (e) SMES-inductor
current.
169.7V. During a sag with no phase jump, the supply voltage
falls to Vd_sag which is 80% of Vd_nom or 135.8V. Two cases of
sag duration have been considered; a sag lasting two seconds VI. CONCLUSIONS
and a sag lasting two minutes. The energy required by the
DVR inverter during the two second sag was found to be 638J The performance a novel hybrid DVR system topology has
from (20)-(26), whereas the energy required to support the two been assessed experimentally and shown to effectively provide
minute sag was found to be 38251J based on IGBT module voltage compensation for short-term sags and long-term
loss parameters given in [37]. The design procedure described under-voltages. A prototype system has been developed which
in [38] was then used to estimate the superconducting material demonstrates an effective method of interfacing SMES and
requirement assuming a working temperature of 50K. The coil battery energy storage systems to support a three phase load.
design for the different cases and overall superconducting The system has been shown to autonomously prioritise the use
material cost based on the price of 2G HTS (high temperature of the short-term energy storage system to support the load
superconductor) of $400/kA-m [39] is given in Table 1. A during deep, short-term voltage sags and a battery for lower
complete techno-economic analysis of the benefits of the depth, long-term under-voltages. This can have benefits in
hybrid system is beyond the scope of this work but the terms of improved voltage support capability and reduced
simplified calculation above shows that a significant reduction costs compared with a SMES-based system. Additional
in superconducting material cost can be achieved by benefits include reduced battery power rating requirement and
hybridisation with a battery. This would be offset by the an expected improvement in battery life compared with a
additional cost the battery system but worsened by the cost of battery-only system due to reduced battery power cycling and
SMES refrigeration equipment which are not included in the peak discharge power.
analysis.

0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This articleDownloaded
has been accepted for http://iranpaper.ir
from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 9

VII. APPENDIX [5] B. Otomega and T. Van Cutsem, "Undervoltage Load Shedding
Using Distributed Controllers," Power Systems, IEEE Trans., vol.
TABLE AI 22, Nov. 2007, pp. 1898-1907.
INVERTER SYSTEM PARAMETERS [6] L. Chun, J. Savulak, and R. Reinmuller, "Unintentional Islanding
Switching frequency 8kHz of Distributed Generation: Operating Experiences From Naturally
Injection transformer rating 6kVA Occurred Events," Power Delivery, IEEE Trans., vol. 29, Jan.
Injection transformer boost ratio 2:1 2014, pp. 269-274.
IGBT Modules [34] 2MBi100TA-060-50 [7] S. Tohidi, H. Oraee, M.R. Zolghadri, S. Shiyi, and P. Tavner,
Filter inductance 3mH "Analysis and Enhancement of Low-Voltage Ride-Through
DC Bus capacitor 2 x 4700uF in series. Capability of Brushless Doubly Fed Induction Generator,"
PLL PI gains Kp = 200 Ki = 50 Industrial Electronics, IEEE Trans., vol. 60, March 2013, pp.
Load nominal voltage 120Vac 1146-1155.
Aux. Supply LAB-SMS 5300 [8] R. Lawrence, "Voltage optimization," Industry Apps. Mag., IEEE,
Sag detect threshold: Vthreshold 0.1 vol. 12, Sept. 2006, pp. 28-33.
[9] P. Kanjiya, B. Singh, A. Chandra, and K. Al-Haddad, ""SRF
TABLE AII Theory Revisited" to Control Self-Supported Dynamic Voltage
SMES-EMULATOR SYSTEM PARAMETERS Restorer (DVR) for Unbalanced and Nonlinear Loads," Industry
Switching frequency 500Hz Applications, IEEE Trans., vol. 49, May 2013, pp. 2330-2340.
MOSFET (Q1, Q2) 2 in parallel in each leg. IXFN140N30P [10] A.M. Rauf and V. Khadkikar, "An Enhanced Voltage Sag
Diode (D1, D2) 2 in parallel in each leg. DSEI2X121-02A Compensation Scheme for Dynamic Voltage Restorer," Industrial
Inductor LSMES 15mH Electronics, IEEE Trans., vol. 62, Oct. 2015, pp. 2683-2692.
DC Bus capacitor 1.5mF [11] V.K. Ramachandaramurthy, C. Fitzer, A. Arulampalam, C. Zhan,
Droop gain 0.65 A/V M. Barnes, and N. Jenkins, "Control of a battery supported
CR Snubber 3.3Ohm + 100nF dynamic voltage restorer," Generation, Transmission and
Distribution, IEE Proceedings., vol. 149, Sept. 2002, pp. 533-542.
[12] J.G. Nielsen and F. Blaabjerg, "A detailed comparison of system
TABLE AIII
topologies for dynamic voltage restorers," Industry Applications,
BATTERY SYSTEM PARAMETERS
IEEE Trans., vol. 41, Sept. 2005, pp. 1272-1280.
Battery voltage 48V (4x12V) [13] D. Somayajula and M.L. Crow, "An Integrated Dynamic Voltage
Battery type Sealed lead acid 75Ah Restorer-Ultracapacitor Design for Improving Power Quality of
IGBT Module MG1502YS50 the Distribution Grid," Sustainable Energy, IEEE Trans., vol. 6,
Power Inductor 2 x 1.8mH in parallel April 2015, pp. 616-624.
Low side capacitor 4700µF [14] W. Bingsen and G. Venkataramanan, "Dynamic Voltage Restorer
DC Bus capacitor 6600µF Utilizing a Matrix Converter and Flywheel Energy Storage,"
Droop gain 0.45 A/V Industry Applications, IEEE Trans., vol. 45, Jan. 2009, pp. 222-
Maximum battery charge current 800mA 231.
Maximum battery discharge current 32A [15] H.J. Kim, K.C. Seong, J.W. Cho, J.H. Bae, K.D. Sim, S. Kim, E.Y.
Lee, K. Ryu, and S.H. Kim, "3 MJ/750 kVA SMES System for
Improving Power Quality," Applied Superconductivity, IEEE
TABLE AIV Trans., vol. 16, June 2006, pp. 574-577.
EXPERIMENTAL SYSTEM PARAMETERS [16] S. Jing, T. Yuejin, Y. Kai, C. Lei, R. Li, L. Jingdong, and C. Shijie,
Load 1.4kW resistive "SMES Based Dynamic Voltage Restorer for Voltage Fluctuations
Line impedance 5 , resistive Compensation," Applied Superconductivity, IEEE Trans., vol. 20,
Step-down transformer ratio 2:1 May 2010, pp. 1360-1364.
Three phase sag fault impedance 3  resistive [17] H.M.S.C. Herath, V.J. Gosbell, and S. Perera, "Power quality (PQ)
Single phase under-voltage fault impedance 25  resistive survey reporting: discrete disturbance limits," Power Delivery,
ADC card NI USB-6463 IEEE Trans., vol. 20, Sept. 2005, pp. 851-858.
Sample rate 10kHz [18] S. Jae Woong, C. Youngho, K. Seog-Joo, M. Sang Won, and H.
Kyeon, "Synergistic Control of SMES and Battery Energy Storage
for Enabling Dispatchability of Renewable Energy Sources,"
VIII. ACKNOWLEDGMENT Applied Superconductivity, IEEE Trans., vol. 23, June 2013, pp.
The author would like to thank financial support from 5701205-5701205.
[19] J. Li, M. Zhang, Q. Yang, Z. Zhang, and W. Yuan, "SMES/Battery
EPSRC EP/K01496X/1 and support from the Royal Academy Hybrid Energy Storage System for Electric Buses," IEEE
of Engineering's Research Fellowship programme. Transactions on Applied Superconductivity, vol. 26, Feb. 2016, pp.
1-5.
[20] J. Deng, J. Shi, Y. Liu, and Y. Tang, "Application of a hybrid
IX. REFERENCES energy storage system in the fast charging station of electric
vehicles," IET Generation, Transmission & Distribution, vol. 10,
[1] P.K. Ray, S.R. Mohanty, N. Kishor, and J.P.S. Catalao, "Optimal March 2016, pp. 1092-1097.
Feature and Decision Tree-Based Classification of Power Quality [21] N. Zanxiang, X. Xi, K. Qing, R. Aggarwal, Z. Huiming, and Y.
Disturbances in Distributed Generation Systems," Sustainable Weijia, "SMES-Battery Energy Storage System for Conditioning
Energy, IEEE Trans., vol. 5, Sept. 2014, pp. 200-208. Outputs From Direct Drive Linear Wave Energy Converters,"
[2] D. Novosel, G. Bartok, G. Henneberg, P. Mysore, D. Tziouvaras, Applied Superconductivity, IEEE Trans., vol. 23, June 2013, pp.
and S. Ward, "IEEE PSRC Report on Performance of Relaying 5000705-5000705.
During Wide-Area Stressed Conditions," Power Delivery, IEEE [22] P. Jayaprakash, B. Singh, D.P. Kothari, A. Chandra, and K. Al-
Trans., vol. 25, Jan. 2010, pp. 3-16. Haddad, "Control of Reduced-Rating Dynamic Voltage Restorer
[3] "IEEE Recommended Practice for Monitoring Electric Power With a Battery Energy Storage System," Industry Applications,
Quality," in IEEE Std 1159-1995, ed. New York, NY: IEEE IEEE Trans., vol. 50, Jul. 2014, pp. 1295-1303.
Standards Board, 1995, p. i. [23] J.G. Nielsen, F. Blaabjerg, and N. Mohan, "Control strategies for
[4] S. Jothibasu and M.K. Mishra, "A Control Scheme for Storageless dynamic voltage restorer compensating voltage sags with phase
DVR Based on Characterization of Voltage Sags," Power jump," in Applied Power Electronics Conference and Exposition,
Delivery, IEEE Trans., vol. 29, July 2014, pp. 2261-2269. 2001. Sixteenth Annual IEEE, pp. 1267-1273 vol.2.
[24] J.G. Nielsen, M. Newman, H. Nielsen, and F. Blaabjerg, "Control
and testing of a dynamic voltage restorer (DVR) at medium

0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This articleDownloaded
has been accepted for http://iranpaper.ir
from publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2016.2609403, IEEE
http://www.itrans24.com/landing1.html
Transactions on Energy Conversion
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 10

voltage level," Power Electronics, IEEE Trans., vol. 19, May Anthony Gee received his BSc degree in
2004, pp. 806-813. Cybernetics and Control Engineering from the
[25] C. Fitzer, M. Barnes, and P. Green, "Voltage sag detection University of Reading, England. He then completed
technique for a dynamic voltage restorer," Industry Applications, a PhD in supercapacitor/battery hybrid energy
IEEE Trans., vol. 40, Feb. 2004, pp. 203-212. storage systems at the University of Bath, England.
[26] C. Gavriluta, J.I. Candela, C. Citro, J. Rocabert, A. Luna, and P. Since then he has undertaken projects with
Rodriguez, "Decentralized Primary Control of MTDC Networks Williams Advanced Engineering, University of
With Energy Storage and Distributed Generation," Industry Bristol Energy Management Group and is currently
Applications, IEEE Trans., vol. 50, April 2014, pp. 4122-4131. working as a power electronics hardware design
[27] J. Schonberger, R. Duke, and S.D. Round, "DC-Bus Signaling: A engineer. His research interests are in power
Distributed Control Strategy for a Hybrid Renewable Nanogrid," electronic converters, motor drives and renewable energy systems.
Industrial Electronics, IEEE Transactions on, vol. 53, Oct. 2006,
pp. 1453-1460.
[28] S. Chattopadhyay and S. Das, "A Digital Current-Mode Control Francis Robinson (M’1988) gained a PhD from
Technique for DC-DC Converters," Power Electronics, IEEE Heriot-Watt University, Edinburgh, Scotland. He
Trans., vol. 21, Nov. 2006, pp. 1718-1726. has been a lecturer at the Department of Electronic
[29] M. Castilla, J.M. Guerrero, J. Matas, J. Miret, and J. Sosa, and Electrical Engineering, University of Bath since
"Comparative study of hysteretic controllers for single-phase 1990, and is a member of the Electromagnetic,
voltage regulators," IET Power Electron., vol. 1, Mar. 2008, pp. Machines and Drives Group. His teaching and
132-143. research are primarily related to power electronics
[30] R. Venkataramanan, A. Sabanovic, and S. Cuk, "Sliding mode and drives. He has previously worked as a power
control of DC-to-DC converters," in IEEE Conference On electronics design engineer with Posidata (Dana)
Industrial Electronics, San Francisco, CA, USA, 1985, pp. 251– LTD, Rutherford Appleton Laboratory, and GEC Small Machines Ltd. Dr
258. Robinson is a chartered engineer in the UK and a member of the IET and
[31] J. Horak, "Power Quality: Measurements of Sags and IEEE.
Interruptions," in IEEE/PES Transmission and Distribution
Conference and Exhibition, 2005/2006, Dallas, TX, pp. 733-739. Weijia Yuan received his Bachelor degree from
[32] M.H. Bierhoff and F.W. Fuchs, "Semiconductor losses in voltage Tsinghua University and his PhD from the
source and current source IGBT converters based on analytical University of Cambridge. He then became both a
derivation," in Power Electronics Specialists Conference, 2004 research associate in the Engineering Department
IEEE 35th Annual, Aachen, Germany, pp. 2836-2842 Vol.4. and a junior research fellow at Wolfson College,
[33] F. Casanellas, "Losses in PWM inverters using IGBTs," Electric both at the University of Cambridge. Dr Yuan
Power Applications, IEE Proceedings, vol. 141, Sep 1994, pp. joined the University of Bath as a lecturer (US
235-239. equivalent: Assistant Professor) in 2011. Dr Yuan’s
[34] 2MBI150U2A-060-50 Datasheet. Fuji Electric Ltd., Edison, NJ, research interests focus on electrical power
Available: applications of superconductivity including fault
http://www.americas.fujielectric.com/sites/default/files/2MBI150U current limiters, machines and power transmission
2A-060.pdf lines. He has extensive research experience in modelling of superconducting
[35] J.H. Yan, W.S. Li, and Q.Y. Zhan, "Failure mechanism of valve- devices and building superconducting magnets.
regulated lead–acid batteries under high-power cycling," J. Power
Sources, vol. 133, May 2004, pp. 135-140.
[36] X.Y. Chen, J.X. Jin, Y. Xin, S. Bin, C.L. Tang, Y.P. Zhu, and
R.M. Sun, "Integrated SMES Technology for Modern Power
System and Future Smart Grid," Applied Superconductivity, IEEE
Trans., vol. 24, Oct. 2014, pp. 1-5.
[37] FF600R06ME3 Datasheet. Infineon Technologies AG,
Neubiberg, Germany, Available:
http://www.infineon.com/dgdl/Infineon-FF600R06ME3-DS-
v03_01-en_de.pdf?fileId=db3a304318a6cd680118e5130f161311
[38] W. Yuan, W. Xian, M. Ainslie, Z. Hong, Y. Yan, R. Pei, Y. Jiang,
and T.A. Coombs, "Design and Test of a Superconducting
Magnetic Energy Storage (SMES) Coil," Applied
Superconductivity, IEEE Trans., vol. 20, May 2010, pp. 1379-
1382.
[39] V. Selvamanickam and J. Dackow, "Progress in SuperPower’s 2G
HTS Wire Development and Manufacturing," in DOE Advanced
Cables & Conductors Peer Review, , Alexandria, VA, 2010.

0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.