New Paradigm of Power System Operation and Control

New Paradigm of Power System Operation and Control

Professor Janusz Bialek

Principal Research Fellow
Imperial College London
 Power systems
Power
Motivation and acknowledgments electronics

• Motivation for the talk: understanding Inverter-Based Resources (IBRs)

• IBRs require interdisciplinary effort
• Most of the literature is from power electronics or control point of view
• My background is power systems
Economics Control
• Some knowledge of control but none of power electronics
• This presentation is the result of long discussions with Tim Green,
Florian Dörfler and others
• Nicking some of their diagrams
• Look at the IBRs from the power system point view
• John F. Kennedy: Ask not what your country can do for you, ask what
you can do for your country
• JB: Ask not what you can do for IBRs, ask what IBRs can do for you
(power systems)
Outline

• Setting the scene for the Summer School

• High-level overview, from the power system point of view, of the challenges posed by high penetration of
Inverter-Based Resources (IBRs)
• (almost) no equations
• Details of the challenges treated by other presenters
• Understanding Grid Following and Grid Forming Inverters
• Frequency control in IBR-based systems
• New services needed
• Conclusions
 1st revolution, end of 19th Century: The Current War DC vs AC

• Tesla won with AC

• Thermal/hydro/nuclear plants used Synchronous Machines (SMs) to
convert mechanical energy into AC electricity
• Technical characteristics and controls of the grid determined by the
physics of SMs

Source: D. Gross
 2nd revolution happening now: zero-carbon generation

2030

Mature, slowly-evolving, risk-averse industry must quickly change to manage

considerable risks

Fast pace of change: System Operators are scared

 Inverter-Based Resources (IBRs)

• Wind, solar PV and batteries produce DC output so they need inverters (converters) to convert DC to AC
and connect to the AC grid
• IBRs also include EVs, HVDC terminals, variable-speed motor drives, statcom, etc as they need inverters
to connect to the AC grid
• Replacing SMs by IBRs changes fundamental technical characteristics of the power system and requires
new controls
 SM: physics-defined behaviour

Source: D. Gross

• Physics dominated by slow electromechanical phenomena (seconds) and

thermodynamic phenomena (minutes)
• SM create 50/60 Hz smooth AC voltage from a rotating electromagnet
• Self-synchronisation and inherent damping of oscillations around the synchronous
speed
• Physical link between power balance and frequency
• Large inertia provides a buffer to absorb generator trips
• Primary frequency response automatically restores power balance
 SM: fault currents, grid strength and modelling

• SM produce large 2-4 pu fault (short-circuit) currents that decay with time
• High fault currents support voltages during a fault (essential for fault-ride- Source: D. Gross
through)
• Easy to detect, useful for protection
• SMs maintain grid “strength”: voltage and frequency independent of real and
reactive power changes at a given location
• Related to fault levels: a high short-circuit current at a given location
indicates that there are many SMs nearby which maintain system
strength
 DC Inverter 3 phase AC

+ Filter
What is an inverter? DC
energy
source Grid
• Converts DC input into 3-phase AC output
-
• Inverter is essentially a programmable switching device
PWM
• Fast switching (kHz) converts DC into a quasi-AC signal:
Pulse-Width Modulation PWM V, ω, θ

• High flexibility and fast actuation (milliseconds) ☺

• Controllable amplitude (voltage V), frequency ω and phase angle θ
• Hence can control
• real power 𝑃 ≅ 𝑉1 𝑉2 sinθ/𝑋
• reactive power 𝑄 ≅ (𝑉12 − 𝑉1 𝑉2 cosθ)/𝑋
• How many degrees of freedom?
3-level 5-level 7-level
• Depends on the controls (GFL, GFM) used to close the
feedback loop
Low-pass filter needed to filter out harmonics
 Inverters are great but… DC Inverter 3 phase AC

+ Filter
DC
energy
• No physics-based behaviour: no inherent self-
source Grid
synchronisation, primary frequency response,
damping of oscillations etc
-
• Everything must be programmed
PWM
• One of the main negative effects: little energy
storage– just DC-side capacitor (“electrical inertia”) 
• Problems for frequency response

Reduced system inertia

in GB 2009-2021 Source: NG ESO
IBRs: fault currents and grid
strength
• Inverters do not provide fault currents
• Cannot be overloaded temporarily due to low
thermal mass of semiconductors
• Deteriorated fault-ride-through
Source: NG ESO
• Protection has to be re-thought
• Is it a red herring?
• PV/wind plants are usually not fully loaded so there is a regional spare capacity to provide fault current
• RTE: zonal analysis, average load of IBRs 30%, max load in a given zone about 60%
• With more wind/solar commissioned to meet net-zero targets, spare capacity will grow
• Reduced system strength
• In SM-based systems it is related to short-circuit levels
• “weak grid” - problems with operation of inverters that require fixed voltage and frequency
• New metrics needed for system strength in IBR-based systems
 Small-signal stability
• IBR controls create new potential modes of
instability: control interactions and weak grid
• Sub-synchronous oscillation (SSO) events
• Exact reasons are often not clear, EMT replication
not easy
• Risks to the security of supply
• Additional costs as SOs have to operate the grid
more conservatively, e.g. 23 stability constraints in
ERCOT
• Uncertainty increases the costs even more

S. Dutta “SSO events and Learnings” PES GM 2023. SSCI- Sub-Synchronous Control Interactions

Sub-synchronous Oscillations (15-20 Hz, 2% pu) in West Murray Zone (WMZ), Australia
AEMO replicated the events using EMT after two years
Source: Australian Energy Market Operator AEMO
 A case study of SSO: Scottish “wobbles”
• 8 Hz spontaneous oscillations in Northern Scotland 2021, 2023, 2024
• 2023: 5 days, tripping of assets (generation, HVDC link, transm. line)
• A mystery: no link to levels of inertia, fault levels, demand, wind levels,
generation mix and system flows
• ESO still not sure about the causes
• Could not replicate or analyse as no white-box models of IBRs available due to proprietary reasons
• Lack of sufficient real-time monitoring (PMUs) in GB
• Operational measures (and costs): holding increased levels of response, reserve, and synchronous generation
 IBR research challenge
• SM-based system: slow but robust as physics-based
• IBR-based system: agile but fragile as control-based
• Inverter challenge: operate the grid efficiently and to the same reliability standard but using IBRs
• Turn a threat of IBRs into an opportunity to remove the shackles imposed by slow and bulky SMs
• Exciting times for research!
• Lack of full understanding how the future IBR-based power system will work
• Rethink power system control
• Multidisciplinary approach needed

Power Power systems

electronics

Economics Control
Acknowledgement: Yunjie Gu
Set up by 6 Founding System Operators and research community in
2019 to address the scale and urgency of the IBR challenge
Goal: ability to operate 100% IBR transmission system by 2025 and
share the learnings
Addresses all the main problems due to IBRs
 Global Centre: Electric Power Innovation in a Carbon Free
Society (EPICS)

• Funded jointly by NSF, UKRI and CSIRO

• Main task: address the technical,
economic and societal aspects of the
inverter challenge
 Comparison of modelling and analysis of IBR-based
and SM-based systems
• System size:
• SM: limited number (<200 for GB) and low order models (<15)

• IBRs: (hundreds of) thousands, each of a high order model (>20)

• Standardisation:
• SM: models of machines are standard and similar regardless of size and manufacturer

• IBRs: Inverters take very many forms with a wide range of design choices

• Model availability
• SM: white-box models available (state-space, block diagrams)
• IBR: manufacturers usually release only black-box models (compiled EMT code)

• Model dependence on the operating point: EMT

• SM: models valid in a wide operating range
• IBR: model switching due to voltage and current limiters
 Comparison of modelling and analysis of IBR-based
and SM-based systems: time-scale separation
• SM: time-scale separation between fast and slow
dynamics simplifies modelling

• Simpler RMS/phasor models for slower

electromechanical dynamics
• Computationally-intensive EMT for faster
electromagnetic dynamics
• IBR: overlapping and interacting dynamics, no time-
scale separation possible
• EMT-type simulation necessary
• Poor scalability as EMT is computation-intensive
 Comparison of modelling and analysis of IBR-based
and SM-based systems – stability analysis

Source: T. Green

• Small-signal stability analysis

• SMs: white-box models enable eigenvalue analysis with participation factors pointing to root-causes of
instabilities

• IBR: running many black-box EMT simulations – infeasible

• Emerging grey-box stability analysis: impedance-based method (T. Green et al)

• No analytical tools for large-signal stability analysis for IBRs

 Understanding Grid Following Inverters (GFL) and Grid Forming
Inverters (GFM)
• A lot of discussion and disagreement in power electronics and control communities about defining GFL and GFM
• GFL: a current source
• GFM: a voltage source behind a (virtual) impedance
• What does it mean from power system perspective?
• Here: high-level explanations about their functionality without going into power electronics or control details
• Fundamental question: which are the main variables that a grid controller can generally control?
• Real power P strongly related to the phase angle θ and frequency Δω = 𝜃ሶ
• Reactive power Q strongly related to voltage V P, Δω = 𝜃ሶ
• Weak link between the two pairs Controller
Q, V
Grid

• 2 degrees of freedom:
• only 1 variable from each pair can be independently controlled
• the other one depends on grid operating conditions
 Grid Following Inverters (GFL) vs Grid Forming Inverters (GFM)
P, ω
• Grid Following Inverter GFL: Grid
GFL
Q, V
• Controllable set-points P and Q
• V and ω set by the grid - GFL “follows” the grid
• GFL is a current source that synchronizes its output with the grid's voltage V and frequency ω to inject or absorb
a required active and reactive power
ω,P
• Grid Forming Inverter GFM: Grid
GFM
V,Q
• Controllable set-points V and ω: a voltage source
• GFM “forms” the grid like SM
• GFM is a voltage source behind a (virtual) impedance which can set voltage and frequency
• (P, Q) are dependent variables – IBR output depends on the grid conditions – cannot be set
• Not a binary divide: a spectrum with many types and flavours, crossovers etc.
 P, ω
GFL Grid
Grid Following Inverter (GFL) Q, V

IBR
• Practically all IBRs now use GFL - a mature
technology DC 3 phase AC
Inverter
• Can contribute to voltage control by controlling Q + Filter
DC
• PWM needs reference signals: V* θ* ω* energy
• Synchronisation: frequency ω* for PWM is set by source Grid
the grid using Phase-Lock-Loop (PLL)
-
• PLL requires a strong grid (fixed voltage and
frequency) PWM V,
• 100% GFL grid is not possible V * θ* ω *
GFL control P, Q
• Yunjie has a different view PLL
• (V* θ*) is provided by the current controller that ensures the desired (Pref, Qref) P, Q, I
Pref
control
• GFL can provide Fast Frequency Response (FFR) Qref

• change quickly output power P in response to frequency changes A conceptual model of GFL
• FFR requires a headroom
 ω,P
GFM Grid
Grid Forming Inverter (GFM) V,Q

• GFM provides a controllable voltage V and frequency ω similarly as SM

IBR
• 100% IBR grid must have some GFMs DC 3 phase AC
Inverter
• Many voltage/frequency control schemes proposed
+ Filter
DC
• Here the simplest: droop control similarly as in SM energy
source Grid
• Droop necessary for parallel operation of GFMs/SMs and
synchronisation -
PWM P, Q
• GFM measures grid (P, Q) and sets (V*, ω* , θ *) for PWM PWM
GFM control
• Similarly as for SM but the other way around V* θ* ω*
Droop
V control
• The output P, Q depends on grid conditions – GFM must have a headroom
A conceptual model of GFM
• Battery a natural energy source as headroom for wind/PV means a revenue loss
• GFM may operate in an islanded grid and provide additional services (blackstart, Droop characteristic
damping of oscillations)
• Just a few experimental GFM installations worldwide – a subject of intensive R&D
 Open research questions about GFM Droop characteristic

• Droop control is the simplest

• GFM can emulate SM (Virtual Synchronous Machine VSM) with controllable
inertia and damping
• Should GFM emulate SM?
• Many other control methods have been proposed (see Florian Dörfler presentation)
• How many GFMs, and where, do we need in a 100% IBR system?
• How to tune multiple control loops of (hundreds of) thousands of IBR controllers?
• Small-signal and large-signal stability of parallel operation of GFMs/GFLs and SMs
• a compliance test (e.g. passivity)?
Benz, 1885
• Etc.

1905, Panhard
 Primary frequency response: SM
• Maintaining the power balance: the most important grid control
• Traditional centre-of-inertia analysis (neglecting transmission network)
• local behaviour may be important when inertia is not distributed uniformly
• Following an infeed (generation or interconnector) loss, the power deficit is ω
initially covered by stored kinetic energy (inertial response) so frequency ω drops
• Turbine governors sense ω and increase output P as per the droop characteristic
• Large inertia M needed to buy time for slow governors (large T) to act
• Traditionally, two important parameters:
• Nadir: avoid inadvertent activation of under-frequency load shedding (48.8 time
Hz in GB) Adapted from F. Dorfler

• Nadir ~ M/T : large T is counterbalanced by large M ☺

• Rate of Change of Frequency ROCOF Droop characteristic
• Fear of catching the frequency decline too late
• Some countries including UK: loss-of-mains DG ROCOF relays
• ROCOF ~ 1/M : ROCOF is small ☺
Primary frequency response: IBR
ω
• Little “electrical inertia” M but much faster than SM
(smaller T)
• Nadir ~ M/T : small M counterbalanced by small T ☺
• ROCOF ~ 1/M : small M means large ROCOF 
time
• A contributing factor for GB 2019 disturbance
• Rapid frequency decline due to a loss of 2 power
stations
Adapted from F. Dorfler
• 350 MW of DG lost on ROCOF, over 1M customers
lost power on load shedding
Counter measures to deal with with ROCOF

• Planning countermeasures
• Installing synchronous condensers
• Increase physical inertia – simple and also helps with fault currents, but costly
• GB: increasing ROCOF settings of loss-of-mains relays (0.125 Hz/s to 1 Hz/s)
• Operation (dispatch): “spilling” wind/solar - replacing wind/PV by SM-based plants to increase inertia
• Additional costs: more expensive generation dispatched + compensations for constrained-off wind/PV

Cost of managing ROCOF in GB

Source: D. Gross
 ω

Do we really need to have large

physical inertia for IBR-dominated
systems? time

Inertia was needed to buy time for slow turbine

governors to act but inverters are quick

Can IBRs deliver “synthetic inertia”?

Synchronous condenser

Source: D. Gross
 ω
Can GFL-based Fast Frequency Response (FFR)
provide “synthetic inertia”?
time
• Inertia is an instantaneously-acting energy buffer: energy is released
proportionally to ROCOF (rather than just frequency deviation) IBR

• IBRs are fast but not instantaneous: DC

Inverter
3 phase AC

• Measurement delays (PLL) as frequency cannot be DC

+ Filter
measured directly: wait for zero-crossing of AC wave energy
source Grid
• Frequency is a noisy signal, measuring ROCOF requires waiting for
-
many periods
PWM V,
• Delays with activating the energy source V* θ* ω*
GFL control P, Q

PLL
• GFL suffers delays 0.1-0.5 s, depending on the energy source P, Q, I
Pref
control
Qref
• Acceptable for FFR to limit the nadir (occurs after about 10 s)
• Not fast enough to limit ROCOF
• GFLs do need inertia of SMs to buffer the delays
• The term “synthetic inertia” with respect to GFLs is misleading
Can GFM provide “synthetic inertia”?

• GFM measures P instead of ω IBR

DC 3 phase AC
Inverter
• P can be measured directly – reduced measurement
+ Filter
delays ☺ DC
energy
• But there are still some small unavoidable delays: source Grid
processing, activating the energy source 
-
PWM
• Smaller delays can be buffered using a large DC-side PWM
P, Q
supercap (“electrical inertia”) GFM control
V* θ* ω*
Droop
• Can 100% IBR system with GFMs operate without inertia? V control

• Possible but large ROCOF

• Maybe in the future when everything is connected via
inverters Frequency response
in a 100% IBR small
island grid

Source: F. Doerfler
 System services

• SM provides many system services by means of

physics so they are free
• inertia, fault currents, damping of oscillations,
etc
• IBRs have to be programmed to deliver those
services and have to be paid for them – additional
costs
• Open research questions: What are the grid needs
and the services needed to satisfy them in IBR
dominated system?
Source: S. Tweed (EirGrid)
• Proliferation and fragmentation of services
• New codes, standards, tariff structures needed
• New market mechanisms needed
• See the new stability market in GB
 Summary
• Power system technical characteristics are changing rapidly due to replacement of SMs by IBRs
• SM: Physics-based, robust but slow
• IBRs: software-based, fast, agile but fragile
• Inverter challenge: operate the system efficiently and to the same reliability standard but using IBRs
• Significant challenges for modelling and analysis of IBR-systems
Power Power systems
• Operational problems: stability (SSO), fault currents, frequency control
and provision of services electronics

• Great time to do power system research!

• Need for interdisciplinary effort

Economics Control