C4 TECHNICAL BROCHURE

Power system technical

performance

Electromagnetic transient
simulation models for large-scale
system impact studies in power
systems having a high penetration
of inverter-connected generation

Reference: 881

September 2022
TECHNICAL BROCHURE

Electromagnetic transient simulation

models for large-scale system impact
studies in power systems having a
high penetration of inverter-connected
generation
WG C4.56

Primary members

B. BADRZADEH, Convenor AU M. DAVIES AU

S. GOYAL AU S. GROGAN AU
J. LU AU F. VILLELLA BE
J. MAHSREDJIAN CA D. MUTHUMUNI CA
F. FERNANDEZ DE A. KURI DE
L. DALL DK H. SAAD FR
M. VAL ESCUDERO IE J. SCHMALL US

Corresponding members

A.S. NETO BR A. ISAACS CA

D. KELL CA J. V. NAVA CA
A. REZAEI ZARE CA O. SAAD CA
Y. CHEN CN A. PASHAEI GB
S. SAHUKARI GB P. POURBEIK US
L. XU US

Copyright © 2022
“All rights to this Technical Brochure are retained by CIGRE. It is strictly prohibited to reproduce or provide this publication in any
form or by any means to any third party. Only CIGRE Collective Members companies are allowed to store their copy on their
internal intranet or other company network provided access is restricted to their own employees. No part of this publication may
be reproduced or utilized without permission from CIGRE”.

Disclaimer notice
“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any responsibility, as to the
accuracy or exhaustiveness of the information. All implied warranties and conditions are excluded to the maximum extent permitted
by law”.

WG XX.XXpany network provided access is restricted to their own employees. No part of this publication may be
reproduced or utilized without permission from CIGRE”.

Disclaimer notice ISBN : 978-2-85873-586-0

“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any
responsibility, as to the accuracy or exhaustiveness of the information. All implied warranties and
conditions are excluded to the maximum extent permitted by law”.
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Executive summary
Power systems around the world are transitioning to significantly higher shares of inverter-based
resources (IBR) with few synchronous generators remaining online. IBR and synchronous generators
have fundamentally different dynamic performance characteristics. System dynamics and technical
needs are therefore vastly different between synchronous and IBR dominated power systems, and these
differences will become greater as IBR uptake increases in the power system. Reductions in system
strength, inertia, damping of small-signal oscillations, fault contribution, and other synchronous
characteristics are the result of the transition to higher penetration of IBR in comparison with
conventional power systems with a dominance of synchronous generators.
These changes in system characteristics have caused new and emerging power system phenomena.
These phenomena either did not previously exist or when they did, they manifested themselves on a
limited scale and only in a small part of the power system. Energy transition infers that these phenomena
are likely to occur more often, will have the potential to impact a larger part of the power system, and
their absolute impact will be greater than it used to be. This could potentially result in major supply
disruptions such as system black events if such phenomena are not understood and addressed pre-
emptively. A testament to this widespread impact is the growing need for IBR remediations in real
generator connection applications including the need for installation of devices such as synchronous
condensers or control system tuning to avoid instabilities.
Worldwide experiences indicate that conventional power system stability analysis tools have often not
been able to predict these phenomena due to the simplifications inherent in these tools. The
consequence of this inability to predict the problems early enough is that problems might be first
experienced during actual power system operation, at which point it is more difficult to address and more
disruptive and costly to the connecting party; i.e. if the system operator or network owner needs to
invoke a constraint to pre-empt the impact on power system stability or nearby network users. Another
approach sometimes adopted is conservative power system operation in the absence of accurate
answers based on accurate models, which is an equally inefficient approach and not in the best interests
of consumers.
Detailed whole-system modelling has therefore been increasingly used in recent times in particular in
countries/regions with higher IBR penetration to address the problems discussed above. This will
facilitate accurate long-term power system planning allowing the identification and resolution of new and
emerging phenomena before they manifest in real power system operation. It will also permit a more
accurate albeit more involved assessment of the impact of connecting new IBR such as battery energy
storage systems, hydrogen electrolyser, solar and wind generation on power system planning and
operation. This detailed modelling will also facilitate better understanding of the performance of
emerging technologies such as grid-forming inverters and how best they can be designed to meet
emerging power system needs and technical requirements in power systems with significantly higher
IBR penetration.
The objectives of this Technical Brochure are to discuss
▪ Comparison of conventional phasor-domain transient (PDT) tools against electromagnetic transient
(EMT) tools in power systems with high IBR penetration and their applications and limitations
▪ The level of details to be included in an IBR EMT model to make it suitable for wide-area system
impact studies
▪ Approaches for whole-system EMT model development
▪ Model testing and validation for IBR EMT model for high IBR penetration studies
▪ Practical case studies on how these wide-area EMT models have been used across the world and
the insight provided by them which could not have been otherwise provided by conventional
phasor-domain models.

The Technical Brochure is organised as follows:

▪ Chapter 1 discusses relevant ongoing and concluded activities worldwide in particular those of
CIGRE and IEEE, elaborates the need for wide-area EMT modelling and outlines recent examples
of various jurisdictions worldwide who have used wide-area EMT modelling.

3
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ Chapter 2 compares PDT and EMT modelling with regard to factors such as the level of modelling
details included, the accuracy achieved, and examples where PDT is not suitable for power system
planning and operation, as well as those where PDT is sufficiently accurate. Various screening
methods are discussed, and their merits and applications are compared for making a decision on
the necessity of wide-area EMT modelling for each specific scenario, or limiting the number of EMT
studies in situations where several tens or hundreds of EMT studies may be required.
▪ Chapter 3 highlights the importance of vendor-specific site-specific IBR EMT models for generator
connection and power system operational studies, and discusses the applications where generic
IBR EMT models might be prudent to use including for long-term power system planning studies.
Adequate modelling of the IBR and network components including associated control and
protection systems, loads and distributed energy resource (DER) are discussed. Lastly a
discussion on model visibility/transparency/portability is presented recognising the confidential
nature of these models due to intellectual property rights contained in those models, and at the
same time challenges to make practical and meaningful conclusions from the black-box models.
▪ Chapter 4 presents a systematic and streamlined approach for developing wide-area EMT network
models from the load flow cases, and provides guidance on the principles of network equivalencing
for instances where a large part but not the whole power system is represented in EMT. The
advantages and disadvantages of other approaches including co-simulation, phasor-domain, and
real-time EMT simulation (as opposed to offline EMT simulation) are elaborated on.
▪ Chapter 5 highlights the importance of model acceptance testing to ensure that individual EMT
models are fit for purpose and do not exhibit any unexpected responses before they are integrated
into wide-area EMT models. This is done at early stages and before connecting the IBR to the
actual power system where no real system measurements is available. Several practical case
studies are then presented discussing the use of staged tests, natural network disturbances and
hardware-in-the-loop simulation after the IBR gets connected to the network.
▪ Chapter 6 overviews practical examples of using wide-area EMT model in different countries
worldwide including Australia, Belgium, Canada, mainland Europe, and the US. These examples
include system strength and adverse control system interactions studies between multiple IBR,
designing special protection schemes for the whole power system, interarea oscillations in
continental Europe system, protection system studies, and system separation studies for a normally
interconnected power system.

4
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Contents
Executive summary ............................................................................................................. 3

Figures and Illustrations ..................................................................................................... 8

Tables ................................................................................................................................. 12

Equations ........................................................................................................................... 13

Acronyms and abbreviations ........................................................................................... 14

1. State-of-the-art ......................................................................................................... 17
1.1 Relevant activities ................................................................................................................................... 17
1.1.1 CIGRE working groups ...................................................................................................................... 17
1.1.2 IEEE Working Groups and Task Forces ............................................................................................ 19
1.2 The need for wide-area EMT modelling ................................................................................................. 19
1.2.1 Overview ............................................................................................................................................ 19
1.2.2 Balancing the use of wide-area EMT modelling ................................................................................. 20
1.2.3 Prerequisites to enable wide-area EMT modelling ............................................................................. 20
1.3 Recent international experiences........................................................................................................... 20
1.3.1 Australian National Electricity Market - Mainland ............................................................................... 21
1.3.2 Tasmania ........................................................................................................................................... 23
1.3.3 Belgium .............................................................................................................................................. 24
1.3.4 Ireland and Northern Ireland .............................................................................................................. 25
1.3.5 Denmark ........................................................................................................................................... 26
1.3.6 France................................................................................................................................................ 27
1.3.7 Quebec, Canada ................................................................................................................................ 29
1.3.8 Texas ................................................................................................................................................. 31
1.3.9 UK ...................................................................................................................................................... 32
1.3.10 Brazil .................................................................................................................................................. 33

2. The role of EMT and PDT dynamic simulation for wide-area system studies ..... 35
2.1 Fundamental principles .......................................................................................................................... 35
2.1.1 Examples ........................................................................................................................................... 36
2.1.2 Summary ........................................................................................................................................... 39
2.2 Level of modelling details ....................................................................................................................... 39
2.2.1 IBR control systems ........................................................................................................................... 39
2.2.2 Network components ......................................................................................................................... 42
2.3 Comparison of simulated responses ..................................................................................................... 42
2.3.1 Examples indicating acceptable correlation ....................................................................................... 43
2.3.2 Example indicating unacceptable correlation ..................................................................................... 47
2.4 GUI considerations for wide-area EMT studies..................................................................................... 52
2.4.1 Hierarchy ........................................................................................................................................... 53
2.4.2 Layers ................................................................................................................................................ 54
2.4.3 Scripting and attributes ...................................................................................................................... 54
2.5 The role of screening methods............................................................................................................... 55
2.5.1 Available Fault level ........................................................................................................................... 55
2.5.2 Improved analytical multi-infeed interaction factor ............................................................................. 58
2.5.3 Advanced short-circuit strength metric ............................................................................................... 64
2.5.4 Comparison of screening methods .................................................................................................... 66

3. Model adequacy ....................................................................................................... 67

3.1 Importance of vendor-specific site-specific models ............................................................................ 67

5
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

3.2 Application of generic EMT models ....................................................................................................... 68

3.2.1 Long-term planning studies ................................................................................................................ 68
3.3 Large-scale inverter-connected generators .......................................................................................... 71
3.3.1 Prime mover ...................................................................................................................................... 72
3.3.2 Power electronic switching ................................................................................................................. 72
3.3.3 Control systems ................................................................................................................................. 73
3.3.4 Model aggregation ............................................................................................................................. 74
3.4 Network .................................................................................................................................................... 76
3.4.1 Static network elements ..................................................................................................................... 76
3.4.2 HVDC and FACTS ............................................................................................................................. 78
3.4.3 Protection systems............................................................................................................................. 78
3.5 Loads and Distributed Energy Resources (DER) modelling ................................................................ 80
3.5.1 General load and DER modelling....................................................................................................... 80
3.5.2 Large variable speed motor drives ..................................................................................................... 81
3.6 The need for more detailed DER modelling........................................................................................... 81
3.7 Model visibility/transparency/portability ............................................................................................... 83
3.7.1 Model visibility and confidentiality ...................................................................................................... 83

4. Large-scale EMT simulation.................................................................................... 87

4.1 Network development ............................................................................................................................. 87
4.1.1 Load flow case creation ..................................................................................................................... 87
4.1.2 Dynamic model initialisation ............................................................................................................... 88
4.1.3 Network equivalencing ....................................................................................................................... 89
4.1.4 Static voltage sources ........................................................................................................................ 90
4.1.5 Dynamic voltage sources ................................................................................................................... 90
4.1.6 Frequency dependent network equivalent (FDNE) ............................................................................ 91
4.1.7 Comparison of different static voltage source network equivalencing points ..................................... 92
4.2 Co-Simulation techniques ...................................................................................................................... 94
4.2.1 Hybrid simulation ............................................................................................................................... 95
4.3 Real-time EMT simulation ..................................................................................................................... 100

5. Acceptance testing and validation of EMT models ............................................. 102

5.1 Pre-commissioning model acceptance on individual models ........................................................... 102
5.1.1 Pre-requisite test – SMIB flat run ..................................................................................................... 102
5.1.2 Balanced fault – large disturbance test ............................................................................................ 102
5.1.3 Unbalanced fault – large disturbance test ........................................................................................ 102
5.1.4 Sequential fault – large disturbance test .......................................................................................... 103
5.1.5 Temporary overvoltage test ............................................................................................................. 103
5.1.6 Voltage reference step change test ................................................................................................. 103
5.1.7 Active power controller reference test .............................................................................................. 103
5.1.8 Grid frequency – controller test ........................................................................................................ 103
5.1.9 Inertia – frequency control test ......................................................................................................... 103
5.1.10 Grid voltage change – response test ............................................................................................... 104
5.1.11 Oscillatory rejection test ................................................................................................................... 104
5.1.12 Grid phase angle change – response test ....................................................................................... 104
5.1.13 Extremely weak network tests ......................................................................................................... 104
5.2 Cross-platform and long-run checks on wide-area models ............................................................... 105
5.2.1 Long-run stability checks ................................................................................................................. 106
5.3 Examples of validation against staged field test results ................................................................... 107
5.3.1 November 2019 staged test – voltage oscillation in Northwest Victoria, Australia ........................... 107
5.3.2 March 2020 test – Southwest New South Wales, Australia ............................................................. 109
5.3.3 April 2020 test – Northwest Victoria Australia, with revised inverter settings ................................... 110
5.4 Validation of wide-area EMT models against actual system disturbances ...................................... 111
5.4.1 Australian examples......................................................................................................................... 112
5.4.2 Gaspésie Peninsula Wind turbine model validation ......................................................................... 114
5.5 Hardware-in-loop testing ...................................................................................................................... 116
5.5.1 Software-based validation................................................................................................................ 116
5.5.2 Hybrid use of software and control replica ....................................................................................... 121

6
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

6. Simulation case studies ........................................................................................ 124

6.1 System strength studies ....................................................................................................................... 124
6.1.1 Impact of a proposed inverter connected generator on system strength ......................................... 124
6.1.2 Remediation for System Strength Adverse Impact .......................................................................... 127
6.1.3 Quantity and combination of synchronous generators/condensers.................................................. 131
6.1.4 Developing operational constraints .................................................................................................. 132
6.2 Adverse control interactions and harmonic instabilities ................................................................... 133
6.2.1 Sub-synchronous oscillations .......................................................................................................... 133
6.3 Designing system-wide control and protection schemes .................................................................. 134
6.4 Low Frequency Inter-area oscillations studies ................................................................................... 137
6.4.1 Introduction ...................................................................................................................................... 137
6.4.2 Event Study in Real-Time Simulations Using Replica ...................................................................... 137
6.4.3 On-Site Modification of the C&P by RTE and REE .......................................................................... 140
6.5 Protection studies ................................................................................................................................. 140
6.5.1 Introduction ...................................................................................................................................... 140
6.5.2 Power Swing Protection ................................................................................................................... 142
6.5.3 Expected misoperation issues due to IBRs ...................................................................................... 142
6.5.4 Inverter modelling requirements ...................................................................................................... 145
6.5.5 Conclusion ....................................................................................................................................... 148
6.6 System separation study ...................................................................................................................... 148

7. Conclusions and further work .............................................................................. 150

7.1 Key conclusions .................................................................................................................................... 150
7.2 Suggestions for future work ................................................................................................................. 151

References ....................................................................................................................... 153

APPENDIX A Dynamic Phasors ................................................................................... 161

A.1 Dynamic Phasors................................................................................................................................... 161

APPENDIX B EMT Studies for Geomagnetic Disturbances ....................................... 167

B.1 EMT Studies for Geomagnetic Disturbances ...................................................................................... 167
B.2 GMD Analysis Methods ......................................................................................................................... 168
B.2.1 Load flow ................................................................................................................................................ 168
B.2.2 PDT ......................................................................................................................................................... 168
B.2.3 EMT ......................................................................................................................................................... 169
B.3 Test system ............................................................................................................................................ 169
B.3.1 Load-Flow Model Implementation ........................................................................................................ 169
B.3.2 PDT Model Implementation ................................................................................................................... 170
B.3.3 EMT Model Implementation .................................................................................................................. 170
B.3.4 Cross-Examination of Implementations (Without GMD) .................................................................... 170
B.4 Comparison of GMD simulation results .............................................................................................. 170
B.5 Geomagnetically-Induced Currents ..................................................................................................... 170
B.6 Transformer VAR Loss.......................................................................................................................... 171
B.7 Voltage Stability Results ....................................................................................................................... 171
B.8 GMD Impacts on Wind Farms ............................................................................................................... 174
B.9 Conclusion ............................................................................................................................................. 175
B.10 Simulation Results ................................................................................................................................ 175

7
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figures and Illustrations

Figure 1.3-1 Simplified layout of Tasmanian power system .................................................................. 23
Figure 1.3-2 Historical development of the Belgian transmission system ............................................. 25
Figure 1.3-3 SCADA to EMT interface flow diagram [R10] ................................................................... 28
Figure 1.3-4 Hydro-Quebec grid for EMT studies: top view .................................................................. 30
Figure 2.1-1 Series R-L circuit ............................................................................................................... 35
Figure 2.1-2 Series R-C circuit .............................................................................................................. 35
Figure 2.1-3 Example circuit .................................................................................................................. 36
Figure 2.1-4 Response of an electric circuit simulated on an EMT platform – DC offsets, harmonics are
captured in the form of instantaneous time domain response .............................................................. 36
Figure 2.1-5 DC content in the bus voltage - Typical situation when clearing faults near series
compensated lines ................................................................................................................................. 37
Figure 2.1-6 Schematic representation of a two level VSC converter................................................... 37
Figure 2.1-7 Illustration of PLL response – The PLL is expected to lock to the grid frequency and provide
an accurate estimate of the instantaneous ‘angle’ of the voltage (and current) .................................... 38
Figure 2.1-8 Typical voltage and current waveforms experienced at the connection point of an IBR
following a system fault ......................................................................................................................... 39
Figure 2.2-1 PLL behaviour under phase jump (left) and unbalanced fault (right) tests in PDT and EMT
models [R24].......................................................................................................................................... 40
Figure 2.2-2 Frequency range of different power system phenomena and limits of PDT modelling .... 41
Figure 2.3-1 Study Comparison Loop ................................................................................................... 43
Figure 2.3-2 Solar farm active power response overlay: PDT vs. EMT ................................................ 43
Figure 2.3-3 Solar farm reactive power response overlay: PDT vs. EMT ............................................. 44
Figure 2.3-4 Sub-Test A: Active power output decrease comparison ................................................... 45
Figure 2.3-5 Sub-Test B: Active power output increase ........................................................................ 45
Figure 2.3-6 Sub Test A: Active power reference decrease comparison .............................................. 45
Figure 2.3-7 Sub Test B: Active power reference increase comparison ............................................... 45
Figure 2.3-8 Sub-Test A: Reactive power output change ..................................................................... 45
Figure 2.3-9 Sub-Test B: Reactive power output change ..................................................................... 45
Figure 2.3-10 Sub Test A: Reactive power reference change .............................................................. 46
Figure 2.3-11 Sub Test B: Reactive power reference change .............................................................. 46
Figure 2.3-12 Comparison of QT (upper) and PT (lower) of a synchronous machine for PDT (blue) and
EMT (red) models during a three-phase fault ........................................................................................ 46
Figure 2.3-13 Voltage oscillation in EMT simulation, not observed in PDT simulation ......................... 47
Figure 2.3-14 Active power response of LCC HVDC link in PDT and EMT simulation ......................... 48
Figure 2.3-15 Voltage profile at connection point of HVDC link in PDT and EMT simulation ............... 48
Figure 2.3-16 Voltage profile at the connection points of HVDC link in PDT and EMT simulation ....... 49
Figure 2.3-17 Voltage oscillation in EMT simulation, not observed in PDT simulation ......................... 50
Figure 2.3-18 HVDC Fault Response (Blue: Offline EMT, Green: RTS) .............................................. 51
Figure 2.3-19 PDT and EMT model performance comparison with suspect PDT model response ..... 52
Figure 2.3-20 PDT and EMT model performance comparison with suspect EMT model response ..... 52
Figure 2.4-1 Hierarchical circuit: subnetworks containing subnetworks with masks ............................. 53
Figure 2.4-2 A snapshot from a 735 kV series-compensated network ................................................. 53
Figure 2.4-3 Multimachine power plant representation ......................................................................... 54
Figure 2.5-1 Calculation of local IBR impact on connection point capability ......................................... 57
Figure 2.5-2 Calculation of remote IBR impact on connection point capabilities .................................. 57
Figure 2.5-3 Determination of MIIF factor between two connection points [R34] ................................. 59
Figure 2.5-4 Power electronics devices embedded in Northern France Network ................................. 62
Figure 2.5-5 Reduced network in EMT including the four power electronics components ................... 64
Figure 2.5-6 EMT simulation results of three-phase fault at Wind Farm POI to validate IBR-CCT
screening method – (a) blue curve: fault duration = 101.5 ms, (a) red curve: fault duration = 104.2 ms,
(b) red curve fault duration = 130.8 ms ................................................................................................. 65
Figure 3.1-1 Comparison between EMT model and product test .......................................................... 67
Figure 3.2-1 Concept of aggregate windfarm model ............................................................................. 70
Figure 3.3-1 Type 3 WTG model sub-systems...................................................................................... 72
Figure 3.3-2 Dynamic response of wind turbine with full (red) and average switch models (black) ..... 73
Figure 3.3-3 Converter control model for type 3 turbine used for wide-area studies ............................ 73
Figure 3.3-4 Model of the rotor-side converter control for a type 3 wind turbine including drive train
torsional damping .................................................................................................................................. 74

8
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 3.3-5 Aggregated Inverter-Based Resource (IBR) plant model ................................................. 75

Figure 3.4-1 Node-Breaker modelling (left) vs Bus-Branch modelling (right) ....................................... 76
Figure 3.5-1 DER and load model aggregation ..................................................................................... 81
Figure 3.6-1 Comparison of DER model response between PDT and EMT simulation for a 100 ms fault
resulting in a residual voltage of 0.7 pu ................................................................................................. 82
Figure 3.7-1 Common approach for decoupling of control elements from EMT tool ............................ 84
Figure 3.7-2 Control layers of an HVDC station [R54] .......................................................................... 84
Figure 4.1-1 Initialisation and dynamic blocks in an EXST1 exciter ...................................................... 88
Figure 4.1-2 Isolated EMT generator initialisation technique. Left: Prior to steady state. Right: Post
steady state. .......................................................................................................................................... 89
Figure 4.1-3 Frequency response (positive sequence impedances) for L-Network (orange) R-Network
(blue) ...................................................................................................................................................... 92
Figure 4.1-4 Comparison between synchronous generator response in different network models ...... 93
Figure 4.1-5 Comparison between transmission bus frequency in different network models ............... 93
Figure 4.2-1 Application case of a hybrid co-simulation in power system............................................. 94
Figure 4.2-2 Co-simulation of two sub-systems. The diagram on the left shows the signal exchange at
the co-simulation boundaries. The diagram on the right illustrates the synchronization of both sub-
systems at discrete times ...................................................................................................................... 95
Figure 4.2-3 Geographical representation of the synthetic network model of Texas, showing the split of
the network for the co-simulation (green area as EMT, grey area as PDT-balanced). ......................... 96
Figure 4.2-4 Transient voltage behaviour for a busbar fault, showing PDT simulation (dashed blue), EMT
(green) and PDT-EMT co-simulation (red). The diagram on the right shows a zoomed scope of the
diagram on the left. ................................................................................................................................ 97
Figure 4.2-5 Different components that can be imported as DLL files in an IBR generation plant ....... 97
Figure 4.2-6 Scheme used to perform EMT-EMT co-simulations with multiple targets ........................ 98
Figure 4.2-7 EMT-EMT simulations to address confidentiality concerns .............................................. 98
Figure 4.2-8 Modified EMT-EMT simulations to reduce model management resource requirements .. 99
Figure 4.3-1 Simplified process of RTS solution ................................................................................. 100
Figure 5.1-1 Oscillatory rejection tests on PDT grid voltage (1-10Hz, 1Hz per step) ......................... 104
Figure 5.2-1 Active and reactive power comparisons for dynamic behaviour ..................................... 105
Figure 5.2-2 Bus voltage phasor module comparisons, dynamic performance .................................. 106
Figure 5.2-3 Long term simulation: synchronous generator speed signal with and without transformer
magnetisation branches ...................................................................................................................... 107
Figure 5.3-1 Voltage oscillation overlay between simulation and actual measurement (overview) .... 108
Figure 5.3-2 Voltage oscillation overlay between simulation and actual measurment (zoomed view) 108
Figure 5.3-3 Transmission bus voltage profile - March 2020 test ....................................................... 109
Figure 5.3-4 Transmission bus voltage profile - March 2020 test (zoomed view) ............................... 110
Figure 5.3-5 Voltage profile at solar farm connection point ................................................................. 111
Figure 5.3-6 Voltage profile at solar farm connection point (zoomed view) ........................................ 111
Figure 5.4-1 Generator 1 active power response comparison ............................................................ 112
Figure 5.4-2 Generator 1 reactive power response comparison ......................................................... 112
Figure 5.4-3 Generator 2 active power response comparison ............................................................ 113
Figure 5.4-4 Generator 2 reactive power response comparison ......................................................... 113
Figure 5.4-5 Generator 3 active power response comparison ............................................................ 114
Figure 5.4-6 Generator 3 reactive power response comparison ......................................................... 114
Figure 5.4-7 Measurement points of the Type 3 Wind Turbine ........................................................... 114
Figure 5.4-8 Recorded positive- and negative-sequence voltages at stator level; comparisons of currents
for different time frames, for recorded and simulated waveforms at the Hydro-Québec Type 3 WG level
for event 6. The simulation results and measurements are overlayed................................................ 116
Figure 5.5-1 Overlay of actual and RTS simulated L-G voltages at 220 kV busbar during the fault and
its recovery period ............................................................................................................................... 117
Figure 5.5-2 Overlay of actual and RTS simulated HVDC AC currents during the fault and its recovery
period ................................................................................................................................................... 117
Figure 5.5-3 Overlay of actual and RTS simulated HVDC active power during the fault and its recovery
period ................................................................................................................................................... 117
Figure 5.5-4 Overlay of actual and RTS simulated HVDC reactive power during the fault and its recovery
period ................................................................................................................................................... 118
Figure 5.5-5 Variation of HVDC Power and AC Frequency within ± half hour of the fault .................. 118
Figure 5.5-6 Overlay of actual and RTS simulated L-G voltages at a remote 110 kV busbar during the
fault ...................................................................................................................................................... 119
Figure 5.5-7 Overlay of actual and RTS simulated HVDC power flows during and for 15s after the fault
............................................................................................................................................................. 119

9
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.5-8 Overlay of actual and RTS simulated AC Frequency during and for 15s after the fault . 119
Figure 5.5-9 Overlay of actual and RTS simulated Type 3 windfarm power flows during and for 15s after
the fault ................................................................................................................................................ 119
Figure 5.5-10 Overlay of actual and RTS simulated Type 4 windfarm power flows during and for 15s
after the fault ........................................................................................................................................ 120
Figure 5.5-11 Overlay of actual and RTS simulated voltages at Type 4 windfarm during the fault .... 120
Figure 5.5-12 Overlay of actual and RTS simulated Type 4 windfarm currents during the fault ........ 120
Figure 5.5-13 Overlay of actual, RTS and PDT simulated voltages at 110 kV PCC during and for 10 s
after the fault ........................................................................................................................................ 120
Figure 5.5-14 Overlay of actual, RTS and PDT Active & Reactive powers at 110 kV PCC during and for
10 s after the fault ................................................................................................................................ 121
Figure 5.5-15 a) Network representation of the INELFE link, b) HIL of INELFE Replicas with Real Time
simulator .............................................................................................................................................. 122
Figure 5.5-16 Active power step – Active power (a), DC current (b) [R65] ........................................ 122
Figure 5.5-17 Converter block/deblock – Comparison with on-site measurements [R65] .................. 123
Figure 5.5-18 Converter deblock – Comparison with on-site measurements (zoom) ......................... 123
Figure 6.1-1 Map of the system under consideration for the system strength studies for a new IBR in
Queensland, Australia ......................................................................................................................... 125
Figure 6.1-2 System performance before adding the new IBR – Voltage at a key node in the network
............................................................................................................................................................. 126
Figure 6.1-3 System performance after adding the new IBR – Voltage at a key node in the network 126
Figure 6.1-4 System performance after adding the new IBR – Voltage at a key node in the network 127
Figure 6.1-5 Output of the new IBR ..................................................................................................... 127
Figure 6.1-6 System performance after adding a Synchronous Condenser with the new IBR – Voltage
at a key node in the network................................................................................................................ 128
Figure 6.1-7 System performance after adding a Synchronous Condenser with the new IBR – Voltage
at a key node in the network................................................................................................................ 128
Figure 6.1-8 Output of the new IBR after adding a Synchronous Condenser ..................................... 129
Figure 6.1-9 System performance with modified control of the IBR - Voltage at a key node in the network
............................................................................................................................................................. 129
Figure 6.1-10 Output of the new IBR with modified control ................................................................. 130
Figure 6.1-11 System performance with old and new control parameters of the IBR – Voltage at a key
node in the network ............................................................................................................................. 130
Figure 6.1-12 Output of the IBR with old control parameters .............................................................. 131
Figure 6.1-13 Output of the IBR with new control parameters ............................................................ 131
Figure 6.2-1 Observed sustained voltage in Northwest Victoria, Australia from field monitors .......... 133
Figure 6.2-2 Voltage oscillation measurement obtained from tests in Northwest Victoria, Australia .. 134
Figure 6.2-3 Comparison between constraining MW output and total number of connected inverters
............................................................................................................................................................. 134
Figure 6.3-1 Synthetic sketch of the situation analysed ...................................................................... 135
Figure 6.3-2 Study approach for planning ........................................................................................... 135
Figure 6.3-3 High Level Comparison between assessment of stability of classical units vs PE based in
the case of analysis ............................................................................................................................. 136
Figure 6.4-1 Measured frequencies in different locations of central Europe during the 1st Dec 2016 event
[R68] .................................................................................................................................................... 137
Figure 6.4-2 AC and DC interconnections between France and Spain Transmission grids and mode
shape ................................................................................................................................................... 137
Figure 6.4-3 Simplified Model used in Real-Time Studies with Replica [R69] .................................... 138
Figure 6.4-4 Results of 3 simulations regarding the time constant of the P-mode-3 filter [R69] ......... 139
Figure 6.5-1 PV Solar Farm ................................................................................................................. 141
Figure 6.5-2 FSC WTG........................................................................................................................ 141
Figure 6.5-3 DFIG WTG ...................................................................................................................... 142
Figure 6.5-4 PSB mis-operation due to a faster swing under IBR ...................................................... 143
Figure 6.5-5 OST Malfunction due to changed swing trajectory under IBRs ...................................... 144
Figure 6.5-6 Impact of IBR on EC ....................................................................................................... 144
Figure 6.5-7 Comparison of power swing simulation results of the IEEE PSRC D29 test system under
0% IBR in EMT and PDT simulations. ................................................................................................. 147
Figure 6.5-8 Comparison of power swing simulation results of the IEEE PSRC D29 test system under
25% IBR in EMT and PDT simulations. ............................................................................................... 147
Figure 6.5-9 Comparison of power swing simulation results of the IEEE PSRC D29 test system under
50% IBR in EMT and PDT simulations. ............................................................................................... 148
Figure 6.6-1 James Bay system voltage oscillations due to an extreme disturbance ......................... 149

10
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure A-1 Basic RL circuit energization ............................................................................................. 161

Figure A-2 Simple Circuit ..................................................................................................................... 162
Figure A-3 Time-domain solution time-points for an ODE solution ..................................................... 163
Figure A-4 Inductor companion circuit model for time-domains solution ............................................ 164
Figure A-5 Three subnetworks separated by transmission lines ........................................................ 164
Figure A-6 Sparse matrix for circuit diagram of Figure A-5, the number of non-zeros is 60 ............... 164
Figure A-7 Comparison of time-domain (EMT) in blue and phasor-domain (TS, voltage phasor
magnitude) in red, voltages (kV).......................................................................................................... 165
Figure B-1 GMD vulnerability assessment process of TPL-007 [R104]. ............................................. 168
Figure B-2 Definition of x and y coordinates, the direction of GEF vector, and the orientation of
transmission line. ................................................................................................................................. 170
Figure B-3 GIC flow in line and transformer neutral under GEF=1 V/km, simulated in EMT .............. 172
Figure B-4 The dependence of GIC on ac voltage level: (a) instantaneous value of magnetic flux in
transformer core; (b) the dc component of magnetic flux; (c) instantaneous value of transformer
magnetization current; (d) the dc component of transformer magnetization. ...................................... 173
Figure B-5 Transformer var loss under GEF=1 V/km simulated in EMT. ............................................ 173
Figure B-6 Simulation of bus voltages in EMT under: (a) GEF=4V/km (stable) and (b) GEF=5.5V/km
(unstable). ............................................................................................................................................ 173
Figure B-7 Wind farm under study with 18 DFIG turbines .................................................................. 174
Figure B-8 MOT operating conditions under GIC of 200 A at neutral a) voltage, b) current ............... 174
Figure B-9 Wind farm THDs versus GIC at full load............................................................................ 175
Figure B-10 MOT hotspot temperature during the benchmark GIC event, a) benchmark GIC signature,
b) MOT winding hotspot temperature under the GIC benchmark with various magnitudes. .............. 175
Figure B-11 Response of the SM on bus Breed to a three-phase 100-ms bolted fault on bus 4 of 118-
GMD within EMT and PDT: (a) machine speed; (b) electric power output; (c) turbine mechanical power;
(d) field voltage; (e) PSS control signal output; and (f) OEL control ................................................... 178
Figure B-12 GIC flow in line and transformer neutral under GEF=1 V/km and θGEF=35° simulated in
EMT ..................................................................................................................................................... 179
Figure B-13 The dependence of GIC on ac voltage level: (a) instantaneous value of magnetic flux in
transformer core; (b) the dc component of magnetic flux; (c) instantaneous value of transformer
magnetisation current; (d) the dc component of transformer magnetisation ....................................... 183
Figure B-14 Transformer var loss under GEF=1 V/km and θGEF=35° simulated in EMT.................. 184
Figure B-15 Voltage of bus 9 simulated in PDT under: (a) GEF=8V/km (stable) and (b) GEF=9V/km
(unstable). ............................................................................................................................................ 186
Figure B-16 Simulation of bus voltages in EMT under: (a) GEF=4V/km (stable) and (b) GEF=5.5V/km
(unstable). ............................................................................................................................................ 187

11
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Tables
Table 2-1 Adequate representation of wind turbine components in PDT and EMT tools as function of
studies of interest .................................................................................................................................. 41
Table 2-2 Defined Terms ....................................................................................................................... 56
Table 2-3 AFL Calculation steps for example (2) .................................................................................. 58
Table 2-4 MIIF Amplitude ...................................................................................................................... 59
Table 2-5 Weighted MIIF ....................................................................................................................... 60
Table 2-6 Weighted MIIF amplitude in percentage ............................................................................... 60
Table 2-7 MIIF Angle ............................................................................................................................. 60
Table 2-8 Weighted MIIF angle in percentage ...................................................................................... 61
Table 2-9 Peak load - MIIF weighted amplitude .................................................................................... 62
Table 2-10 Off peak - MIIF weighted amplitude .................................................................................... 62
Table 2-11 Peak load - MIIF weighted angle ......................................................................................... 63
Table 2-12 Off peak - MIIF weighted angle ........................................................................................... 63
Table 2-13 Comparison of three screening methods ............................................................................ 66
Table 5-1 Recorded events used to validate type 3 wind turbine model............................................. 115
Table 6-1 Comparison between PDT and Real-time simulation results ............................................. 140
Table 6-2 Key modelling difference between cross-examined PDT and EMT models of a Type 3 WTG
model ................................................................................................................................................... 146
Table 6-3 Calculated PSB time delay from the PDT and EMT simulations as a function of IBR integration
level ..................................................................................................................................................... 148
Table B-1 Comparison minimum GEF amplitudes causing voltage collapse calculated by the LF, TS,
and EMT methods ............................................................................................................................... 173
Table B-2 Minimum GEF amplitude causing voltage collapse calculated by the PDT under different
saturation time delay values. ............................................................................................................... 173
Table B-3 Cross-examination of load-flow solution of the EMT and LF/PDT implementations (without
GMD) ................................................................................................................................................... 175
Table B-4 Cross-examination of GIC in the neutral of load substation transformers under GEF=1 V/km
and θGEF=35° calculated by the LF/PDT and EMT methods ............................................................... 179
Table B-5 Cross-examination of GIC in the neutral of load substation transformers under GEF=1 V/km
and θGEF=35° calculated by the LF/PDT and EMT methods ............................................................ 181
Table B-6 Cross-examination of GIC in the neutral of generator substation transformers under GEF=1
V/km and θGEF=35° calculated by the LF/PDT and EMT methods ................................................... 182
Table B-7 Cross-examination of effective GIC in the neutral of three-winding grid transformers under
GEF=1 V/km and θGEF=35° calculated by the LF/PDT and EMT methods. ...................................... 183
Table B-8 GIC calculation error of the LF/PDT methods as a function of GEF amplitude .................. 183
Table B-9 Cross-examination of var loss of load substation transformers under GEF=1 V/km and
θGEF=35° calculated by the LF/PDT and EMT methods ...................................................................... 184
Table B-10 Cross-examination of var loss of generator substation transformers GEF=1 V/km and
θGEF=35° calculated by the LF/PDT and EMT methods. .................................................................. 185
Table B-11 Cross-examination of var loss of three-winding transformers under GEF=1 V/km and
θGEF=35° calculated by the LF/PDT and EMT methods. .................................................................. 185
Table B-12 Total var loss of transformers calculated by the LF, TS, and EMT methods.................... 186
Table B-13 Error of calculated transformer var loss by the LF/PDT methods as a function of GEF
amplitude. ............................................................................................................................................ 186
Table B-14 Comparison of the minimum GEF amplitude causing voltage collapse calculated by the LF,
TS, and EMT methods......................................................................................................................... 186
Table B-15 Minimum GEF amplitude causing voltage collapse calculated by the PDT under different
saturation time delay values. ............................................................................................................... 187

12
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Equations
Equation 1.............................................................................................................................................. 35
Equation 2.............................................................................................................................................. 35
Equation 3.............................................................................................................................................. 35
Equation 4.............................................................................................................................................. 35
Equation 5.............................................................................................................................................. 36
Equation 6.............................................................................................................................................. 36
Equation 7.............................................................................................................................................. 36
Equation 8.............................................................................................................................................. 36
Equation 9.............................................................................................................................................. 59
Equation 10............................................................................................................................................ 65
Equation 11.......................................................................................................................................... 137
Equation 12.......................................................................................................................................... 161
Equation 13.......................................................................................................................................... 161
Equation 14.......................................................................................................................................... 161
Equation 15.......................................................................................................................................... 162
Equation 16.......................................................................................................................................... 162
Equation 17.......................................................................................................................................... 162
Equation 18.......................................................................................................................................... 163
Equation 19.......................................................................................................................................... 163
Equation 20.......................................................................................................................................... 163
Equation 21.......................................................................................................................................... 164
Equation 22.......................................................................................................................................... 165
Equation 23.......................................................................................................................................... 165
Equation 24.......................................................................................................................................... 165
Equation 25.......................................................................................................................................... 165
Equation 26.......................................................................................................................................... 165
Equation 27.......................................................................................................................................... 166
Equation 28.......................................................................................................................................... 166
Equation 29.......................................................................................................................................... 168
Equation 30.......................................................................................................................................... 169

13
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Acronyms and abbreviations

AC Alternating Current
AEMO Australian Energy Market Operator
AFL Available Fault Level
AVR Automatic Voltage Regulator
C&P Control and Protection
CCT Critical Clearance Time
CP Constant Parameter
CPU Central Processing Unit
CSC Coupled Sequence Control
DC Direct Current
DER Distributed Energy Resources
DFIG Doubly-Fed Induction Generator
EC Electrical Centre
EMT Electro-Magnetic Transients
FACTS Flexible AC Transmission System
FDNE Frequency- Dependent Network Equivalent
FMI Functional Mock-up Interface
FRT Fault Ride Through
FSC Full-Size Converter
GEF Geo-Electric Field
GIC Geo-magnetically Induced Currents
GMD Geo-Magnetic Disturbance
GPU Graphics Processing Unit
GSC Grid Side Converter
GUI Graphical User Interface
HIL Hardware In the Loop
HSM High Speed Monitoring
HV High Voltage
HVDC High Voltage Direct Current
HVRT High Voltage Ride Through
IBR Inverter Based Resource
IM Induction Machine
IP Intellectual Property
JWG Joint Working Group
LCC Line Commutated Converter
LF Load-Flow
LV Low Voltage

14
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

LVRT Low Voltage Ride Through

MATE Multi-Area Thevenin Equivalence
MIIF Multi-infeed Interaction Factor
MMC Modular Multilevel Converter
MPPT Maximum Power Point Tracking
MSCR Minimum Short Circuit Ratio
MV Medium Voltage
OEM Original Equipment Manufacturer
NEM National Electricity Market
NERC North American Electric Reliability Council
NGET National Grid Electrical Transmission
OOS Out-Of-Step
OST Out of Step Tripping
PDT Phasor-Domain Transients
PE Power Electronics
PLL Phase Locked Loop
PCC Point of Common Coupling
PoC Point of Connection
POI Point of Interconnection
PSB Power Swing Blocking
PV Photo-Voltaic
PWM Pulse Width Modulation
RCP Rapid Control Prototyping
REE Red Eléctrica de España
RMS Root Mean Square
RoCoF Rate of Change of Frequency
RSC Rotor Side Converter
RTE Réseau de Transport d'Électricité
RTS Real Time Simulation
SaaS Software as a Service
SCADA Supervisory Control And Data Acquisition
SCL Short Circuit Level
SCR Short Circuit Ratio
SG Synchronous Generator
SIL Software In the Loop
SM Synchronous Machine
SMIB Single Machine Infinite Bus
SNSP System Non-Synchronous Penetration
SSCI Sub Sub-Synchronous Control Interaction
SSG Synchronous Three-Phase Fault Level [MVA]

15
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

SSRI Sub Sub-Synchronous Resonance Interaction

SSTI Sub-Synchronous Torsional Interactions
STATCOM Static Synchronous Compensator
SVC Static Var Compensator
TF Task Force
TN TasNetworks
TOV Temporary Over-Voltage
TRV Transient Recovery Voltage
TS Transient Stability
TSO Transmission System Operator
UDM User-Defined Models
VSC Voltage Source Converter
VSG Virtual Synchronous Generator
WG Working Group
WP Wind Park
WSCR Weighted Short Circuit Ratio
WTG Wind Turbine Generator

16
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

1. State-of-the-art
Power system operators and network owners rely on power system modelling and simulation to maintain
secure operation of power systems. Conventional power system simulation models, typically referred to
as phasor-domain transients (PDT)1 models, have been used worldwide by all major network owners
and system operators for predicting the response of power systems when subjected to credible and
non-credible contingency events.
These types of models represent a trade-off between acceptable accuracy and simulation speed and
have proven to be acceptable for power systems having a large amount of conventional synchronous
generation and a limited penetration of non-synchronous energy sources. However, such system
models lose accuracy as the ratio of synchronous to inverter-based generation (including wind turbines,
solar inverters, battery energy storage systems, and variable speed pumped storage units) declines.
This primarily stems from the fast control systems used in non-synchronous generation, the dynamics
of which cannot be adequately represented in PDT simulation tools. It follows that such models are not
suitable for predicting the response of power systems to major disturbances, e.g. causation chain that
may result in a major supply disruption, or during extreme operating conditions which may include
islanding events that follow loss of major transmission in-feeds.
Electromagnetic transient (EMT) simulation tools address the deficiencies of PDT models. Such
modelling tools are in use by all major power system equipment manufacturers for designing equipment,
however, to date they have found limited application for large-scale power system studies. This is due
to the computational burden associated with running large numbers of EMT models in parallel as well
as the difficulties in many jurisdictions of sourcing such models from the original equipment
manufacturers (OEMs). To address the speed of simulation issues associated with EMT models, state-
of-the-art solution techniques are being progressively developed by software developers. Concurrently,
improvements are being applied to the speed and robustness of the simulation models developed by
OEMs. Power system modeling engineers in regions with a high penetration of inverter connected
generation are already observing limitations in the use of PDT models. Some regions have already
developed large-scale EMT model of large parts of their systems. This includes Australia and Texas,
USA where EMT models are used extensively for making operational decisions. This Technical
Brochure serves as a platform for the dissemination of knowledge, lessons learned, recommended
practices, intended applications, and underlying reasons for the use of EMT models for large-scale
stability studies in power systems having a high penetration of inverter connected generation.
Note that best endeavors have been applied to use the terms PDT and EMT consistently throughout
this document including in the figures presented. However, some of the figures were originally developed
by some of the members of this WG several years ago, which have occasionally referred to PDT as
RMS (root mean square). An editable version of some of these figurers was not available.

1.1 Relevant activities

1.1.1 CIGRE working groups
1.1.1.1 JWG C4/C6.35/CIRED: Modelling of inverter based generation for power
system

dynamics studies
The objective of JWG C4/C6.35/CIRED was to review and report on the latest developments relating to
inverter based resources (IBR) modelling for power system dynamic studies. Its main outcome, TB 727
(“Modelling of Inverter-Based Generation for Power System Dynamic Studies”), was published in May
2018.
TB 727 covers two types of models; being the PDT and EMT and provides guidance on the selection of
appropriate IBR models at a plant level, including the required characteristics and functions that should
be represented, as well as the type of model that is most appropriate for each type of study. The benefits

1
These models are sometimes referred to as root-mean square (RMS) or positive-sequence fundamental frequency models. See Chapter 2 for
further discussion on the terminology used in this TB.

17
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

and limitations of PDT models are discussed and cases where the use of EMT is warranted are
identified.
While similar in scope, the intent of WG C4.56 is not to expand on the work of JWG C4/C6.35/CIRED
but rather focus on the use of EMT models for system-wide studies related to integration of high
penetration of IBR.
1.1.1.2 CIGRE JWG B4.82/IEEE: Guidelines for use of real-codes in EMT models for
HVDC, FACTS and inverter based generators for power system analysis
To facilitate the effective use of EMT models, new approaches such as “real-code” models are being
used. These models incorporate the exact code used in the equipment installed in the field and can be
traced back to specific inverter code versions and settings changes. This joint IEEE and CIGRE WG is
leading an effort to standardize the approaches for using real-code models that ensure improved
usability, compatibility, and interoperability.
This method has been tested and used commercially in a number of supplier models (both EMT and
PDT tools) with good success and will be generally available when the IEEE/CIGRE Joint Task Force
has completed its work.
The output of this WG is an enabler for the work focused on WG C4.56 relying on more robust and
accurate EMT models.
1.1.1.3 CIGRE WG B4.81: Interaction between nearby VSC-HVDC converters, FACTs
devices, HV power electronic devices and conventional AC equipment
CIGRE WG B4-81 “Interaction between nearby VSC-HVDC converters, FACTs devices, HV power
electronic devices and conventional AC equipment” started in 2019 and is planned to be finalized in
2022. The outcome of this WG will provide insight into interactions between VSC-HVDC converters and
other power electronics devices or passive HV components installed on the network, that can occur over
a wide range of frequencies: from inter-area oscillations, to sub-synchronous interaction (as SSRI and
SSTI) and even to high frequency interaction (between 100 Hz to several kHz). In addition, interactions
due to non-linear behaviors such as transformer saturation and control non-linearity are also covered.
This WG aims to provide an overview on the interaction phenomena and to provide recommendation on
the appropriate simulation tools (PDT, EMT, real-time simulation, small-signal stability, etc.) to analyse
such phenomena.
1.1.1.4 WG C4.49 Multi-frequency stability of converter-based modern power systems
CIGRE WG C4.49 entitled “Multi-frequency stability of converter-based modern power systems” was
established in 2018 with the main objective of improving the understanding of new phenomena relating
to sub-synchronous and super-synchronous (harmonic) stability issues of grid-connected power
electronic devices. The main activities of the working group focus on providing clear explanation of the
phenomena, consolidating definitions, and describing available methods for modelling, analysis,
evaluation, and mitigation techniques. Guidelines regarding the general approach to such studies and
the availability as well as choice of tools are addressed by the working group. These methods include
but are not restricted to EMT simulation.
1.1.1.5 CIGRE JWG C4/B4.52 "Guidelines for Sub-synchronous Oscillation Studies in
Power Electronics Dominated Power Systems”
CIGRE WG C4/B4.52 entitled “Guidelines for Sub-synchronous Oscillation Studies in Power Electronics
Dominated Power Systems” was established in 2019 and it aims to conclude its work in early 2022. This
WG focuses on sub-synchronous oscillations in power system as opposed to overall wide-area stability
studies, which is the focus of WG C4.56. The Technical Brochure begins with a classification of sub-
synchronous oscillations ranging from classical to more contemporary forms. This is followed by a
summary of industry practices, challenges and experiences. The WG then investigates various
screening methods for identifying the risk of sub-synchronous oscillations before conducting EMT
studies. The objective of these screening methods is limited to identifying sub-synchronous oscillations
and as such differ from those considered in WG C4.56 on the choice between PDT and EMT modelling
and studies. Examples of EMT studies for assessing the risk of sub-synchronous oscillations are then
discussed. However, EMT model adequacy, network model development, model acceptance testing
and validation are not considered. The WG also discusses different practical mitigation methods of sub-
synchronous oscillations.

18
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

1.1.1.6 WG C4.60 Generic EMT-Type Modelling of Inverter-Based Resources for Long

Term Planning Studies
CIGRE WG C4.46 was established in 2020 with the objective of developing a set of publicly available
generic EMT IBR models aimed for long term planning studies. This WG complements the work of
C4.56, which focuses on vendor specific models. The new generic models will enable exploratory and
futuristic IBR-related EMT studies and will provide a tool for researchers and engineers to compare
simulation results, methods and solutions on the same basis and, thus, have more confidence in results
and conclusions. These new generic models will be validated and benchmarked across various EMT
software platforms to ensure model consistency.
1.1.2 IEEE Working Groups and Task Forces
1.1.2.1 Task Force on Modeling and simulation of large power systems with high
penetration of inverter-based generation
This Task Force (TF) looks at the availability, applicability, usability and reliability of present state-of-
the-art mathematical models for representation of inverter-based resources (IBR) in the various
simulation platforms ranging from positive-sequence phasor-domain simulation tools to EMT simulation
tools. This TF was formed by and under the Power System Dynamic Performance Committee of the
IEEE Power and Energy Society. The main goal of the TF is to look at all of the state-of-the art tools
and techniques for simulation of large scale power systems with high penetration of IBR, and to make
recommendations on potential future needs in the modeling and simulation domain. As stated, the TF
is looking at all types of models, with an equal emphasis on positive-sequence phasor-domain stability
tools and modeling. Thus, the main focus is not on just EMT modelling. Furthermore, developing
industry standard methods for creating and validating wide-area EMT models is not a key focus of this
taskforce.

1.2 The need for wide-area EMT modelling

1.2.1 Overview
EMT analysis has been used for decades for many applications such as HVDC and FACTs design, SSO
analysis, transient analysis, and has become relatively common in evaluation of renewable
interconnection in weak systems. The ability of EMT tools to accurately model fast inner controls and
protections is well recognized as being required for these studies, and there are myriad examples of
reliability or operability risks which have been predicted and mitigated with the use of EMT tools. As
renewable penetration is increasing, the applicability of EMT tools for a wider range of studies is
becoming evident. For example, recent system events in North America [R1] have caused the North
American Electric Reliability Council (NERC) to strengthen its language on requirements for EMT
modelling and studies as it was determined that conventional PDT tools were inadequate in predicting
the events. PDT tools serve an important function, and will remain in our processes for years to come,
but as IBR penetration increases and these events become more common, it is increasingly clear that
EMT models must play a larger role in future planning and analysis of large power grids.
An increasing number of utilities are adopting wide-area EMT analyses to supplement their reliability
studies in the planning process. Examples of these include:
▪ Hawaiian Electric (HECO) has recently performed full planning studies in both PDT and EMT for
their entire island-wide networks, anticipating 100% renewable penetration scenarios by 2023 [R2].
These studies were able to present highly specific recommendations on operation and controls to
improve grid reliability. Island-wide EMT studies continue to be performed in the HECO system.
An important component of these studies was to evaluate the use of Grid Forming (GFM) control
technology in battery applications as a mitigation for high penetration/weak grid issues.
▪ ISO-NE has been requiring performance of EMT studies for many years for transmission connected
renewable generation, but recently has been requesting evaluation of large clusters of Distributed
Energy Resources (DER) [R3].
▪ ERCOT requires EMT models for every interconnection and has been utilizing EMT for large high
renewable pockets for several years, and continues to increase their use of EMT in routine planning
work.
▪ ATC (within MISO footprint) now routinely performs wide-area EMT analysis as part of their
planning process as a result of very high renewable penetrations.
▪ At the direction of NERC and other entities, numerous large regional power grids have begun
preparing themselves for increasing EMT analysis by putting in place requirements for EMT models

19
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

as part of their interconnection requirements. In North America, these entities include MISO,
CAISO, TVA, and others. Many other entities are in the process of developing requirements in
anticipation of a larger role for EMT in planning.
▪ All new and existing generators connecting or connected to Australia’s eastern interconnected
power system (National Electricity Market or NEM) are required to provide both PDT and EMT
manufacturer-specific and site-specific models of their plant for all connection and planning studies,
with only limited exceptions to the requirement (e.g. very small plant in strong networks).
1.2.2 Balancing the use of wide-area EMT modelling
Wide-area EMT analysis is capable of reducing exposure to serious network events and increasing grid
reliability. Accordingly, many transmission entities are incorporating EMT to various degrees, however
this does not come without challenges. EMT modelling is complicated, requires special training or
specialist consultants, special computer software and hardware, new rules and regulations, and a lot
more time and effort on the part of the engineering groups. Model creation, data collection and checking
is technically difficult and often requires extensive quality control and iteration between planners and
generators. Once the models are acceptable, the studies themselves require significant technical
investment, and come with significant schedule and cost implications. This often presents challenges to
align the expectations and the effective involvement of the different stakeholders, from those responsible
for the system development and operation to the asset manufacturers and the owners of the
installations.
This fundamental conflict between the pressure to increase the speed of renewable interconnections
and the “slowing-down” effect of additional detail in the study process required to maintain grid reliability
is difficult to reconcile. Section 4.2 in this document presents concepts which may help in this area, but
in many cases this fundamental conflict may not be resolved in the short term without compromise.
In any case, a wide-area EMT model and simulation is a toolbox that has to complement and,
considering the performance and model complexity, cannot completely substitute other classical wide-
area system modelling and simulation approaches, such as load flow and PDT simulation.
1.2.3 Prerequisites to enable wide-area EMT modelling
1.2.3.1.1 Regulatory environment
In order to implement EMT studies on a wider scale, entities charged with conducting the studies need
the ability to request / demand site-specific models of sufficient quality from proponents and, in some
cases, retrospectively. Generic or standardized models are often not appropriate for detailed analysis
but, in some cases, non-project specific vendor (blackbox) models can be obtained at early stage of
project development and can be used in initial screening analysis. To be fair to generator proponents,
the modelling requirements must be clear and reasonably achievable, and additional support may be
required from the transmission utility in regions where these requirements haven’t been implemented
previously.
1.2.3.1.2 Additional schedule and resource requirements
As discussed in section 1.2.2, the addition of EMT analysis into the routine planning practices of existing
study groups presents a significant additional technical burden on these groups. These engineering
groups likely require additional training, computing resources, and human resources in order to cope
with the increased study burden. In addition, the study timeframes are likely to be longer than previously,
at least until such time as the tools and human resources are caught up.
Since the engineering and software skills required to conduct wide-area EMT studies are new and still
developing, investment in human resources is required at all levels, including at the engineering and
computer science academic level.

1.3 Recent international experiences

This section provides an overview of the current situation and practices adopted in ten different
jurisdictions with various levels of IBR integration and system interconnection (or system strength).
While Australia and some Independent System Operators (ISOs) in the USA have already adopted EMT
models and tools for system-wide stability analysis, a trend is emerging with many other countries
starting to develop capabilities and models with the view of adopting this tool in short to medium terms.
The common driver for introducing EMT in planning and operational security analysis is the integration
of high levels of renewable generation (IBR) and the decommissioning of conventional plant. In addition,

20
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

the advancement of EMT tools and computer hardware have supported the introduction of EMT based
simulations in wider applications.
1.3.1 Australian National Electricity Market - Mainland
The Australian National Electricity Market (NEM) power system has experienced a rapid uptake of IBR
in the past few years. This increase of energy generated from IBR has displaced synchronous
generators, resulting in fewer synchronous generators being routinely dispatched. With today’s
commonly-installed grid-following technology, most IBR require sufficient levels of system strength to
maintain stable operation, which are usually provided by synchronous generators and condensers. This
situation might change as technology innovates (such as increased prevalence of grid-forming
inverters), but such deployments are not foreseeable in the short term, and it is reasonable to expect
that the current technology will remain as it is for power system planning and operation purposes.
The full range of interactions between IBR, synchronous generators and wide area power systems are
more complex, and less widely known, than those pertaining to traditional power systems dominated by
synchronous generation technologies.
1.3.1.1 Situations prompting the need for electromagnetic transient models
The Australian Energy Market Operator (AEMO) has been using EMT simulation models for several
years, including for black start studies, sub-synchronous control interactions between series
compensated lines and IBRs, and stability analysis of one to two remote and radially connected IBRs
under low system strength conditions.
These applications shared two common features: the need to simulate a small part of the power system
under consideration, and a clear indication of previously known power system phenomena that had
occurred globally and were well understood in the international community. In Australia’s National
Electricity Market (NEM) power system, there has been an increasing number of incidents stemming
from unknown plant behaviour. Examples include:
▪ 2015: extended commutation failures of a line-commutated high-voltage DC (HVDC) link due to
interaction with the power system to which it was connected under remote rather than close-in
network faults and the impact of protection settings deployed in the HVDC link.
▪ 2016: protection settings applied to some IBRs which limited the number of network faults they
could ride through, despite network voltages and frequency recovering after each fault as they
occurred.
▪ 2017-18: insufficient synchronous machines online and excess system-wide IBR output, which led
to inadequate system strength under conditions where secure levels of inertia were available.
▪ 2019: sub-synchronous control interactions between multiple IBR without a series compensated
line or HVDC link.
In situations such as the above, it is not feasible to predetermine the extent of the power system that
must be modelled to produce an accurate outcome. Further, for the latter two instance, the phenomena
of interest and associated dominant frequencies are such that the use of root-mean square (PDT)
modelling is not suitable. Examples will be presented in Chapter 6.
There is very high penetration of IBR in some NEM regions (or areas within NEM regions), and new
generation connection applications are almost exclusively IBRs, of a range of fast-developing
technologies. The resulting increased potential for adverse interactions, together with the need to
understand and address unknowns before events occur in practice, are key factors contributing to
AEMO’s development and use of wide-area EMT models to the same extent it has used PDT models.
1.3.1.2 Development of an integrated EMT model
The investigation of the 2016 South Australia black system event prompted AEMO to begin developing
wide-area models of the NEM power system in 2017. By September 2019, AEMO had developed wide-
area EMT models for each of the five NEM regions, using learnings from the earlier South Australian
studies.
AEMO found that simulation studies required a wide-area model of one region along with a small part
of the adjacent region. The need to better understand suspect intra- and inter-area modes of oscillations
experienced in some EMT studies demonstrated the importance of developing an integrated EMT model
of the whole mainland NEM power system (excluding its smallest region, Tasmania, due to decoupling
via an HVDC link).

21
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

This integrated EMT model of the whole mainland NEM was completed in June 2020. It comprises
approximately 3,000 busbars and 200 detailed EMT dynamic models, and covers a total geographical
area of over 2 million square kilometers (0.78 million square miles). Advanced methods applied by
AEMO ensured that the increased computational burden is limited to a maximum of 20 percent
compared to a wide-area model of one region. This means that an EMT dynamic simulation run of this
system lasting 30 seconds can be completed in three hours.
The use of this model has demonstrated very high levels of accuracy when compared against measured
system responses, giving AEMO confidence in the veracity of studies for making decisions on what-if
scenarios.
The synchronous generators models were developed by AEMO’s in-house capabilities, based on
information provided by generator owners, currently stored in AEMO’s database, and data sought from
sites. The models of IBR were mostly manufacturer- and site-specific, provided by generator owners.
EMT models for a few legacy wind farms in the NEM were developed by AEMO using the OEM’s generic
models of similar technology type, and then validated against measured power system disturbances to
ensure these models will behave in the same way as the real equipment when used in EMT studies.
Before EMT studies can commence, a load flow case representing a specific power system operating
condition was developed in a PDT platform, and this load flow case is converted to AEMO’s designated
EMT tools. There are multiple software package providing such a capability, and in AEMO’s case, this
conversion process was facilitated by the same EMT study tool. After this conversion, the EMT model
is tested in steady state to ensure it can be initialised to the original load flow condition. For ongoing
studies where a significantly different power system operating condition needs to be examined, the
above process is repeated, otherwise the same EMT model will be used with minor modifications to suit
the need of each study case.
1.3.1.3 Expansion of operational as well as planning applications
Regulatory frameworks determined by the Australian Energy Market Commission in 2017 [R4], and a
range of guidelines and requirements developed by AEMO in 2018 [R5], have further expedited the
need for wide-area EMT modelling for different applications, by AEMO and other organizations.
Applications have included:
▪ Determining whether a new or modified generator connection would adversely impact system
strength, and assessing the veracity of different solutions if an adverse impact was identified.
▪ Calculating background levels of system strength and inertia determined by AEMO and maintained
by transmission network owners.
▪ Developing operational advice for real-time power system operators under system intact and
outage conditions, including when operating a normally interconnected power system as a
sustained island.
▪ Determining the system operability envelope under non-credible contingency events.
The growing application of EMT tools in the NEM power system has resulted in an increase of
approximately 60 percent in the number of Australian users compared to two years ago.
These EMT studies have been predominantly operationally focused. However, a number of mid- to long-
term planning applications have emerged recently requiring EMT dynamic analysis. These include:
▪ Determining the system security benefits of a proposed interconnector between two states where
intended benefits relate to phenomena which cannot be simulated by PDT modelling.
▪ Determining whether a large-scale power system can be operated without synchronous generators.
▪ Assessing the impact of transmission network asset retirements on available system strength for
already connected IBR.
▪ Planning for renewable energy zones.
▪ Designing system-wide special protection schemes.
These wide-area EMT models cannot be considered as a one-off model development exercise. Several
initiatives are currently being undertaken by AEMO to:
▪ Improve the total time taken to conduct simulation studies.
▪ Perform more extensive modelling of the distribution network to account for: (1) increased uptake
of MW range IBR in distribution networks, often in remote and sparse areas, (2) increased uptake
of kW-range distributed photovoltaics with inferior responses to those of MW-range IBRs, and (3)
changing load characteristics due to increased use of inverter-based loads.

22
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ Streamline production of any given wide-area EMT model from state estimator data.
In summary, with the increasingly higher penetration levels of IBRs, there is a consensus in Australia’s
NEM power system that large EMT system models are both necessary and practical, and will
increasingly become only more so.
1.3.2 Tasmania
Refer to Figure 1.3-1 which gives a simplified illustration of the Tasmanian electrical power system as
of October 2020. It is a moderately sized island system with an average load of approximately 1200
MW. Hydroelectric generation is the dominant energy source with wind energy becoming an ever more
significant contributor. Tasmania has been connected to mainland Australia via a 500 MW subsea
HVDC link (line-commutated converter) since 2006. The HVDC link can have very high bi-directional
flows, depending upon energy market conditions, but under normal hydro lake inflows it operates with
near neutral annual energy flows. The seldom used gas fired generation is effectively a back-up in case
of HVDC cable failure or drought. Due to its unusual generation profile, this power system regularly
operates with a very high penetration of asynchronous/inverter-based generation: a record 85 % was
sustained for several hours in August 2020. When cheaper energy resources are available, the hydro
generation backs off, saving its water, but continues to provide the inertia, fault level and capacity
reserves needed for secure power system operation. AEMO and TasNetworks currently study and
manage the power system with non-real time PDT and EMT simulation tools. PDT models remain
critical, e.g., in determining dispatch constraints to ensure that the system operates within the secure
technical envelope. While offline EMT integration studies are now mandatory for all new IBR projects
in Tasmania due to the ubiquitous low system strength conditions. EMT tools are particularly needed
to check for stable operation and interaction with other IBR, including the use of real-time simulators to
perform such studies.

Figure 1.3-1 Simplified layout of Tasmanian power system

23
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

1.3.2.1 Why Real Time Simulation (RTS)?

As part of the Australia’s power system evolution, Tasmania aims to double its renewable energy
generation, which will require a fourfold increase in HVDC interconnector capacity with the mainland -
from 500 MW to 2000 MW. This will be a massive change for the Tasmanian system and will
unquestionably require full “hardware-in-the-loop” testing of the control and protection systems of the
new HVDC Links.
1.3.2.2 What to Model?
HVDC systems were traditionally tested in real time but (by necessity) the model of the power system
was usually very simplified. Due to the Tasmanian system’s moderate size and uniquely variable
dispatch profile – dominated by hydro, wind and the HVDC Link a simplified representation poses too
high risks. Fortunately, the requirement for detailed modelling can now be met, at tolerable cost, with
modern RTS tools. Therefore, TasNetworks decided to model its complete transmission network (116
lines), all major generators (63) and all substation loads (55) on its RTS tool.
1.3.2.3 What Were the Challenges?
For the existing system, the core challenge was to reproduce a real-time power system model with the
same highly accurate performance as TasNetworks (TN) non-real time models (PDT and offline
EMT). The most difficult aspect of this was to faithfully replicate the hundreds of user-built governor
and exciter models. To achieve it, TN developed a software conversion program that populated its
RTS base case model from TN’s own PDT model data library files. In doing so, TN retained its single
(and already proven) source of truth. The existing windfarm models did not have complete block
diagrams and were therefore represented by tuned generic models. For the future power system
(circa 2028), the challenge was to estimate models for the new control systems of the HVDC links and
wind farms that could stably integrate these new IBR equipment into a power system with extremely
low inertia and fault levels. Recent publications on Virtual Synchronous Generation (VSG) give
templates for various future control arrangements. However, the strategy TN took was to use only
some capabilities of a full VSG, mainly:
▪ The new windfarms and HVDC Links were controlled as synchronised voltage sources.
▪ Synchronisation was still provided by traditional Phase-Locked Loops; however
▪ Overcurrent limitation was implemented with closed loop droop control (virtual resistance).
Providing that the existing minima for fault level and inertia were maintained this strategy proved
sufficient. Importantly, changing the control principle from current source injection to voltage source
generation reduced the system’s susceptibility to resonance issues.
1.3.2.4 What are the Expected Future Challenges?
It is already clear that EMT modelling of today’s power systems is essential for secure power system
management and RTS does give an important additional advantage of being able to validate the actual
control & protection systems (as well as their models). Clearly, it is impractical to use actual hardware
systems for more than just a small sample of equipment - most of the power system, in particular the
power electronic based systems, must still be represented by models. There are two main issues:
▪ For the Suppliers, being able to provide models that can run on multiple platforms.
▪ As with other EMT tools ensuring that the provided models are accurate.
All jurisdictions face these ever growing and often conflicting issues. The challenge for the power
industry is to determine a standardised modelling approach that allows accurate model usage across
platforms while also protecting supplier IP.
1.3.3 Belgium
ELIA is the TSO for the Belgian network, it owns and operates the transmission level from 36 kV to 380
kV, including the offshore 220 kV meshed network.
The Belgian transmission network is very meshed and has been historically well connected with the
neighboring countries. Its energy mix was comprised of mostly large controllable synchronous
generators connected to the highest level of the transmission system and located close to the load
centers in most of the cases.
This historical situation did not justify the development of systematic EMT simulations to assess the
system stability. PDT simulations have been and are the preferred tool from long term planning to
operational and near-to real-time system stability assessment.

24
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

In the last years, a fast evolution of the energy mix happened and brought some part of the system to a
remarkably high concentration of power electronics devices and moved the production centers farther
from the load. Large scale HVDC and offshore wind parks with high power density connected long AC
cable systems are constituting the majority of concentrated IBR. On top of these also distributed
resources, mostly based on power electronics (i.e. onshore wind and PV) is spread among the whole
Belgian transmission and distributions system.
This tendency is increasing with additional concentration of power electronics and reduction of classical
rotating units that impacts system strength. These developments are summarised in Figure 1.3-2.

Figure 1.3-2 Historical development of the Belgian transmission system

ELIA is moving towards development of the EMT capabilities to perform system-wide system stability
studies to correctly handle the risk. The priority is given to long term planning to try to reduce the risks
in an early stage and reduce the need, if possible, of detailed EMT simulations and assessment in
operational environment.
This effort goes in parallel via development of knowledge and of system-wide models allowing correct
simulation and understanding of the phenomena.
The highly meshed network topology and the concentration of power electronic devices from many
different vendors introduce significant challenges with regards to the definition of the system size and
the details to be included in the model together with the performance and numerical stability of the
simulation platform.

1.3.4 Ireland and Northern Ireland

The Ireland and Northern Ireland power system is a small synchronous system with an all-time peak
load of 6.9 GW and a maximum all-time wind output of 4.5 GW, which occurred in February 2021. At
present (2022), there are two HVDC interconnections between the island of Ireland and Great Britain
with a combined capacity of 1 GW, and there is no interconnection with continental Europe. There is
over 5.5 GW of wind capacity installed on the power system and there is approximately 10 GW of
dispatchable capacity, including the two HVDC interconnectors. This power system is regularly operated
with very high shares of wind generation with a record penetration of 147% of demand in Northern
Ireland (May 2021) and 96% of demand in Ireland (May 2020) [R6].
EirGrid and SONI, the Transmission System Operators in Ireland and Northern Ireland, have a multi-
year strategy focused on the transformation of the power system and electricity market in order to ensure
that renewable energy targets adopted under the European Union Clean Energy Package are met [R7].
The connection of up to 10 GW of additional renewable generation, new HVDC interconnection with
continental Europe and regular operation with non-synchronous penetrations of 95% (SNSP) are
required to meet the target of 80% renewable electricity by 2030. This ambitious journey will require a
holistic approach and, specially, a clear understanding of new forms of system instability driven by the
displacement of conventional plant with IBRs as well as the strength and limitations of the analysis tools
used to predict them.

25
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

1.3.4.1 Experience with EMT studies

Detailed EMT studies have been regularly performed over the years in the Irish power system to analyse
specific known phenomena of concern, including switching and lightning transients, ferroresonance,
resonance issues related to shunt compensation, assessment of black-start plans, post-fault analysis
investigations, TRV in circuit breakers, etc. These studies share a common feature of previous
knowledge and understanding of the phenomena to be investigated and a requirement for limited extent
of network modelling (albeit with a high level of detail and complexity).
Over the past few years, the installation of long cables associated with new renewable generation
projects or with expansion/reinforcement of the transmission grid, has driven the need to perform many
detailed EMT analysis to investigate and mitigate risks of harmonic resonances and Temporary Over-
Voltages (TOV) caused by transformer energisation or fault clearance ([R8],[R9]). These studies require
a larger portion of the network to be represented in detail, but still limited to regional models, and the
scope is restricted to assessment of overvoltage stress in equipment and energy dissipation in surge
arresters.
A comprehensive EMT study was recently conducted to investigate risks of sub-synchronous
interactions driven by the planned installation of series compensation in three existing 400 kV circuits.
This study required very detailed and accurate modelling of a number of generators, their controls and
the mechanical components of the shafts. Screening studies in the frequency domain were performed
to identify critical topologies and to limit the extent of the network and power electronic devices
required to be included in the EMT model. Frequency-Dependent Network Equivalent (FDNE)
models were used outside the main area of study. The initialisation of the EMT model was
validated against steady state load flow and short-circuit analysis performed with standard planning
models.
1.3.4.2 Next Steps
System-wide stability analyses in planning and operational timeframes are currently performed using
PDT positive sequence models only. This includes on-line dynamic security assessment in the Control
Rooms as a decision support tool to ensure the stability of real-time operation as well as near-time
market schedules (look-ahead).
To date, no incidents related to converter stability or interaction of power electronic controls have been
observed in the All Island power system, suggesting that the adopted operational metrics, tools and
models have been adequate for the levels of IBRs and system strength experienced so far. However
medium to long term studies with increased share of non-synchronous generation have shown areas
of concern that require further investigation.
The System Operators (EirGrid and SONI) are mindful of the limitations of PDT tools and models to
capture the behaviour of fast control loops and their importance in weak system conditions. However,
EMT simulations are very complex and time consuming and it is not always practical to perform
system-wide studies with this tool, especially in the real-time operational timeframe. Work is
progressing in a number of areas to bridge the gap between the required level of modelling and
the practicalities of analysis:

▪ Improving understanding of new form of stability challenges and the relationship between high
shares of IBRs in the generation mix and the underlying system strength.
▪ Review of the adequacy of traditional tools and models, such as PDT, to ensure integration of
higher levels of renewable generation in a safe a secure manner.
▪ Development of accurate and robust metrics and screening methodology to assess the risk of
converter driven instabilities.
▪ Specification of model and data requirements, including EMT, applicable to new connections and
to existing plant undergoing material changes.
▪ Development of EMT modelling and analysis capabilities for system-wide stability assessments to
investigate selected scenarios, as a last resort, and to support operational decisions in a timely
manner.
1.3.5 Denmark
The Danish power system is a small system divided across two synchronous areas: The Continental
European synchronous area, and the Nordic synchronous area. The Danish power system is often
operated at high levels of IBR, and frequently the penetration of IBR exceeds the total load in Denmark.
However, secure operation of the Danish power system is directly dependent on a strong
interconnection to the neighboring countries. Besides HVDC LCC and HVDC VSC interconnector

26
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

capacity, multiple parallel 400 kV AC interconnections to Germany and Sweden in combination with a
highly meshed transmission grid, contributes to the short-circuit strength of the Danish power system
being kept relatively high. Although, with the increasing IBR penetration, and the consequential
decommissioning of large centralized synchronous machines in Denmark and the neighboring AC-
interconnected countries, the system inertia and the short-circuit strength is facing a significant decrease
in the near future. The last large synchronous machine in Denmark is expected to be decommissioned
before year 2030.
1.3.5.1 Experience with EMT studies
At the time of writing EMT analysis has not yet been applied to study the system wide stability of the
Danish power system. In Denmark EMT analysis has typically been applied for classical EMT purposes
such as lightning and switching-transients for insulation coordination and power system design. For
now, phasor-based time domain analysis (PDT) is the primary tool applied for general stability
assessment purposes on the full system level. Although, EMT analysis is generally applied in project
specific cases for local stability analysis. In these cases, the boundaries around the studied area is
represented by Thevenin equivalent sources, considering different levels of short circuit power in order
to investigate the likelihood of having oscillation relevant events e.g. control interaction. In addition to
this, EMT analysis has been applied to perform black start related studies to emulate the sequence of
the energization process and soft start functionalities.
However, partly motivated by a significant undergrounding of the transmission grid and partly by the
increasing penetration of IBR, the necessity for EMT modelling and analysis is becoming evident.
In 2020, the Danish TSO Energinet commissioned the Kriegers Flak Combined-Grid-Solution, which is
a hybrid interconnector with a total of 936 MW offshore wind power divided between four wind power
plants together with a 400 MW HVDC VSC back-to-back interconnector between Denmark and
Germany. An offline EMT model was developed for the project, which quickly proved useful to study
several stability related issues, such as control interactions. In several cases the PDT model of the
Kriegers Flak system, using vendor-specific PDT models of the inverters, was unable to reproduce the
phenomena of interest. Identification of stability risks by use of the full EMT model of the Kriegers Flak
project led to control tuning of both HVDC VSC and wind power plants.
1.3.5.2 Recent incidents and next steps
In 2019 and 2020 the Danish power system experienced various severe incidents. In one case, a fault
in a substation led to a partial trip of a distant wind power plant, which in combination with the trip of a
HVDC LCC converter and a central power plant led to a local unbalance of more than 900 MW. In
another case, a solid fault in a substation without bus-bar protection led to a three-phase fault for more
than 450 ms, which led to a system wide voltage suppression. In all cases the Danish power system
survived, but it is deemed that the stability of the system was near a limit. Common for the two mentioned
cases is that the post-incident analysis has shown that there is a poor match between the measured
values of the real incident and the simulated response of the full system PDT model due to reduced
order models and insufficient details of protection / trip functions implemented in PDT models. Motivated
by the recent incidents, and the prospect of increasing IBR and cable penetration, Danish TSO Energinet
is investigating the opportunities of utilizing EMT modelling for large-system stability analysis and impact
assessment.
1.3.6 France
RTE uses EMT simulation tools to perform offline time domain studies and frequency scan studies.
Engineers conduct offline EMT simulations at different stages of a project. The following stages require
study for projects involving power electronics equipment connected to the French transmission grid
[R10].
▪ At the planning and specification stage, identify potential technical issues and assess solutions.
This stage uses a generic model of the new installation in a reduced grid model and equivalent.
For example, generic HVDC/FACTS model (including control and protection (C&P) representation)
may be initially used for the new installation. Engineers can then begin developing and tuning
settings based on experience and functional specifications.
▪ At the design stage, evaluate solutions proposed by manufacturers. At this stage, studies can use
preliminary models provided by manufacturers in addition to available generic models.
Manufacturer models are usually black boxed due to Intellectual Property (IP) issues.
▪ At the testing / commissioning stage, verify the system performance with the device model to
prevent potential adverse interactions with the existing grid.

27
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ When the manufacturer’s offline models of power electronic devices are sufficiently accurate, as
shown by agreement with the testing /commissioning stage, offline simulation studies may be used
to simulate the final design and during the commercial operation of the project. However, because
RTE acquires replica controllers, most system studies use real time simulation using HIL to ensure
high accuracy and reliability of results.
Real-time EMT simulation using control and protection replicas is also possible for each stage
depending on the context, but it is mainly restricted to analysis during the testing / commissioning stage
and commercial operation.
1.3.6.1 Creation and maintenance of large networks
The user determines the extent of the surrounding network to model explicitly for EMT or frequency
domain simulations. Grid models for EMT simulations are usually restricted to a few substations beyond
the point of interest. However, due to the growing complexity of transmission grids with complex control
and protection systems, studies use larger and more detailed network models. This is especially true
for frequency scan studies. The use of a large-scale model around the study area improves the accuracy
of the results. However, building such large-scale models is time consuming and error prone, and also
requires accounting for the various network topologies and generation/consumption scenarios in
network operation. This is why RTE developed an automated interface to import network data from its
System Control and Data Acquisition (SCADA) tool. This tool uses a Common Information Model (CIM)
description-based interface.
Several years ago, RTE identified the need for a platform that would gather network data and for tools
that would simulate steady-state conditions. This platform models the entire RTE network from 400 kV
to 63 kV, including sequence impedances of lines (positive, negative, and zero-sequence), transformer
parameters (impedances, tap changer positions), generators (sequence impedances, voltage, and
reference power settings), loads, FACTS parameters, and substation configurations.
The CIM files export module of the tool is compliant with the Common Grid Model Exchange Standard
(CGMES). Figure 1.3-3 shows the flow diagram of the tool to the EMT program interface.

Figure 1.3-3 SCADA to EMT interface flow diagram [R10]

The entire 400 kV and 225 kV network model is comprised of 1280 substations, 2049 lines, 244
autotransformers, 12 phase-shifting transformers, 841 synchronous machines, 2007 loads, 115 shunt
capacitors and reactors, and 7 SVCs.
Load flow comparisons between the tool and a commercially available EMT programs validate this EMT
model. Results show a good accuracy within the specified 5% relative error bound. The main differences
result from slight imbalances on AC overhead lines and transformer representations. The studies
conducted at RTE demonstrated acceptable computing times for load-flow solutions, along with
electromechanical and electromagnetic transient simulations.
The creation and maintenance of large networks in a single Graphical User Interface (GUI) environment
with the capability to perform unbalanced and multiphase load-flow solutions, and to simulate both

28
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

electromechanical and electromagnetic transients, offers major advantages to RTE users. It provides a
unified and validated data set with a very high level of accuracy; it allows users to extract data for various
applications and to perform aggregations when necessary at different locations in a system. When
network reduction is performed, depending on the considered study, the omitted network is represented
either by Thevenin equivalent sources or by a frequency dependent network equivalent (FDNE).
1.3.6.2 Modelling of power electronics bases systems and related controls
Several voltage source converter (VSC) based HVDC and FACTS devices are in service, in the planning
or in construction phases, and most of them use new technologies such as Modular Multilevel
Converters (MMC). The MMC based HVDC link between France and Spain (project INELFE: France-
Spain ELectrical INterconnection) is, to date, the highest rated VSC link at 2 × 1,000 MW. Avoidance of
possible adverse interactions between controls for HVDC systems and FACTS devices requires many
EMT studies using detailed system models.
Advanced expertise in modelling and simulation of VSC based systems has a crucial impact on the
planning and delivering of numerous HVDC and FACTS projects, as is the case for RTE. Moreover, the
long-term maintenance of HVDC and FACTS installations operated by RTE is under its responsibility.
1.3.6.3 Detailed modelling of power plants
EMT simulation studies are also conducted by EDF with focus on electrical systems of power plants. A
complete database of detailed low frequency models has been developed and the maintained
configurations include a representation of the main generator, the control system, transformers and the
mechanical behaviour of the shaft with relevant parameters [R11]. The power plant models are used to
study ferro-resonance, faults and lightning transients with appropriate surrounding network model. All
models are validated through PDT models, site tests and measurements.
1.3.7 Quebec, Canada
1.3.7.1 Background
At Hydro-Quebec (TransÉnergie) the level of modelling in EMT studies is continuously escalating
since the addition of a massive series compensation on the 735 kV system in the 1990’s. The
inclusion of machine voltage regulation controls allows studying 60Hz over voltages in conjunction with
automatic system separation controls [R12]. It is also required for various studies, such as, motor start-
up, network islanding, single pole reclosing analysis, statistical studies, harmonic analysis for industrial
installations and research activities related to power swing detectors.
The creation and maintenance of a large utility network in a single Graphical User Interface (GUI)
environment with load-flow, stability and EMT data has many advantages. It provides a unique and
validated data set, allows data extraction and aggregation for various applications and allows performing
different types of analysis from different locations. The first experiments and results on the simulation of
an extra-large network (Hydro-Quebec grid from 735 kV to 25 kV loads, including synchronous machines
at 13.8 kV) in an EMT software [R13] have been presented in [R14]. Since the presentation of this paper
computing times/methods have evolved significantly and continue to improve [R15]. It is now even more
feasible to simulate transmission grids with more than 30,000 electrical nodes and more than 20,000
control block signals. Nonlinearities can be represented as well. In fact, modern software codes do not
impose limitations on the size of simulated grids. Very largescale distribution grids can be also simulated
efficiently [R16] [R17]. Tests with more than 300,000 nodes have been successful.
1.3.7.2 Network models maintained
Two main network models are maintained at Hydro-Quebec. Other networks can be extracted according
to specific needs. The first extra-large network for EMT studies represents the complete (100%) Hydro-
Québec power grid. The second smaller network is a reduced version of the first network focusing mainly
on power plants, 735-315 kV transmission lines and major loads. It does not represent medium and low
voltage transmission lines and regroups some loads into large centers. These networks are continuously
updated at Hydro-Quebec for performing EMT studies.
Figure 1.3-4 provides a top view to the complete Hydro-Quebec grid.

29
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Ontario

Vermont

New
New York New
Hampshire
Brunswick

Massachusetts

Figure 1.3-4 Hydro-Quebec grid for EMT studies: top view

1.3.7.2.1 L-Network
The first network, named hereinafter L-Network (very large), constitutes a reference for obtaining
network equivalents for various study purposes. It is also the previously discussed unified environment
for maintaining and extracting data for various applications.
The L-network is organized using a multilevel hierarchical design structured on 6 pages in the GUI.
There are a total of 94,706 physical devices and 93,234 signals. The list of physical devices includes
42474 power devices and 52232 control block diagram devices. The signal count adds 29,797 power
nodes to 63,437 control system signals. The top-level listing (subnetwork contents are not counted) of
main devices is:
▪ 1,259 transmission lines representing the existing 1,560 lines and derivations.
▪ 2,098 three-phase transformers
▪ 916 load models representing a total of 40.5 GW. All medium and high voltage shunt capacitors
and inductors were modeled separately. Some loads were modeled with transformers and shunt
capacitors at the lower voltage level.
▪ 11 SVC (Static Var Compensator) models of 300 Mvar and 600 Mvar. The SVCs have been
combined on some buses by creating 600 Mvar models. An average value model is used by default.
▪ 9 HVDC models to export 5,300 MW out of Quebec. Average value models are used by default.
▪ 32 series-capacitor MOVs
▪ 213 nonlinear inductances used for high voltage power transformer saturation representation
▪ 349 synchronous machines (SM) with associated controls representing more than 49 power
stations and four synchronous compensators. All hydraulic generators are modeled with a single-
mass mechanical part.
The transmission lines modeled using the constant parameter model (CP) include propagation delay.
The more advanced frequency-dependent model layer can be selected for specific higher frequency
content studies. The list of synchronous machines is augmented with 20 more machines related to

30
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

distributed generation. In the context of dynamic stability studies, it is not recommended to use even a
single fixed voltage source with internal impedance since such a source generates fictitious reactive
power during voltage swings and erroneously accentuates voltage excursions.
The equivalent of the above L-network is available in a phasor-domain package [R18] (PDT). Although
simplified, it serves to verify the overall lower frequency electromechanical behaviour of the grid.
1.3.7.2.2 R-Network
The reduced second network, named hereinafter R-Network, was constructed as an alternative to the
first L-network for achieving reduced computer timings in simulations. Most of the UHV studies can be
conducted using this reduced network version. It represents the entire 735 kV system and includes the
main parts of 315 kV to 120 kV systems. Transmission lines are combined when necessary and
equivalent load devices are used for lower voltage derivations for preserving the power-flow conditions.
A particular effort is made to model shunt capacitors at the 120, 230 and 315 kV levels in order to obtain
the most faithful frequency response compared to the complete network model. It has been found that
this approach provides good results with minimal effort when the study zone is electrically sufficiently
far from the location of the equivalent.
The reduced network has a total of 28,000 physical devices including control diagram blocks. There are
6,200 power devices and 3,400 power nodes. The listing of top-level devices is:
▪ 170 lines, with 75 lines at the 735 kV level, 53 at 315 kV, 23 at 230 kV and 19 at 120 kV
▪ 90 three-phase transformers
▪ 128 load models, for a total of 35,000 MW
▪ 11 SVC models
▪ 9 HVDC models to export 5,300MW out of Québec.
▪ 57 synchronous machines with AVRs for a total of 37,000 MW of generation.
1.3.7.3 Applications
EMT models are used on a daily basis by Hydro-Québec for design and planning studies. The list of
applications includes:
▪ Information database for detailed simulations for extraction for various types of studies. Equivalents
can be derived at any node: frequency-dependent network equivalents or 60Hz.
▪ Maintenance of network data, topology and loading conditions.
▪ Geomagnetic disturbance studies: accurate solution with harmonics, transformer models.
▪ Multi-terminal HVDC transmission system study.
▪ Transient stability studies with a detailed network representation: verification of phasor-domain
solutions, auto-excitation of synchronous generators, single-phase and two-phase fault studies,
overvoltage and protection, unbalanced grid, sub synchronous resonance conditions in series
compensated network, load modeling (including exponential model for transient stability studies).
▪ Network separation studies: overvoltages due to long lines, study of protection systems, sacrificial
arresters.
▪ Control of TOV magnitudes and duration following a system separation in 735 kV grid James Bay
system. Arresters are used to limit the TOV to 1.6 pu during system separation. The arresters are
switched on for a short period of time during system disturbances by local power swing detection,
remote detection of over-frequency or open-corridor detection.
▪ Network Frequency response: equivalents for interconnecting wind parks, equivalents for
interconnecting industrial clients.
▪ Integration of renewable energies: usage of network equivalents, wind park controls from different
manufacturers.
▪ Power quality problems.
▪ Series compensation: arrester energy absorption, impact on power system dynamic performance,
short-circuit studies, fault clearing, delayed current-zero conditions during fault clearing.
The majority of the above studies cannot be conducted sufficiently accurately with PDT tools.
1.3.8 Texas
ERCOT first utilized an EMT tool for wide-area stability analysis in 2015. The primary concerns were
the ability of positive sequence fundamental frequency PDT simulations and models to accurately reflect

31
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

system response under conditions of low system strength in the Panhandle region of the ERCOT
system. As the system has evolved, ERCOT has continued to re-evaluate the Panhandle region with
EMT analysis every year or two. ERCOT has also utilized wide-area EMT analysis for other areas
where low system strength could be a concern, such as the South Texas portion of the ERCOT system.
Defining a system strength metric to determine exactly when an EMT analysis is necessary remains a
significant challenge [R19].
The Panhandle region of the ERCOT system includes approximately ten 345 kV buses with a meshed
345 kV network. In 2015, this portion of the ERCOT system was far away from any load centres and
any synchronous generators. Projected wind generation interconnections in the Panhandle region
exceeded 5,000 MW. ERCOT EMT studies confirmed that maintaining a weighted short circuit ratio
(WSCR) of 1.5 based on wind generation output in the Panhandle was adequate to maintain reliable
operations [R20]. It was also found that EMT simulation results generally were consistent with positive
sequence fundamental frequency simulation results when the WSCR was at least 1.5 [R20][R21][R22].
EMT analysis of the South Texas portion of the ERCOT system determined that application of the WSCR
metric was not appropriate for that area. Since the WSCR calculation is based on relatively simplistic
assumptions pertaining to system strength provided by synchronous machines, it does not consider
potentially beneficial impacts provided by local load and voltage support provided by SVCs or
STATCOMs or even other IBR that may be located on key transfer paths. Additionally, determining
adequate boundaries for WSCR calculations was difficult because the wind generation is dispersed
across a larger area of the network compared to the Panhandle. Thus, it was concluded that the
Panhandle presented a unique situation (a coherent wind generation cluster far away from load and
other generators) where application of the WSCR metric was appropriate.
Recent developments affecting the Panhandle region include the interconnection of the Lubbock Power
and Light (LP&L) system to the ERCOT transmission grid and significant generation development just
beyond the Panhandle region in what is referred to as the Nearby Panhandle region. The integration of
LP&L will provide a local load sink and an additional transmission export path for the Panhandle region.
The interconnection of wind generation in the Nearby Panhandle region increases overall exports from
the wider area, but also provides voltage support along primary export paths. It is difficult to say if these
developments improve system strength, but they make application of the WSCR metric [R23] less
suitable to make that determination. In other words, EMT evaluations may be become more important
and the areas where detailed EMT modelling is necessary may be expanding. This presents challenges
for computational capabilities as well as additional burdens on resource owners who may need to
develop and submit detailed EMT models for their facilities.
1.3.9 UK
National Grid Electrical Transmission (NGET) uses EMT tool and offline simulation for different type of
system studies including SSTI, Control Interaction and Dynamic Performance Studies. Potential for
interactions between a planned new connection of controlled based devices and other existing devices
in the system can be evaluated through EMT system studies. The first step for these studies is to identify
a study area and derive an equivalent model which includes retained study area.
Recently in NGET, north and south of the NGET grid has been modelled in detail in time domain for
conducting two different set of studies for installation of three STATCOMs in the south and deployment
of Power-Flow control devices (Series Compensation FACTS devices) in three different locations in
north. The defined study areas for both studies were derived from full GB system in PDT and converted
into time domain model in an EMT tool. All the components and devices including generators and their
controllers, lines, transformers and power electronic devices are modelled in detail. Regarding the power
electronic based devices, the preference was to use vendor provided models in the studies. An EMT
model was developed and validated to meet the technical requirement in case if vendor model for a
power electronic based device was not available.
All the power flows and voltages in steady state are validated against the reference model. In terms of
dynamics validation, different approaches are taken in order to makes sure the responses received from
the controlled based device matches with the reference model. A large number of substations and lines
are included in the study area and they are modelled in detail in the EMT tool. In the EMT model, short
circuit level and impedance are checked in all substations to make sure that the responses to the faults
are valid.
The next step after the validation of the EMT model is to identify the list of studies. As the model in both
studies include a high number of substations and lines, an exhaustive list of credible contingencies can
be studied. In all the contingencies including those with and without applied faults, dynamic performance

32
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

of the controlled based devices is simulated using offline simulation. The results from the EMT
simulations demonstrate that how the devices in the grid reacts to the continuous dynamical changes
and supporting stability of the system.
1.3.10 Brazil
1.3.10.1 Current applications
In Brazil, one of the main duties of the National Electric System Operator - ONS is to undertake power
system studies, aiming to quantify the impacts on the National Interconnected System - SIN and its
agents, in order to guarantee the security of continuous supply throughout the country.
The development, implementation and delivery of the models of wind and photovoltaic parks for
conducting studies of electromagnetic transients, are part of the process of integrating new generation
facilities into the electrical system. It should be noted that when delivering the model to ONS, it may be
made available to any agent in the electricity sector, enabling the representation of this undertaking in
other studies of electromagnetic transients.
For a wind or photovoltaic park to be declared fit for operation, among the various requirements required
by ONS, it is necessary that the entire plant of the park (internal network), wind turbines or photovoltaic
panels (cell and converter) and elevator and connection transformers, be represented in all simulation
programs used by ONS, specifically for electromagnetic transients, considering the information and
models sent by the proprietary agents of the equipment that make up the operation network installations.
The analyses are made considering the nearby wind and / or photovoltaic parks represented and with
maximum (nominal) generation. If it is not possible to simulate these plants with maximum generation,
given the difficulty in representing the parks in aggregate, consider the representation of the plants
present close to the event site, as a source of voltage or current, in series with the equivalent of short-
circuit.
The analysis is carried out considering the minimum load condition for the project's integration period.
If possible, the maximum permissible voltage limits on the bus should be used, as long as these limits
are not violated in the other buses of the system.
The network equivalents (boundaries in the representation in EMT model) are calculated usually
accounting for fundamental frequency behaviour of the system, and the allocation of these will consider
the distance of two buses where the contingency is analysed.
1.3.10.2 What are the challenges?
The characteristic of solar and wind generation in Brazil results in a high IBR concentration in specific
parts of the country. Wind generation is characterized by a large representation in the Northeast (about
85%) and in the South (about 15%) in Brazil. The other regions have practically no significant share of
generation. Regarding solar generation, the area with the best energy capture is in a region between
the northeast and southeast of the country.
In line with these characteristics, some local particularities present challenges both from the point of
view of studies of electromechanical transients and studies of electromagnetic transients. Below we
present the biggest challenges faced and which are still present in the integration of wind and solar
parks in the Brazilian transmission system.
1.3.10.2.1 Inertia
The value of wind generation today is greater than 15 GW (representing 24% of the country's load).
Considering the geographic, seasonal and type of generation differences in Brazil, it appears that the
resource of energy transfer from one region to another is widely used to service the load. However,
considering the large amount of renewable generation using frequency converters, both in the
transmission network and in the distribution network, the accurate estimation of the system's inertia
becomes essential to guarantee a safe operation.
1.3.10.2.2 Network Representation
The characteristic of installing wind farms, usually focuses on very specific regions. In some situations,
SVC is present in these parks and in very particular cases, more than one SVC is connected by short
transmission lines, where there may be less than 20 km between SVCs.

33
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

1.3.10.2.3 System Features

Historically and currently, a major difficulty in integrating new renewable sources is the mismatch
between generation and reinforcement in transmission. With the development of technologies and
logistics, wind and solar parks are built and are able to energise very quickly. When there is a large
concentration of wind farms, there is a delay in the start-up of transmission lines and other equipment
to transport this generation. This ends up leading to problems in operation and in some cases even
restriction of generation in order to limit shipments of equipment and ensure systemic safety.
Furthermore, delay in the transmission network means that it is not possible to maximize the transfer of
energy between regions, thus reducing associated costs.

34
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

2. The role of EMT and PDT dynamic simulation for

wide-area system studies
2.1 Fundamental principles
The focus of this document is on Electromagnetic Transient (EMT) modeling. To date, and still in many
regions with extremely large systems, such as North America and continental Europe, the vast majority
of system stability simulations are performed in positive-sequence phasor-domain transient simulation
tools. Phasor-domain transient simulation tools are sometimes referred to by entities as RMS simulation
tools or simply as transient stability simulation tools. In this document, however, we adopt the term
phasor-domain transient (PDT). The reason is that the term RMS refers to root-mean square and thus
true RMS analysis can only truly be performed on symmetrical three-phase waveforms, which may
include harmonics etc. Thus, we believe that PDT is a more appropriate term.
The fundamental difference between EMT type simulation and PDT type simulation is in the way circuit
elements are mathematically represented to form the network admittance matrix. Equation 1 and
Equation 3 show the representation of the three basic circuit elements in EMT simulations while
Equation 2 and Equation 4 show the representation of the three basic circuit elements in PDT
simulations.
Circuit equations for an R-L circuit
𝑑
𝑣(𝑡) = 𝑅𝑖(𝑡) + 𝐿 𝑖(𝑡) Equation 1
𝑑𝑡
𝑉(⍵) = 𝑅𝐼(⍵) + 𝑗(𝐿 ⍵)𝐼(⍵) Equation 2

i(t) R L

v(t)

Figure 2.1-1 Series R-L circuit

Circuit equations for R-C circuit
1
𝑣(𝑡) = 𝑅𝑖(𝑡) + ∫ 𝑖(𝑡) 𝑑𝑡 Equation 3
𝐶
𝐼
𝑉(⍵) = 𝑅𝐼(⍵) + Equation 4
𝑗(⍵ 𝐶)
i(t) R C

v(t)

Figure 2.1-2 Series R-C circuit

It is apparent that in PDT representation, the differential equations that describe the dynamic behaviour
of an electric circuit (Equations (1) and (3)) are approximated by algebraic equations (Equations (2) and
(4)). It should also be noted that in the PDT implementation, the network elements are represented by
the corresponding impedance at the system fundamental frequency. Thus, network resonance
conditions (other than fundamental frequency) are inherently absent in the PDT formulation.
EMT network equation formulation (Equation 5 to Equation 8) and the circuit response are further
illustrated through the simple circuit example below. In the results shown below in Figure 2.1-4, a switch
(not shown) connecting resistance (R2) parallel to the capacitance is closed at time instant t = 1.00 s
and opened at time instant t = 1.1 s. The following important points should be noted.
▪ Even in steady state, the solution provides the instantaneous values of voltages and currents
▪ Harmonics in the waveforms are captured
▪ DC offset in the waveforms (specially the current in this example) are captured
▪ Damping (decay) of transients are accurately captured

35
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

1 2
R1 L1 I1 I3
I2

R2
E C

Figure 2.1-3 Example circuit

𝑑𝐼1
𝐸 = 𝐼1 𝑅1 + 𝐿1 + 𝑉2 Equation 5
𝑑𝑡
𝐼
𝑉(⍵) = 𝑅𝐼(⍵) + Equation 6
𝑗(⍵ 𝐶)
𝑑𝑉2
𝐼3 = 𝐶 Equation 7
𝑑𝑡
𝐼1 = 𝐼2 + 𝐼3 Equation 8

Figure 2.1-4 Response of an electric circuit simulated on an EMT platform – DC offsets, harmonics are
captured in the form of instantaneous time domain response
As described, the EMT solution provided the instantaneous values of voltages and currents. PDT
simulations do not capture the instantaneous values of voltages and currents. This is a key difference
in view of simulations to verify the response of power electronic interfaced devices.
All of the above points are important when analyzing the response of power electronic interfaced
devices, where instantaneous voltage and current waveforms form inputs to control systems and
dictate the operation of power electronic components.
2.1.1 Examples
Specific illustrative examples of the need for EMT modelling are described below for the benefit of the
reader.

36
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

2.1.1.1 Impact of DC offset on inverter operation:

Figure 2.1-5 shows the voltage waveform appearing at an ac system location close to a series
compensated double circuit line. In this example, the presence of a DC offset on the bus voltage is
illustrated.

Figure 2.1-5 DC content in the bus voltage - Typical situation when clearing faults near series
compensated lines

Figure 2.1-6 Schematic representation of a two level VSC converter

Consider the AC voltage in Figure 2.1-5 being the terminal voltage at an IBR. A typical arrangement of
a wind turbine with DC choppers is shown in Figure 2.1-6. In order for the VSC to function as designed,
the DC link voltage has to be kept at a value greater than the peak ac voltage at the VSC terminal. This
condition is satisfied under normal operation, however, during the transient shown in Figure 2.1-5, this
requirement can be violated. This DC offset can cause the diodes of the inverter Figure 2.1-6 to conduct.
The DC link capacitor will thus get charged, raising the DC link voltage. The DC link chopper is designed
to dissipate energy from the DC circuit and to maintain the DC link voltage within design ratings. The
DC chopper will be activated by the inverter protection systems to limit the voltage rise. If the DC offset
of the system does not decay fast enough (mostly due to damping from the AC system), the chopper
may reach its thermal limits and a trip signal to disconnect the WTG will be issued. This is an example
of an event that can only be captured in an EMT platform as PDT simulations do not capture power
system (network) dynamics such as DC offsets and the damping of such transients.
2.1.1.2 Phase Locked Loop (PLL) response:
The instantaneous angle of the inverter bus voltage is a critical input to the protection and control system
of a VSC. This angle is commonly derived using a PLL. The instantaneous voltage is an input to the
PLL. In the illustrative example in Figure 2.1-7, it is seen that the PLL is able to track a frequency change
using the voltage waveform as its reference.

37
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

-
Figure 2.1-7 Illustration of PLL response – The PLL is expected to lock to the grid frequency and provide
an accurate estimate of the instantaneous ‘angle’ of the voltage (and current)
However, depending on the PLL that has been implemented and the nature of the disturbance, large,
rapid changes to the voltage waveform may cause the PLL to lose tracking and cause unexpected plant
behavior.
Figure 2.1-8 shows a response of a power system and a wind turbine following a fault and fault clearance
near the windfarm. The PLL is expected to estimate the instantaneous angle variation of the fundamental
frequency component of such transient waveforms (see Figure 2.1-7).
The performance of the PLL can only be accurately established through EMT simulations where its
principal input measurement quantity, the voltage waveform, is available. The performance of the PLL
is more critical under ‘weak grid’ connections (and where multiple IBR are connected in close vicinity),
thus, adding to the importance of EMT studies when considering wind farm connections under
challenging grid conditions.
The example in Figure 2.1-8 shows that for the given disturbance, the wind turbine recovers to a stable
state. However, the power reversal observed soon after the fault is cleared is a result of the PLL re-
locking delay. The turbine has also recovered to a substantially reduced output power compared to its
pre-fault value, with (en-masse) implications for power system balancing.

38
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.1-8 Typical voltage and current waveforms experienced at the connection point of an IBR
following a system fault

2.1.2 Summary
In summary, EMT simulations provide benefits due to the following key factors.
▪ EMT captures the instantaneous values of voltages and currents.
▪ EMT simulations capture the network dynamics including DC offsets and harmonics and represent
accurately the damping of such transients.
▪ Key control system components such as PLL, fast acting control loops and nonlinear control loops
can be accurately represented in EMT platforms.
▪ Nonlinear effects phenomenon of power system equipment such as transformer saturation can be
readily represented in EMT platforms if such phenomena is deemed to have an impact on inverter
response.

2.2 Level of modelling details

2.2.1 IBR control systems
Several areas of the world are beginning to experience issues associated with a high penetration of IBR.
Key contributing factors include reduction of synchronous generators online, large electrical distances
between areas of concentrated IBR and large synchronous generators as well as the increasing share
of IBR compared to the local demand. Depending on the extent of interconnection, these issues could
be characterised as either system strength or inertia challenges.
For interconnected power systems, lack of system strength is often the limiting factor as opposed to lack
of inertia. Simplified metrics for quantifying low system strength conditions are presented in as well as
technical issues that can be expected in such networks. The key issue can be thought of as follows.
In a weak grid, the equivalent Thevenin impedance looking into the grid at any node becomes very
large. Thus, the smallest change of injected current (power) at a node can lead to a large change in
voltage. Thus, when current-regulated IBR are introduced into a weak grid this can lead to difficulties
with voltage control, due to reduced gain margins in closed loop voltage control, and the potential for
control interactions among nearby active devices with fast voltage control loops.

39
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

As system strength decreases, control loops of IBR could become unstable. They can therefore
experience control instabilities even without any other IBR nearby. The impact of PLL parameters on
the small-signal behaviour of voltage source converters is discussed in [R22]. To operate under low
system strength conditions, not only PLL parameters may need to change, but also other controller
parameters may need to change. This is not limited to converter controllers, but also applies to plant
controllers. A controller that is requested to be tuned acting fast to disturbances is actually not beneficial
for weak systems. It can easily introduce undesired consequences in weak systems. In some
circumstances stable operation may not be feasible without significant control modifications or may also
need additional equipment such as synchronous condensers under worst case scenarios. Furthermore,
operation of multiple IBR in close proximity could give rise to adverse control interactions and
instabilities.
Modelling of the power system, plant and controllers involves the construction of mathematical solution
of a physical system and performing numerical simulation. Simulation consists of inevitable numerical
error. It is pertinent to consider and minimize this error to achieve results and interpretation, which
gives a better overview. As mentioned previously, due to the large integration of IBR, PDT
simulations are no longer sufficient. EMT modelling may be required when studying the impact of IBR
under low system strength conditions, where the local AC voltage amplitudes and phase
displacements have a higher sensitivity to small changes. For example, IBR often rely on a PLL to
maintain synchronism between their injected currents and local network voltage and studying the
accuracy and stability of this PLL response requires EMT simulations especially under low system
strength conditions, since PDT models are inherently unable to represent such key components.
The response of a typical all pass filter PLL to a network phase angle jump and an unbalanced fault as
obtained in PDT and EMT tools is shown in Figure 2.2-1 demonstrating the deficiencies of the PLL
model in the PDT tool. PDT representation of other control loops such as DC link current and voltage
controllers can exhibit similar inaccuracies. Without representation of such components, PDT models of
IBR may fail to show a possible control instability and hence, in such cases could yield results that would
likely lead to inaccurate conclusions being drawn. As the results of these types of dynamic studies feed
into investment decisions of IBR during planning phase and decisions in the operational phase, this
could cause the power system exposed to risk of instability or being operated insecurely.

Figure 2.2-1 PLL behaviour under phase jump (left) and unbalanced fault (right) tests in PDT and EMT
models [R24]
Figure 2.2-2 illustrates the frequency range of different power system phenomena ranging from a few
Hz to several kHz often associated with pulse width modulation (PWM) in voltage source converters.

40
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

PDT models PDT models

capable of incapable of
representing representing

Figure 2.2-2 Frequency range of different power system phenomena and limits of PDT modelling
EMT models are replacing PDT models for large-scale power system studies in circumstances where
the PDT models fail to predict the phenomena of interest. This often occurs when the phenomenon of
interest has a dominant frequency deviating by more than ± 5 Hz with respect to the network
fundamental frequency, or when the system strength available to IBR approaches close to or drops
below the withstand capability of IBR. The latter application will apply even if the dominant frequency of
interest is at or near the fundamental frequency. Examples will be presented in this section
demonstrating important differences between the responses obtained from the PDT and EMT models
under low system strength conditions.
Such an approach allows for accurate and adequate methods to manage the impact of new, modified
or existing generation and other power system plant on power system security and network transfer
capability.
To gain the maximum accuracy, EMT models should have a complete representation of all fast, inner
controls as implemented in the real equipment. It is possible to create models which embed (and
encrypt) the actual hardware code into an EMT component providing a direct one-to-one representation
of the actual plant control codes without any potentially erroneous assumptions or manual
implementation errors.
As an example, model content requirements for wind turbine generators (WTG) in PDT and EMT tools
as function of studies of interest is shown in Table 2-1 (i.e. what the model must represent in its contents
for a given study type). Items highlighted with an asterisk depend on wind turbine type and make, and
could be very important or have no impact on studies of interest. The need to represents such aspects
would need to be determined in a case-by-case basis in liaison with the respective OEM. Note that
analysing high-frequency transients is out of scope of this TB, and modelling needs for such studies are
provided for comparison purposes only.
Table 2-1 Adequate representation of wind turbine components in PDT and EMT tools as function of
studies of interest
Transient Sub- High-frequency Harmonics#
stability synchronous transient
interactions
Aerodynamics PDT*, EMT* EMT* - EMT*
Components

Pitch controller PDT*, EMT* EMT* - EMT*

Mechanical drive train PDT*, EMT* EMT - EMT*
Torsional damping PDT*, EMT* EMT - EMT*
A
Electrical generator PDT, EMT EMT EMT EMT

41
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Transient Sub- High-frequency Harmonics#

stability synchronous transient
interactions
Dynamic braking resistor / chopper PDT, EMT EMT - EMT
DC link PDT, EMT EMT EMT* EMT
IGBT switches and PWM switching - - EMT EMT
Unit transformerB PDT, EMT EMT EMT EMT
Internal filters PDT, EMT EMT EMT EMT
Inner loop converter control EMT EMT EMT* EMT
Outer loop converter control PDT, EMT EMT EMT* EMT
C
Phase locked loop EMT EMT EMT* EMT
D
Frequency control PDT, EMT EMT - EMT
High voltage ride-through PDT, EMT EMT EMT -
Low voltage ride-through PDT, EMT EMT - -
Multiple fault ride-through PDT, EMT EMT - -
limitations
Protection PDT, EMT EMT EMT -

* Requirement determined on a case-by-case basis in consultation with OEM

#
It is noted that load-flow harmonic models may address steady-state harmonic requirements
A. Fifth-order generator.
B. Including saturation for EMT models. For PDT models it is acceptable to represent three-winding transformers as two-winding equivalents.
C. Explicit representation.
D. Including frequency raise and lower, frequency droop and deadbands.

2.2.2 Network components

Unlike PDT studies where often full-scale model of a region/country/continent is often used without
having a significant impact on simulation speed, in typical EMT studies only a portion of the power
system network is represented in detail. Even in the case of large system representation in EMT, it is
foreseen that some parts of the system (‘external network’) will be represented by network equivalents.
When the external network is reduced using mathematical techniques such as Krone reduction, there
is no guarantee that the resultant network is passive. The mathematical reduction method may not
guarantee passivity. As well, most utility network data bases contain non passive elements (such as
negative resistances) to represent various network conditions. Such representations are generally
suitable and work well for PDT calculations. However, when they are converted to EMT domain, such
non passive (non-physical) elements can impact the accuracy and the numerical stability of the overall
simulation.
A more accurate method is the expansion of the reduced network (i.e. adding few more external nodes),
so that the resultant network is passive and hence can represent using positive passive elements such
as resistors, inductors and capacitors (e.g. [R25]). This is further discussed in Section 4.1.3.
The condition of the original data form should be given due consideration even where the data is used
to develop the network model in the EMT platform. Presence of unrealistic data (for example negative
resistances) in the EMT network model can lead to unrealistic results or numerical instability. When data
is converted from an existing database to the EMT platform, the interpretation of specific data items
should also be given due consideration. Some examples include, base values used for ‘Per Unit’
conversion, interpretation of machine mechanical damping parameters and saturation parameters.

2.3 Comparison of simulated responses

A key matter to confirm for models used in conventional power systems (with a relatively low share of
IBR) is that regardless of the tool, they should give rise to the same results. It is understood that different
tools have very different algorithms, but they should all give comparable results such that the study
engineer can trust the results regardless of the platform they are using.
As discussed in various case studies in this section, operation under low system strength conditions
could mean that comparison of PDT and EMT model will not produce identical results. However, the
intent of model comparison remains, regardless of the available system strength for the IBR under
consideration.

42
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

To segregate external influences and IBR interactions, the first stage of benchmarking typically involves
the use of a single machine infinite bus (SMIB) case such that a like-for-like comparison can be made.
This way, in subsequent studies the engineer has confidence in the model. Figure 2.3-1 shows a project
study cycle, including the various verification (v) loops to ensure that once the project is integrated, it is
in fact a correct representation.

Actual Controls
or Customer V
V Simulator On Site Tests

Basic Design Basic Design RTS

V
Studies Studies Real network
10 bus AC
PDT PDT V
V
Full AC 10 bus AC
network Plant model

EMT

10 bus AC

Figure 2.3-1 Study Comparison Loop

One must also consider that comparing results of different simulation tools for the same model (even
PDT vs PDT or EMT vs EMT) can present slight discrepancies. It is important to establish the “reference
waveforms” and compare based on this reference. Factors such as the solver and the time step of
simulations will have an influence on the results
2.3.1 Examples indicating acceptable correlation
2.3.1.1 Example 1
The following figures present one example of comparing solar farm response to the same fault in SMIB
system. In this example, site-specific PDT and EMT solar farm models are connected to the same SMIB
test systems separately in PDT and EMT simulation tools, with the same equivalent grid impedance.
The same three-phase disturbance was applied to the solar farm models at the same location, in both
PDT and EMT simulation, and the response of the solar farm models from both PDT and EMT simulation
were overlaid in the following plots:

Figure 2.3-2 Solar farm active power response overlay: PDT vs. EMT

43
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.3-3 Solar farm reactive power response overlay: PDT vs. EMT
Figure 2.3-2 and Figure 2.3-3 showed similar response from the solar farm in both PDT and EMT
simulation, although the responses are not exactly the same. The solar farm demonstrated similar
reactive power contribution during the fault, and similar post fault active power recovery in both PDT
and EMT simulation. However, it should be noted this comparison is tested in SMIB environment, where
only one solar farm is considered. In a wide area model where multiple generators are present, other
phenomenon such as control system interaction might be identified, where PDT and EMT simulation
can produce very different results. This is elaborated further in the next section.
2.3.1.2 Example 2
The figures below display the results of site-specific models vs. measurement data at a PV plant in
Vietnam. The total plant capacity is 205.2 MVA, consisting of 8 blocks and a total of 45 inverters. Each
block consists of either 6 or 5 inverters of 4.65 MW each.
User defined models of inverters and park controller were developed for network analysis. In order to
demonstrate a reasonable functioning of the user defined models built in PDT and EMT domain, the
following tests were performed and matched with the site measurements provided:
▪ Test: Active power response test – Step response test
▪ Test: Reactive power response Test – Step response test
The results display a configuration of 6 PV inverter of 4.65 MW amounting to a total of 27.36 MW, to
match the site measurement response at the Point of Connection (POC). In these tests, the response
of the plant controller is validated in conjugation with the inverter model. The present configuration was
tested with a Short Circuit Ratio (SCR) at the point of interconnection of value 3. Change in active power
and reactive power reference setpoint is simulated and compared with site measurements.
An increase or decrease in the active or reactive power reference results in a corresponding increase
or decrease in the active or reactive power. Various combinations of steps have been executed to
check the performance of the User-Defined Model (UDM) vs. measurement.

44
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.3-4 Sub-Test A: Active power output Figure 2.3-5 Sub-Test B: Active power output
decrease comparison increase

Figure 2.3-6 Sub Test A: Active power reference Figure 2.3-7 Sub Test B: Active power reference
decrease comparison increase comparison
Both the simulation (PDT and EMT) and measured responses show a good match with regard to the
change in active and reactive power references.
However, it should be noted that minor deviations may be seen between the simulated and
measurement data. This may be attributed to either the limitations of simulation, error in measurement
data and not the model itself. For example, in Figure 2.3-8 the minor deviation between the PDT
simulation response and measurement is attributed to the fact that voltage in the network increases with
the increase in reactive power. However, while simulating the current scenario, the POC voltage is held
constant at 1 pu, since the prime purpose here is to validate the plant control model implementation and
reactive power response between the measurement and the developed models.

Figure 2.3-8 Sub-Test A: Reactive power output Figure 2.3-9 Sub-Test B: Reactive power output
change change

45
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.3-10 Sub Test A: Reactive power Figure 2.3-11 Sub Test B: Reactive power
reference change reference change
The above results demonstrate that both vendor specific PDT and EMT models can provide accurate
results which may be used to perform dynamic network analysis. To display a plausible response with
PDT models in IBR dominated systems the need for vendor specific models validated against
measurements is essential compared to generic PDT models.
2.3.1.3 Example 3
This example compares the response of the PDT and EMT models of a synchronous machine during a
three-phase fault. The events observed in the figures below correspond to the application of a
symmetrical fault at t = 5 s on a bus close to the synchronous machine, tripping of an adjacent line 4.5
cycles later at t = 5.075 s and fault clearance at t = 5.0917 s. In the upper part of the Figure 2.3-12, the
comparison of the reactive output power of the machine measured at the terminals (QT) for both models
is observed. In the lower part of the figure, the comparison of the active output power of the machine
measured at the terminals (PT) is presented. Red curves correspond to EMT model of the synchronous
machine and blue curves to PDT model.

Figure 2.3-12 Comparison of QT (upper) and PT (lower) of a synchronous machine for PDT (blue) and EMT
(red) models during a three-phase fault
For a three-phase fault applied at the terminal of a synchronous machine, 3 components can be
expected in the phase currents of the machine: 1) a fundamental frequency component, 2) a second
harmonic component and 3) a DC component [R26], [R27]. The second harmonic component is due to
the subtransient saliency (Xq” ˗ Xd”) which has been typically neglected in some PDT models [R28].
Another simplification consists in the removal of the flux linkage derivative term in the stator equations,
leading to the omission of stator transients. Omitting these transients then leads to the removal of the
DC component in the stator currents [R27].
The most used PDT models for the synchronous machines correspond to the well-known GENROU,
GENSAL, GENTPJ and GENTPF models [R28], [R29] which represent only the positive sequence
dynamics seen on the machine terminals. In EMT models, on the other hand, all the components are

46
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

considered from the general formulation of the dq0 synchronous machine models for salient pole or
round rotor.
In most cases these simplifications of the PDT models have been acceptable for stability studies.
However, with a high penetration of IBRs with more sensitive and fast controls and protections, in grids
with lower inertia and lower short-circuit capability, it may be important to consider the loss of information
and potential instability scenarios which may not be considered using the PDT models.
2.3.2 Example indicating unacceptable correlation
Simplified modelling of IBR control systems in PDT tools means that there are always differences
between the responses obtained from the PDT and EMT models. As discussed above in some instances
these differences may be negligible. When comparing PDT and EMT models the size of the error is not
necessarily the main point of interest, but whether the PDT model can provide an indication of IBR
instabilities, and whether or not the use of PDT models may result in incorrect power system planning
and operation. The term ‘’unacceptable corelation’’ discussed in the examples below refers to these
situations.
2.3.2.1 Australian examples
Two examples are provided in this section, where PDT and EMT simulation produced very different
results for the same power system contingency. The following results were evaluated in a wide-area
EMT model of the power system.
2.3.2.1.1 Example 1
Figure 2.3-13 shows a transmission bus voltage profile following a disturbance simulated in both PDT
and EMT tools. Both simulations show similar transient voltage reduction during the fault. However, in
PDT simulation, the voltage overshoot upon fault clearance is much higher compared with the EMT
simulation. More importantly, EMT simulation identified post contingency sustained voltage oscillation,
which is not observed in PDT simulation. This example demonstrates that the use of PDT models is not
appropriate when the focus is on instabilities stemming from fast control systems deployed in the IBR.

.
Figure 2.3-13 Voltage oscillation in EMT simulation, not observed in PDT simulation
2.3.2.1.2 Example 2
Figure 2.3-14 shows the response of a high voltage direct current line (HVDC) with line commutated
converter (LCC), following the same power system disturbance, simulated in both PDT and EMT tools.
The results are dramatically different between PDT and EMT simulation, where the HVDC link rode
through the disturbance and recovered successfully in PDT simulation, but tripped in EMT simulation.

47
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.3-14 Active power response of LCC HVDC link in PDT and EMT simulation

Figure 2.3-15 Voltage profile at connection point of HVDC link in PDT and EMT simulation
Further investigation showed temporary voltage reduction due to a fault is very similar between PDT
and EMT simulation. However, EMT models demonstrated a system collapse compared with the PDT
simulation. Figure 2.3-15 shows comparison of voltage profile at HVDC connection point between PDT
and EMT simulation (note that the EMT model did not have all protection mechanisms represented).
In this scenario where tripping of the plant for the given scenario had been seen in practice, the PDT
model was seen to be overly optimistic and the EMT model more likely to represent actuality (noting the
absence of protection mechanisms in the model).
2.3.2.2 North American examples
2.3.2.2.1 VSC HVDC example
Figure 2.3-16 demonstrates a comparison between representations of an embedded VSC HVDC
system in an EMT tool and a PDT tool. The VSC HVDC system is embedded in a system with a SCR

48
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

of approximately 1.5, which is considered relatively weak. Following a fault, the PDT tool depicts stable
recovery, while the EMT tool depicts a loss of synchronism and instability in the inverter controls.

Figure 2.3-16 Voltage profile at the connection points of HVDC link in PDT and EMT simulation
2.3.2.2.2 Wind plant example
Figure 2.3-17 shows some characteristic trademarks of EMT and PDT tools when compared against
each other. The plots depict the response of several wind plants to a close-in three-phase fault. The
upper figures show real and reactive power, as well as voltage as the plants enter and exit their Fault
Ride-Through (FRT) modes.
Although there are differences in the details of the responses (particularly during the actual fault period),
the general plant behavior is similar between the tools, including temporary reduction of active power,
ramped recovery to full power, and reactive power response in the pre and post fault periods. When
the fault and immediate post-fault periods are examined however (including in the bottom figure), it is
possible to see that there are characteristic differences between the tools.
In the PDT tool, voltage and power quantities may step between one value and another between
simulation time-steps. However, in the EMT domain, RMS quantities for voltage and power must be
calculated from phase quantities, so a period of time is required for the instantaneous change in phase
quantities to be reflected in the RMS quantities.
Additionally, immediately upon fault clearing the PDT tool shows a direct response to the voltage
controller, which has been accumulating error during the low voltage fault, resulting in an immediate
response in Q upon fault clearing, and a voltage overshoot. The PDT tool also shows perfect, smooth
control of active power during the recovery period. However, in the EMT tool, the more detailed
representation of the converter shows that in fact, the PLL of the firing controls require time to
synchronize to the new system phase angle, and the resulting output of reactive and real power are not
smooth or perfectly controlled.
These differences (and other differences relating to the differences in control representation detail) are
typical in comparisons between these two types of tools. As network interconnections become weaker,
and as control complexity rises, the differences may become more pronounced, to the extent that the
PDT model may no longer be considered accurate in some cases.

49
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.3-17 Voltage oscillation in EMT simulation, not observed in PDT simulation
A second example is shown below showing an acceptable correlation of an HVDC system under test.
The fault was a single-phase fault at the rectifier. The comparison here is between the EMT model and
the Real Time Simulator (RTS) model.

50
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.3-18 HVDC Fault Response (Blue: Offline EMT, Green: RTS)
As can be seen, there was very good correlation between the EMT model and the RTS model. Both
systems include a 10-bus ac equivalent and the main circuit components of the HVDC system. The only
difference here was the RTS system used the actual controls via hardware in loop, whereas the EMT
model had a simulated version of the controls.
2.3.2.2.3 Model anomalies
Some difference in results between EMT simulations and PDT simulations can be expected, especially
for low system strength conditions. However, ERCOT has observed performance differences between
PDT models and EMT models for individual plants when conducting simple model tests. Such results
can erode confidence in system simulation results and create uncertainties about whether EMT or PDT
simulation results are actually more accurate.
Figure 2.3-19 and Figure 2.3-20 show the model response for individual IBRs when subjected to a
voltage profile consistent with the ERCOT low voltage ride-through requirement. In Figure 2.3-19, the
PDT model for the IBR produced a suspect response (i.e., multiple real power dips during the post-
disturbance period) while the EMT model produced a reasonable response to the low voltage event. In
Figure 2.3-20 the EMT model for the IBR produced a suspect response (i.e., absorbing reactive power
rather than injecting to provide voltage support during the low voltage transient) while the PDT model
produced a reasonable response.
Although EMT models are expected to better reflect certain performance details of IBRs that are either
simplified or ignored in the PDT model, the verification of EMT models is imperative to ensure the model
can accurately represent the dynamic response of an IBR and can be used to benchmark the PDT
model.
ERCOT continues to work with stakeholders to approve and implement the proposed dynamic model
validation and verification process (known as PGRR085 within ERCOT [R30]) to improve the quality
and accuracy of both PDT and EMT models. The proposal includes an EMT model validation
component, a verification component to ensure that field settings are properly reflected in the models,
and provisions to ensure that the performance of EMT models and PDT models is consistent. Similarly
in Australia, AEMO publishes a dynamic model acceptance test (DMAT) guideline [R31] to confirm
model adequacy ahead of performing planning and connection studies.

51
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.3-19 PDT and EMT model performance comparison with suspect PDT model response

Figure 2.3-20 PDT and EMT model performance comparison with suspect EMT model response

2.4 GUI considerations for wide-area EMT studies

The GUI (or schematic editor) has major impact on the creation, maintenance, and simulation of large
grids in EMT software. It is the most time-consuming step, but if carefully implemented, can set the user
up for the best ease of use possible when working with very large EMT models. The key ingredients of
the schematic editor are:
▪ Connectivity: the devices must remain connected when moved, geographical positioning options
and signal channelling functions.
▪ Single-line diagram capability with access to individual phases.
The network design should be hierarchical. It should replicate the network model directly without adding
artificial components for underlying numerical methods.

52
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

2.4.1 Hierarchy
In a hierarchical design, the simulated network is assembled using subnetworks and pages. The
subnetworks can contain other subnetworks. The top-level view should replicate the top level view of
the actual physical grid.
This concept is illustrated in Figure 2.4-1. The top design level contains subnetworks numbered from 0
to 5. Subnetworks 0 and 5 are at the second hierarchical level and subnetworks 1 to 4 are contained at
the third level from top down. This is also related to object-oriented development with hiding and
encapsulation. At the top network level the number of circuit details should be minimized and details
moved into subnetworks.

Figure 2.4-1 Hierarchical circuit: subnetworks containing subnetworks with masks

In Figure 2.4-2, the Static Var Compensator (SVC) is a top-level device made of several subnetwork
levels for modelling its internal circuits and controls. It appears as a small symbol with a mask for setting
internal parameters. Its contents become accessible in separate windows.

Figure 2.4-2 A snapshot from a 735 kV series-compensated network

An example of multimachine power plant representation is given in Figure 2.4-3. The shown AVR blocks
contain several encapsulated subcircuits.

53
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.4-3 Multimachine power plant representation

Subnetworks can also be used to create specific models that may be used at different locations in the
grid. The concept of instantiation allows to create unique subcircuits. All instances can be modified at
once by updating only the definition of one subnetwork. If two subcircuits (ST1A in Figure 2.4-3, for
example) are of the same type, then any (data or circuit) change in one of the sub-circuits is
automatically reflected in all other subcircuit instances. This method of managing subcircuits is very
important for maintaining large grids. A unique definition can be used for contents, but subcircuit mask
data allows to change data according to location. Subnetwork data and contents are programmable.
2.4.2 Layers
Given the goal to develop a reference network for various study types, GUI layers can be used for
network reconfiguration. Devices are given a layer attribute that allows them to be excluded and included
through a single layer selection command. Layers provide a convenient approach for precision and
study type selections. If the objective is to perform faster transient stability computations, then non-linear
devices such as surge arresters or magnetization branches can be excluded. Precision can be increased
by including the series-capacitor protective arresters. Layers should be made available at all hierarchical
levels. As an example, in Figure 2.4-3 the load-flow (LF) devices are active only in the load-flow solution
layer, but transmit data to the adjacent synchronous machines for the corresponding steady-state
solution before the time-domain initialization.
2.4.3 Scripting and attributes
Using only the GUI mouse-based functions for entering and maintaining data is not viable for building
and maintaining large networks. Entering data and making data changes in many devices, validating
data, searching for devices, and sending requests to select operational topological conditions and
layers, can be achieved efficiently through scripting.
In many EMT platforms, each GUI component is an object with attributes and methods. A set of existing
attributes (fields) and methods can be extended to include various functionalities according to the needs.
A top circuit object allows to grab and modify data for all components: models interconnected with
signals.

54
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Data attributes contain electrical data, device symbol drawing data and position data. Scripting can be
used to access and modify data attributes. A single script, for example, can be used to enter corrected
data for hundred transmission lines without manually touching the actual lines on the GUI. Device
methods provided with specific tags for identifying triggered scripts, can be used to automatically trigger
data refresh scripts due to changes in database files or due to attribute changes affecting other device
properties. Scripting has been found essential by many organisations for creating and maintaining large
grids. Script methods can also be used also to navigate through the network and locating devices.
The maintained networks should be regularly updated and adjusted to reflect changes on the actual
network and to account for its evolution. The networks should also be regularly synchronized with the
utility database to reflect the changes in the data sources. This is a significant effort requiring the
dedication of engineering resources. Several tools and scripts can be used to facilitate repetitive tasks:
▪ Import and update the data from the databases
▪ Synchronise the load-flow constraints and load profiles for the different configurations
▪ Data manipulation and scenarios studies
▪ Load variation using coefficients or other means
These tasks are required in order to ensure network coherence.

2.5 The role of screening methods

Screening methods are technical evaluations, calculations and simulations that can be performed to
obtain an idea of potential stability and resonance issues. The screening methods are employed due to
the following reasons.
▪ These are less tedious and time consuming compared to detailed large-scale time domain
simulations.
▪ The screening level results provide insight and explanations to results and observations from model
details in time domain analysis.
▪ The screening level results may be used to identify critical scenarios that need specific attention at
detailed simulation stage.
▪ The data requirements for screening level studies may not be as onerous as for the final detailed
study and hence can be accomplished at an early stage of the planning study.
Some of the screening level analysis reported in literature and adopted by planning engineers include;
[R32][R33].
▪ Steady state fault level calculations
▪ Calculation of network harmonic impedance profile to identify network resonance points that may
impact inverter operation
▪ Dynamic frequency scanning techniques to determine the damping characteristics of inverters over
a selected frequency range
▪ EMT simulation-based screening to determine damping on torsional modes of generator shafts that
are in close proximity to inverter based devices.
Screening methods have been traditionally used to determine when wide-area EMT modelling is
required as opposed to PDT modelling. However, in recent times the focus and value proposition has
shifted more towards providing insight on how to reduce the number of EMT studies required where
PDT modelling would not provide an accurate answer. The key driver for this change has been increased
uptake of IBR and emergence of phenomena which cannot be uncovered by PDT modelling. Such an
application of screening methods has been primarily contemplated in regions with established wide-
area EMT models.
However, there still remain some applications for screening methods providing guidance on the use of
PDT vs EMT models in regions where a large-scale system model in EMT may not be available.
2.5.1 Available Fault level
This section uses further terms that have the meanings set out opposite them in Table 2-2. The following
assumptions are made:
▪ For fault level calculations the generating unit terminal voltages are 1 pu
▪ The fault level calculations use the synchronous generation (SG) sub-transient impedance values
(Xd´´, Xq´´).

55
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ IBR require a minimum level of synchronous three-phase fault level which is equal to the minimum
SCR (as advised by the manufacturer) multiplied by the MW rating of the IBR 2.
▪ Note: Fault level calculations made with transient or sub-transient impedances produce somewhat
different fault current levels. Sub-transient values will give a higher estimate of fault currents for
faults cleared in primary protection clearance time. Since the main purpose of this screening
methodology is to assess the need for EMT study due to adverse asynchronous generation
interaction, especially during and immediately after the fault, sub-transient impedance values are
used.
It is inappropriate to use the “steady state” synchronous generation impedance values (Xd, Xq) for SCR
calculations, due to the strong influence of the generating units’ automatic voltage regulators over these
slower time scales.
As system strength reduces, reliable FRT performance will normally falter before the “natural” steady-
state limits of power transmission are reached. This is because the physical limits of equipment (in terms
of rating and speed of response) are most stressed during FRT. Other performance metrics, such as
steady-state stability, also reduce with system strength but it is normally the FRT performance that
determines the limiting response time/stability compromise of the generating unit’s control system.
Table 2-2 Defined Terms
Term Definition
AFL Available Fault Level in MVA
Effective Impedance This IBR impedance is given by V2 ∕ (MSCR × MW rating).
MSCR Minimum SCR: the lowest SCR that the IBR requires to comply with its performance standards.
Synchronous three-phase The three-phase fault level, in MVA, calculated for a network with only synchronous generation
fault level (SSG) plant connected.

2.5.1.1 Screening Concept

Most existing IBR are only specified for operation above a minimum SCR at their inverter terminals,
which must be translated to an equivalent minimum SCR at their connection point by factoring in the
reticulation and transformer impedances. This specification is often driven by the FRT limitations of IBR
under low system strength conditions. The main IBR challenges at low SCRs relate to:
▪ The provision of sufficient fast reactive power support; and
▪ The maintenance of close synchronism with the rapidly changing system phase angle.
The methodology described here shows a practical "screening" process for new IBR connections. The
impact of IBR beyond its connection point is assumed to be proportional to its MW rating multiplied by
its MSCR. Therefore, the IBR is represented as a Thévenin voltage source connected to the network
behind its Effective Impedance. This representation does not generate the actual IBR fault currents but,
instead, produces a current related to the stated performance requirements of the IBR. This concept is
an extension of the calculation method commonly employed where IBR shares a common connection
point. This screening process provides a metric to estimate the need for wide-area EMT modelling and
is described by way of two examples. The first example introduces the concept, with two IBR sharing
the same busbar, while the second example shows how this concept can be simply expanded to provide
a screening metric for more distant IBR interactions.
2.5.1.2 Example (1): concept of AFL at a local busbar
This example is a basic demonstration of how to estimate the capability of the network’s connection
point to support an IBR connection.

2
Recent experiences in particular under high IBR penetration scenarios have indicated that the use of MVA rating would be more important.
However, a discussion on the impact of using MW vs MVA is not the focus of this Technical Brochure.

56
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.5-1 Calculation of local IBR impact on connection point capability

Consider the connection point shown in Figure 2.5-1 where an existing 200 MW IBR is connected. A
second IBR (shaded) wishes to share the connection point.
For ease of explanation, the generation outputs are all 1 pu voltage at zero phase angle, inductances
are given in ohms (pu) and circuit resistance is ignored. The following calculation steps are made:
1. Calculate the synchronous fault level at Connection Point (all IBR disconnected):
1.02 / (0.07 + 0.03) × 100 MVA = 1000 MVA
2. Calculate the required fault level for the existing IBR:
(MSCR × MW rating) = 2.0 × 200 MW = 400 MVA
3. Calculate the AFL (for new IBR connection) = (1) subtract (2):
1000 MVA - 400 MVA = 600 MVA
Now find “prospective” maximum ratings for the new IBR:
4. Maximum rating of new IBR with MSCR of 2.0 is (AFL / MSCR) = (600 / 2.0) = 300 MW
Maximum rating of new IBR with MSCR of 1.5 is (AFL / MSCR) = (600 / 1.5) = 400 MW
Clearly these calculations are quite straightforward when considering IBRs that share the same
connection point. If this screening method calculates that the proposed IBR would consume more than
the 600 MVA of AFL then an EMT study would be triggered to confirm compliant performance of the
new and existing IBR.
2.5.1.3 Example (2): calculation of AFL at nearby busbars

Figure 2.5-2 Calculation of remote IBR impact on connection point capabilities

Using the same principle described in Section 2.5.1.2, the AFL for a possible IBR connection can be
calculated in four steps. These calculations can be made using standard PDT fault level calculation
tools. Note: for SSG fault calculations the sub-transient impedance values (Xd´´, Xq´´) are used.
The following calculation steps are made for each busbar in the region of interest:
1. Calculate the total “fault level” with all generation connected (but represent each IBR as a Thévenin
voltage source behind its Effective Impedance): Stotal (MVA)
2. Calculate the synchronous fault level with only SG connected: SSG (MVA)
3. Find the difference in these two “fault levels”: Δ (MVA) = Stotal – SSG
4. Find the AFL: AFL (MVA) = SSG – Δ

57
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Table 2-3 AFL Calculation steps for example (2)

Step (#) Bus 1 Bus2
(1) Stotal (MVA) 2786 1805
(2) SSG (MVA) 1429 1000
(3) Δ (MVA) 1357 805
(4) AFL (MVA) +72 +195
The calculations show that even with the new larger IBR connection the AFL remains positive at both
its local busbar and at the nearby busbar of the existing IBR. Therefore, this example would not trigger
the requirement for EMT studies but since the AFL is not large any further IBR penetration would likely
warrant EMT studies.
2.5.1.4 Application of this Methodology
In 2018, due to the unprecedented growth in IBR enquiries in Australia, the Australian Energy Market
Operator (AEMO) applied this screening methodology to all new IBR applications. While its technical
application proved straightforward for AEMO and the TSOs to implement, there were several
unexpected and sometimes negative consequences of its usage.
2.5.1.4.1 Impact on Choice of MSCR
Clearly the value taken for MSCR will significantly determine the degree of interaction estimated by this
methodology. However, the market gave no incentive to proponents to offer IBR with a low MSCR
capability. In a competitive environment the complexity of triggering an EMT study can be used to
stymie the connection process for other proponents. Consequently, it was noticed that the stated SCR
values for some IBR were quite high (greater than 3.0).
2.5.1.4.2 Impact of Legacy Plant
Many of the existing IBR connections were commissioned before the requirement for EMT models
became mandatory. Often their performance under low system strength conditions was not known and
was difficult to determine. Therefore, it was decided to apply a conservative MSCR value of 3.0 to each
of these “legacy” systems. The consequence of this was that the screening process triggered many
more EMT studies, which needed the retrospective creation of many EMT models for the legacy plant.
Most of these studies then confirmed that the MSCRs for the majority of legacy plant do not normally
exceed 2.0.
2.5.1.4.3 Impact of Speed of Response Differences
This screening methodology is conservative in that it assumes that the FRT response times of all IBR
are the same. In practice, there will be considerable variations across the IBR fleet, which will reduce
the degree of interaction.
2.5.1.4.4 Impact of Load
While the AFL screening process does highlight the effect of impedance between nearby IBR, in
reducing interactions, it does not take into account a similar benefit that is provided by having local loads
between them. Especially when considering FRT, an IBR should not require the same MSCR, as for
example calculated by a SMIB test, if some of its generated power is consumed locally prior to export
into the grid.
2.5.1.4.5 Reduction of MSCR value
Given the influence of the above-described variables in reducing the need for EMT study, users of the
methodology are recommended to consider whether the stated MSCR, as determined by a SMIB study
for instance, could be reduced. For legacy plant, or new plant where high MSCR requirements are
specified, the screening study may also account for known performance of similar plant to determine if
it would be more realistic to apply MSCR values closer to 2.0.
2.5.2 Improved analytical multi-infeed interaction factor
Before performing detail interactions studies, analytical criteria can be used to estimate the potential risk
of interaction. These analytical criteria can be based on screening method to identify the network and
IBR devices that may have a potential risk of interaction.

58
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

In the Technical Brochure [R34], the assessment criteria that detect the risk of interactions between
LCC based HVDC links were presented. These criteria are based on standardized indicators. However,
there are limitations of using such indicators as the criteria are adapted and validated only for HVDC
LCC links (as stated in [R34]). Therefore, the use of these indicators to study the risks of interactions
between different types of components based on VSC based power electronics as HVDC VSC link,
FACTS and wind farms is not suitable. To study the interactions between several types of new power
electronics components, an improved Multi-Infeed Interaction Factor (MIIF) method is proposed [R35]
and is described the following subsection.
2.5.2.1 MIIF approach for LCC HVDC
The MIIF criterion (Multi Infeed Interaction Factor) defined in [R34] is recalled. This criterion quantifies
the existing electrical connection between two or more power electronics components. By applying a
small voltage drop of 1% (ΔVn) at one of the connection points, the voltage drop on the other connection
points (ΔVe) Figure 2.5-3 is computed. The voltage drop ΔVn is considered small enough in order not to
impact the control and protection system. The ratio between these two voltage differences represents
the MIIF factor:
∆𝑉𝑒
𝑀𝐼𝐼𝐹𝑒,𝑛 = Equation 9
∆𝑉𝑛

Figure 2.5-3 Determination of MIIF factor between two connection points [R34]
By computing this factor for each PCC#N, a square matrix is obtained (Table 2-4). The diagonal terms
of this matrix are necessarily equal to one and the matrix is usually asymmetric.
2.5.2.1.1 Weighted MIIF table
The weighted MIIF factor (Table 2-5) is added to take into account the different components rating:
MIIFj, iPdcj. Multiplying the rated active power (Pdc) of an HVDC LCC connection allows to weight the
voltage drops depending on Pdc: Considering X and Y components; a X component having a higher
power rating than component Y, will have a greater impact on this component Y.
However, the opposite is not necessarily true; i.e. component Y can have a negligible impact on the
component X. The multiplication of MIIFj,I with Pdc proposed in [R34], can be justified as follows: in a
HVDC LCC link, there is a fixed relation between Pdc and reactive power (ratio Q/Pdc ≈ 0.6). Therefore,
a network voltage drop causes a variation in reactive power, which would imply a variation in Pdc.
However, such multiplication is not necessarily true for other power electronic components based on
VSC technologies. For instance, in a HVDC VSC link, there is a decoupling between the active and
reactive power. Therefore, the relationship between P and Q is low and consequently the MIIF j, iPdcj
criterion is not applicable. Moreover, in a FACTS, only reactive power is supplied, hence P dc is zero,
which implies a zero-interaction criterion which is not necessarily true. For FACTS controllers, the risk
of interactions is necessarily operating at the level of reactive power and AC voltage and not on active
power.
Table 2-4 MIIF Amplitude
MIIF
PCC#1 PCC#2 PCC#N
amp.

V1 V2 VN

PCC#1
V
MIIF1,1 = V
MIIF2,1 = MIIFNV,1 =
V1 V1 V1

59
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

V1 V2 VN

PCC#2
V
MIIF1,2 = V
MIIF2,2 = MIIFNV,2 =
V2 V2 V2

V1 V2 VN

PCC#N MIIF1,VN = MIIF2,VN = MIIFNV, N =
VN VN VN

Table 2-5 Weighted MIIF

% PCC #1 PCC #2 PCC #N

V Pdc1 Pdc1
PCC #1 100 MIIF2,1 MIIFNV,1
Pdc 2 PdcN

V Pdc 2 Pdc 2
PCC #2 MIIF1,2 100 MIIFNV,2
Pdc1 PdcN

PdcN PdcN
PCC #N MIIF1,VN MIIF2,VN 100
Pdc1 Pdc 2

2.5.2.1.2 Rules
From the off-diagonal values obtained in percentage in Table 2-5, the risk of potential interactions may
be assessed. The following criteria are based on [R34]:
▪ Ratio less than 15%: risk of interactions is low
▪ Ratio between 15% and 40%: risk is moderate
▪ Ratio above 40%: risk of interaction is high
2.5.2.2 Improved MIIF approach for most IBRs
To work around the limitation of the MIIF classic approach and to allow weighting the impact of a generic
power electronic components over another one, the Pdc is simply replaced by the rated reactive power
Q. This change allows to adapt the criteria for more generic power electronics devices as HVDC VSC
links, FACTS and wind farms. The new weighted MIIF based on amplitude is presented in Table 2-6.
Table 2-6 Weighted MIIF amplitude in percentage
% PCC #1 PCC #2 PCC #N

V Q1 Q1
PCC #1 100 MIIF2,1 MIIFNV,1
Q2 QN

V Q2 Q2
PCC #2 MIIF1,2 100 MIIFNV,2
Q1 QN

QN QN
PCC #N MIIF1,VN MIIF2,VN 100
Q1 Q2

Similarly, to the relationship between the voltage drop and reactive power, a link between angle
difference and active power is established. A new MIIF matrix based on angle differences is deduced in
Table 2-7. Then the weighting is calculated in Table 2-8.
Table 2-7 MIIF Angle
MIIF
PCC#1 PCC#2 PCC#N
angle

 1   2  N
PCC#1 MIIF1,1 = MIIF2,1 = MIIFN,1=
1 1 1

 1  2  N
PCC#2 MIIF1,2 = MIIF2,2 = MIIFN ,2=
 2 2  2

 1   2  N
PCC#N MIIF1,N = MIIF2,N = MIIFN , N =
 N  N  N

60
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Table 2-8 Weighted MIIF angle in percentage

% PCC #1 PCC #2 PCC #N

 P1 P1
PCC #1 100 MIIF2,1 MIIFN ,1
P2 PN

 P2 P2
PCC #2 MIIF1,2 100 MIIFN ,2
P1 PN

PN PN
PCC #N MIIF1,N MIIF2, N 100
P1 P2

2.5.2.2.1 Rules
From the off-diagonal values obtained in percentage in Table 2-6 and Table 2-8 the risk of potential
interactions may be assessed. The threshold criteria is the same as the classical MIIF method [R34]
because there is no sufficient practical example, with VSC technology, that proves that threshold values
should be modified:
▪ Ratio less than 15%: risk of interactions is low
▪ Ratio between 15% and 40%: risk is moderate
▪ Ratio above 40%: risk of interaction is high
2.5.2.2.2 Criteria applied on Northern France projects
A practical example with the improved MIIF approach is presented in this section. The power electronics
devices planned or existing in the North of France is considered hereafter.
Several power electronics components are installed or in planification stage in the north area of the
French transmission network. Below is the list of components included in this study:
▪ IFA2000: 2 bipolar HVDC LCC links rated power 2 × 1,000 MW
▪ ELECLINK 1 monopolar HVDC VSC link rated power of 1,000 MW and ± 300 Mvar
▪ Wind farm (WPP-CSM) with a total capacity of 450 MW and ±150 Mvar
▪ Tourbe SVC with a capacity of +250 and -100 Mvar
▪ IFA2: 1 monopolar HVDC VSC link rated power of 1,000 MW and ± 300 Mvar
▪ FAB: 2 monopolar HVDC VSC links, total capacity of 1,400 MW and ± 560 Mvar
Note that, the future project Eleclink that will be connected on the same bus bar as IFA2000, is not
considered in this study, however, it will not affect the conclusion.
The 400 kV and 225 kV voltage levels in the Northern area of France are considered for the AC network
modelling. The schematic representation of the network including the five projects is shown in Figure
2.5-4.
Time domain simulations are performed using an offline EMT simulation with the manufacturers’ models
for existing equipment and with generic models for future projects. The improved MIIF approach can be
used with EMT tools or PDT tools.

61
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 2.5-4 Power electronics devices embedded in Northern France Network

Variations of voltages and angles at PCC of each power electronics components are applied. To include
the impact of the network configuration, the extreme two network configurations are considered: Peak
load and off peak. In Table 2-9 and Table 2-10, the weighted amplitude MIIF for peak load and off peak
are calculated. In Table 2-11 and Table 2-12, the weighted MIIF angles for peak load and off peak are
presented respectively. Since IFA2000 and ELCLINK are at the same busbar, only IFA2000 is
highlighted in the following tables.
Table 2-9 Peak load - MIIF weighted amplitude
MIIFij.Qi/ HVDC- SVC- WPP- HVDC- HVDC-
Qj (%) IFA2 Tourbe CSM. IFA2000 FAB
HVDC-
100 48.8 38.7 4.7 43.1
IFA2
SVC-
56.3 100 78.6 3.4 26.2
Tourbe
WPP-
39.1 69.5 100 2.3 18.3
CSM
HVDC-
1.5 1 0.8 100 0.7
IFA2000
HVDC-
36.4 19.9 16.0 2.1 100
FAB

Table 2-10 Off peak - MIIF weighted amplitude

MIIFij.Qi/ HVDC- SVC- WPP- HVDC- HVDC-
Qj (%) IFA2 Tourbe CSM. IFA2000 FAB
HVDC-
100 50.6 40.3 10.6 66.8
IFA2
SVC-
65.9 100 78.9 8.7 47.6
Tourbe
WPP-
48.2 72.5 100 6.4 34.9
CSM
HVDC-
3.5 2.3 1.8 100 2.7
IFA2000
HVDC-
38.6 21.2 17.0 4.7 100
FAB

62
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Table 2-11 Peak load - MIIF weighted angle

MIIFij.Pi/P HVDC- SVC- WPP- HVDC- HVDC-
j (%) IFA2 Tourbe CSM. IFA2000 FAB
HVDC-
100 0 31.6 5.5 42.5
IFA2
SVC-
inf inf inf inf inf
Tourbe
WPP-
42.8 0 100 3.1 19.4
CSM
HVDC-
1.6 0 0.7 100 0.7
IFA2000
HVDC-
34.1 0 12.1 2.2 100
FAB

Table 2-12 Off peak - MIIF weighted angle

MIIFij.Pi/P HVDC- SVC- WPP- HVDC- HVDC-
j (%) IFA2 Tourbe CSM IFA2000 FAB
HVDC-
100 0 34.4 11.2 63.7
IFA2
SVC-
inf inf inf inf inf
Tourbe
WPP-
51.4 0 100 7.2 37.0
CSM
HVDC-
3.3 0 1.5 100 2.4
IFA2000
HVDC-
36.1 0 14.3 4.8 100
FAB

From the results in Table 2-9 to Table 2-12 the following can be concluded:
▪ The off peak and peak load give relatively close results. In general, the off peak gives higher risks
of interaction.
▪ Between the wind farms (WPP-CSM) and the SVC-Tourbe, there is a potential interaction (>70%)
at the variation of the voltage amplitude (Table 2-9 and Table 2-10).
▪ For HVDC-IFA2, at the variation of the voltage (Table 2-9 and Table 2-10), there is a high risk
between SVC-Tourbe and HVDC-IFA2. It should be highlighted that the impact of the HVDC-IFA2
link on the SVC is higher (56.3% and 65.9%) than the impact of the SVC on the HVDC-IFA2 link
(48.8% and 50.6%).
▪ There is a moderate risk between WPP-CSM and HVDC-IFA2. It should be highlighted that the
impact of the HVDC-IFA2 (Table 2-9 and Table 2-10) on the WPP-CSM and vice versa is moderate.
However, regarding the exchange of active power, HVDC-IFA2 has an impact on WPP-CSM
(42.8% and 51.4%). A moderate impact of WPP-CSM on HVDC-IFA2 (31.6% and 34.4%) is
noticed.
▪ For HVDC-IFA2000, there is no risk of interaction between IFA2000 and the remaining power
electronics components (i.e. SVC, WPP-CSM, HVDC-FAB and HVDC-IFA2) < 4%.
▪ For HVDC-FAB, there is a risk of interaction on HVDC-IFA2 (for Peak load around 60% and for Off
load around 40%). However, HVDC-IFA2 has a medium impact on HVDC-FAB (<40%). The impact
of the HVDC-FAB over SVC and WPP-CSM are medium (<40%) Furthermore, the impact of SVC
and WPP on HVDC-FAB is low.
Based on the improved MIIF approach, one can conclude that for interaction studies, IFA2000 impact
is negligible due to its remote connection. Therefore, a simplified representation of an equivalent voltage
source of IFA2000 model would be enough. This speed up the simulation time and limits the number of
studies. The EMT model deduced can be simplified as shown in Figure 2.5-5. The remaining AC network
is modeled using the Frequency dependent network equivalents FDNE [R36].

63
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

This section has determined with an analytical approach, the risk of interactions between the various
components of power electronics. It also allowed on one hand, to eliminate the risk of interactions that
could take place between IFA2000 and other projects and, secondly, to show that the risk of interactions
could appear between the four other components.

Figure 2.5-5 Reduced network in EMT including the four power electronics components

2.5.3 Advanced short-circuit strength metric

A key aspect for the stability of IBR in weak grid conditions is associated to the sensitivity of the IBR
terminal voltage to its current injection changes. Hence, metrics based on the equivalent impedance
seen from resource’s terminals or from the point of interconnection (POI) have been traditionally used
as a proxy to quantify system strength. An important limitation of these metrics is that they do not provide
a full picture of the stability and dynamic behaviour of the converter during and after fault conditions.
Other than SCR, the dynamic behaviour of the converters is determined by their operating point
(MW/Mvar output), the internal layout and characteristics of the plant (e.g. transformers impedances,
length of cables, additional voltage support, etc) and their control strategy and parameters.
A new Advanced Short-Circuit Strength metric has been developed by EPRI to supplement SCR based
methods and to provide more information regarding potential IBR control instability to be taken into
consideration before conducting time-consuming EMT simulation studies. The main objective is to
bridge the gap between the required level of modelling and the practicalities of analysis. This new metric
is completely analytical and has been tailored to assess the transient stability margin of the converters 3.
As such, it provides a very fast screening technique that can be implemented in on-line environments
as decision support tool for real time operation. This screening method is capable of flagging conditions
that are prone to transient instability and to quickly evaluate the effectiveness of preventive or corrective
control actions. The only data requirements are positive sequence impedance of the transmission grid
(as per standard load-flow and short-circuit models) and a limited number of parameters from the IBR
controllers. A detailed formulation of the metric, including validation against EMT simulations, is
presented in [R37].
In conventional transient stability studies, the maximum fault duration which can be tolerated by a
synchronous generator to remain stable is referred as Critical Clearance Time (CCT). CCT is highly
related to the fault type, location and operation point of the synchronous generator. A similar concept
can be applied to IBR.
A PLL is normally used in IBRs to track the angle of its terminal voltage. Whenever there is some angle
tracking error, the PLL adjusts its frequency and angle output so that the angle tracking error can be
reduced to zero. When a fault occurs close to an IBR, it can happen that the frequency of the PLL
continues to increase or decrease during the fault and will fail to lock back onto the grid frequency even
after the fault is cleared. When this occurs, the output power and terminal voltage of the IBR can
experience large oscillations and will eventually be tripped by its protection devices.
In brief, the new advanced screening method calculates the IBR-CCT, defined as the maximum duration
that a three-phase fault can be maintained at the POI before the PLL loses the ability to re-synchronise
with the system voltage upon clearance of the fault. The aim of the metric is not to analytically represent
the entire dynamic state relationships of all controls. Instead, the calculation method includes a near-

3
Transient stability of IBR in this context refers to the capability of the PLL to remain “synchronised” to the grid frequency during and after close-in
faults in the transmission grid.

64
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

accurate representation of the trend of these dynamics while reducing the mathematical model
representation, so that a stability margin index can be obtained analytically. These necessary
simplifications introduce a reasonable degree of conservatism in the screening results. For example, it
is intentionally assumed that the IBRs have a lower degree of robustness in their fault ride through
behaviour. It should be noted that, besides the various controller settings, all the information needed for
the IBR-CCT calculation can be obtained from the power-flow solution of the pre-fault system and
without any dynamic simulation [R37].
The calculated IBR-CCT can be compared against the expected fault clearing time te (combined
protection and circuit breaker operation time) to determine the Stability Margin (SM) as shown in
Equation 10 below. A negative SM indicates an unstable scenario whereas a positive SM indicates a
stable case. The higher the SM, the more secure the case is.
100(𝐼𝐵𝑅_𝐶𝐶𝑇 − 𝑡𝑒 )
𝑆𝑀 = Equation 10
𝑡𝑒
The new advanced metric has been recently applied to assess weak grid conditions in the US [R39]
and Ireland [R38]. In both cases, the analyses have highlighted the weak correlation that exists
between the SCR/WSCR and the IBR-CCT. A higher SCR/WSCR does not always guarantee a higher
IBR-CCT. In the Irish example, the investigations revealed cases with SCR in excess of 15 that had
IBR-CCT lower than 60 ms. This observation suggests that SCR alone is not a sufficient indicator of
transient stability risks in IBRs.
Comparisons of EMT simulations against analytical calculations of IBR-CCT showed reasonable
accuracy, within 30 ms, and always on the conservative side [R38]. An example is illustrated in Figure
2.5-6 below where the IBR-CCT calculated analytically was 128.7 ms. The plots on the left (a) show a
stable case in blue with a fault clearance time of 101.5 ms and a marginally stable case in red with a
fault duration of 104.2 ms. The plots on the right (b) show an instable case with a fault duration of 130.8
ms.

Figure 2.5-6 EMT simulation results of three-phase fault at Wind Farm POI to validate IBR-CCT screening
method – (a) blue curve: fault duration = 101.5 ms, (a) red curve: fault duration = 104.2 ms, (b) red curve
fault duration = 130.8 ms
The advanced stability margin (SM) index can provide more information regarding potential IBR
instability in addition to the SCR/WSCR methods for screening the risk of transient stability in IBRs. This
is because SM (IBR-CCT) is tailored especially for transient stability evaluation considering not only the
system strength but also transient stability mechanism and associated control parameters. However, it
should be emphasised that the SCR/WSCR indices should also be used as a high-level indicator of
other types of instability or operational issues that may not be captured by SM. Thus, this advanced
metric is not intended to replace but to complement other SCR based screening methods. Further, it is
understood that many inverter vendors may have more robust and proprietary fault ride through control
algorithms that can improve the actual IBR-CCT whose performance cannot be captured by the SM
metric. However, by assuming a conservative fault ride through behaviour, the SM metric can help
provide system planners with more information regarding potential IBR behaviour and flag scenarios
where EMT analysis would be required.

65
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

2.5.4 Comparison of screening methods

A high-level comparative assessment of the three screening methods discussed in Section 2.5 is
shown in Table 2-13. As can be seen each method presents unique opportunities and limitations which
means that depending on each specific application in hand the use of one method may be more
appropriate.
Table 2-13 Comparison of three screening methods
Summary of screening AFL Improved MIIF Advanced short-circuit
methods for wide-area strength
EMT studies
Key objective When phasor-domain The approach is an Identify operation
studies are not adaptation of the MIIF conditions prone to IBR
adequate and EMT approach presented in transient instability
studies are required previous WG B4-41 to following a close-in
cover VSC technology. fault.
Corroborated against No No No
field measurements?
Corroborated against No Yes No
EMT simulation with
(* corroborated against
vendor-specific
EMT generic models)
models?
Inverter/power system No No Yes (generic behavior
dynamics accounted captured)
for?
Generally more More pessimistic More pessimistic More pessimistic
optimistic or
pessimistic compared
to actual time-domain
results?
Predicts multi IBR No Yes No
control interactions?
Most suitable for Fast calculations of Multiple IBR in a wide On-line screening
when EMT studies are complex area providing decision
a must support for real time
operation
Least suitable for Determining the risk of When comprehensive Identify other forms of
site-specific control network models and converter instability
interactions data are not available. (e.g. multi IBR control
interactions)

66
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

3. Model adequacy
3.1 Importance of vendor-specific site-specific models
Vendor-specific and site-specific EMT models are often necessary when conducting wide-area EMT
dynamic studies. Historically, vendor models and site-specific parameterization were considered for
determining control interaction and harmonic analysis. Generic models or vendor-specific PDT were
usually considered for dynamics studies. Specific applications of generic EMT models are discussed in
Section 3.2. The use of generic models beyond their intended applications negate the benefits of using
EMT tools over PDT tools and require extra time and effort to complete. Specifically, the use of generic
models representing plants in the local study area where a contingency is applied can lead to misleading
conclusions, more so than if such generic models are used in places far from the study area.
With the increasing flexibility and capability provided by EMT tools, vendor-specific IBR models could
be as detailed as a one-to-one representation of the actual installed plant, if required. IBR converter and
prime mover controls and hardware are typically the most complex aspects. These controls are different
depending on the manufacturer, vintage of products, configurations for local requirements, etc. A
simulation model can have different levels of details, depending on how it is implemented. For these
reasons, vendor-specific models are generally developed by respective OEMs due to the level of
knowledge and details required, e.g., using actual IBR control source code. Intellectual property and
country specific requirements are among the reasons why an EMT model from one OEM could behave
differently to that provided by another OEM for the same IBR type, e.g., type 4 wind turbine. In multi-
vendor solutions such as PV or battery or HVDC links, this requirement becomes essential or mandatory
to determine any plant performance. Hence, with reduced system strength, the necessity and credibility
of vendor and site-specific models cannot be overlooked in conduction network interaction analysis.
Figure 3.1-1 shows the comparison of a vendor specific EMT model against an actual product lab test.
The vendor-specific model demonstrates a close match to the actual equipment behavior.

Figure 3.1-1 Comparison between EMT model and product test

For the same product from the same OEM, vendor-specific models could respond differently for different
sites for the following reasons:
▪ The grid codes are different all over the globe. To meet the requirements of different grid codes,
the same product may be configured differently.
▪ The electric grids for the product that is connecting to are unique. For instance, a heavily series
compensated grid may require the activation of special controllers within the product, to reliably
operate.
▪ Within the same regulatory region, the controller parameters of the same product could be set
differently for different sites, as the system strength could be very different seen at the PoC for
each site, and/or the neighboring system characteristics could be quite different because of the
existence of other devices.
The use of a vendor-specific model by itself does not therefore guarantee an accurate outcome.
Practical experiences exist where the use of vendor-specific models without exact representation of site

67
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

parameters could give rise to unreliably optimistic or pessimistic results, hence incorrect decision
making [R40] [R41]. The following examples are noted:
▪ In long cable projects, e.g. offshore wind farms, unacceptable responses including system
instabilities might be experienced if vendor-specific site-specific models are not used.
▪ The absence of vendor-specific site-specific models might mean that adverse interactions between
multiple IBR are not identified.
▪ Vendor-specific and site-specific models may demonstrate certain design constraints, or flexibilities
such as allowance for fluctuations in DC link voltage which cannot be accounted for in generic
models.
▪ The use of models other than vendor-specific site-specific models for determining technical
performance requirements for a specific IBR might result in either over-investment in capabilities
that would not have been required in reality, or inability to achieve study outcomes in practice and
the possibility of a non-compliance with agreed performance requirements. This need will become
more and more pronounced for connections under low system strength conditions.
▪ In a plant solution with multiple vendors, as is often the case of modern PV plants, considering
vendor-specific models parameterized against site values is necessary to perform dynamic and
interaction analysis. Site-specific parameter optimization leads not only improvement in the plant
response but also considers unaccounted network behavior leading to constraints.
In summary, the use of vendor-specific site-specific models is essential for all studies where the specific
performance of an IBR or cluster of IBRs, and impact on overall power system is the key point of interest.
Note that vendor-specific models generally impose a higher computational burden compared to the
generic model. The time taken for each study should be considered beforehand, and judicious decision
to be made on how many simulation runs to be conducted with the vendor-specific site-specific EMT
model.

3.2 Application of generic EMT models

Generic models are less detailed than the vendor-specific models and are not tailored for a particular
IBR product or site. These models are developed with predefined characteristics and specific intended
purposes in mind. The objective is to represent a particular control strategy and selected control loops
within the overall IBR control. The absence of actual control logics might lead to a very different outcome
to that obtained by a vendor-specific and site-specific EMT model. For example, the use of generic EMT
models for low system strength conditions might cause the model not to initialise or conversely result in
too stable a response.
This does not necessarily imply that the performance of generic models would be always unacceptable
when compared against actual equipment tests, especially for higher system strength conditions. It may
be possible to parametrize a generic model to reasonably match the performance of actual product for
some disturbances and grid conditions at a given point in time. However, it should be noted that generic
model responses may no longer be accurate for boundary grid conditions that cannot reasonably be
validated at time of model tuning (major disturbances, unexpectedly low system strength, etc.) or if
additional plant with an adverse impact on system dynamics is installed nearby in subsequent years.
Basic generic EMT models are generally included in EMT tools as examples. However, no internationally
agreed generic EMT models exist at time of writing this TB. They are widely used by research institutions
and universities, and for long-term initial planning studies for areas of high IBR penetration, when no
further information is available. While it is a common practice for IBR OEMs to develop or at least
parametrize generic PDT models, the same generally does not apply for generic EMT models. CIGRE
WG C4.60 is currently addressing this gap and developing, validating and benchmarking generic IBR
models across various EMT software platforms. The outcome of this WG will be publicly available in
2024.
Various use cases where application of generic EMT models is appropriate are described next.
3.2.1 Long-term planning studies
3.2.1.1 Initial scoping studies
The timeframe of interest for long-term planning studies ranges typically from five years to several
decades. The objective of these studies is to assess the adequacy of the transmission grid for future
scenarios and to evaluate alternative expansion plans. These studies are usually carried out with power
flow and fault level analysis only, although PDT dynamic analyses are sometimes used to investigate

68
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

long-term dynamic issues (for example the need for dynamic reactive support) and/or to compare the
robustness of alternative expansion plans.
More complex analyses are also performed on a case-by-case basis, depending on the characteristics
of the network and the planned reinforcements. For example, a network expansion plan comprising
series compensation of transmission lines will normally require sub-synchronous resonance (SSR) and
sub-synchronous controller interaction (SSCI) studies using detailed EMT models. Similarly, for long-
term expansion plans involving a large amount of HV/EHV cables, detailed EMT studies of TOVs and
harmonic resonances are maybe needed to provide a feel on the likely issues to be encountered.
Recently, transmission network owners and independent system operators are increasingly conducting
dynamic studies to have a better basis on which to make decisions considering fundamental differences
between synchronous and IBR technologies. The main areas of concern are a reduction in system
inertia and system strength, with associated risks to power system security. As an example, this concern
is addressed in European Union by legislation that sets requirements on Transmission System
Operators (TSOs) to coordinate the dynamic stability assessments and to conduct a common study per
synchronous area to assess the need for a minimum level of inertia in the region [R42]. In addition,
various EU funded research projects, led by TSOs, have looked into long-term scenarios with up to
100% IBR penetration, e.g. MIGRATE [R43] and EU-SysFlex [R44].
Considering the potential for rapid technological changes in the timeframe of interest and uncertainty
regarding the precise location, type and make of the generation to be connected, the use of site-specific,
vendor-specific dynamic models is not relevant. Generic models in the form of PDT dynamic models are
therefore the most practical approach. However, increased uptake of IBR and reference to some of the
new and emerging phenomena, would likely mean an increasing need for either enhanced PDT models
or generic EMT models for long-term planning studies, depending on the phenomena of interest.
3.2.1.2 Legacy plant with no information available
There are many ways to model various legacy plants, where these legacy models are required for EMT
studies, and have been requested from the Generator owners. In order of priority, the preferred options
are:
▪ Vendor-specific EMT model of the plant provided by its OEM.
▪ A similar model from the OEM of the plant with the same technology, parametrized to reflect site-
specific settings.
▪ A generic model of the correct plant type, for example type 3 or type 4 wind turbines, parametrized
to the extent possible to reflect the capabilities and limitations of specific plant make and site
settings.
Other elements to be modelled include the associated network support devices and relevant protection
functions that might behave differently under reduced system strength conditions, or due to adverse
interactions with other plant.
When utilising generic models, it is recommended to use a model that has the following features, so it
can be tuned to approximate the measured response of the legacy generating system:
▪ Ability to set LVRT and HVRT thresholds.
▪ Ability to set steady state setpoints (e.g. voltage, reactive or power factor setpoints).
▪ Ability to adjust active and reactive power recovery ramp rate.
▪ Ability to limit fault current injection to a specified level.
▪ A generic model of central park level controller if the generating system uses such a controller.
 The general data of the generic model such as plant type, voltage and MVA rating are known
from the data sheet information provided by the vendor. As much as possible, available
information should be adopted for the generic representation of the plant.
 Legacy plants are likely to have been in service over a number of years. Thus, their response
under various system events were likely captured by digital event recorders during their
operation. If such information is available, one possible option is to adjust the control and
protection parameters of the generic model so that the model response may closely follow the
waveforms captured during system event.
▪ All static and dynamic reactive power support plant modelled in addition to the generating unit
models.

69
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

The remainder of this section describes a process for constructing an EMT model for an existing
windfarm, in this example a DFIG type, which has been connected to the network for some time but with
limited information available. In particular, documentation for the control systems of the power electronic
converters is incomplete. There are certain steps that can be followed that will help configure a generic
model to represent the performance of the existing windfarm:
▪ Collect disturbance recordings for a wide range of operating conditions
▪ Collate technical information on the WTG using published literature if necessary
▪ Determine rating limits for WTG and Converter
▪ Locate protection settings
▪ Understand what level of accuracy is needed
High quality disturbance recordings, especially sinusoidal data, are invaluable for this work. However,
sometimes only RMS data are available; clearly when using RMS data the filtering and measurement
delays must be accounted for. Ideally the EMT model will be matched against a range of recorded WTG
outputs but if unavailable the model performance must still be checked over a range of outputs. Generic
models typically possess a limited number of adjustment parameters, so it is important not to “perfect”
the response at one operating point only since this may be detrimental elsewhere.
As the legacy plant ages, its technology may be superseded relaxing the barriers on the publication of
useful technical information. It is worth checking with the OEM whether public domain material is (or
could be made) available.
Normally the EMT model is an aggregate of the complete windfarm, i.e. a single, LV connected WTG
will be scaled up before grid connection. Many rating values for the actual WTG units are available but
often in different formats. It is recommended to use a consistent pu scaling method when defining the
rating limits, noting that the power electronics has only a fraction of the induction machine’s rating.
Although the main protection settings are usually available, care must be taken when interpreting them.
For example, under and over voltage thresholds may be given in pu but check whether they apply to
line-ground, line-line or average voltages etc.
3.2.1.3 Wind Turbine Generator Model
Refer to Figure 3.2-1, which demonstrates a concept for scaling up the output from a single WTG to
create an aggregate windfarm model. It is recommended that the EMT model uses actual parameters
from the legacy plant for the single induction machine (IM) and its low voltage (LV) to medium voltage
(MV) step up transformer. If this data is unavailable, engineering judgement can derive reasonable
default settings. The output from the single WTG is connected to the high voltage (HV) connection point
via a second step up transformer with similar low rating as the WTG transformer. In parallel, the WTG’s
output current, i_wtg1, is multiplied by n (total # of WTGs - 1) to drive a current source which feeds the
main power transformer. This method of aggregation, where the current source connects via a separate
transformer can be more numerically stable than injecting into a single main transformer but it may
involve additional scaling should other plant use the WTG’s MV busbar (See also Section 3.3.4).

Figure 3.2-1 Concept of aggregate windfarm model

70
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

3.2.1.4 Power Electronic Converter Model

Accurate representation of the power electronic converter, principally its control systems, is the main
challenge with this activity since these control systems dominate the dynamic performance of the
windfarm and in particular its FRT response. It is not easy to replicate the performance of commercial
controllers since their requirements are so demanding. There are several linked factors that drive this:
▪ the use of Graetz bridge converters may erode rating margin during unbalanced disturbances
▪ the protection margins are typically tight due to high converter cost
▪ control systems need high bandwidth to operate with tight protection margins
▪ extreme care must be taken with high bandwidth converters due to interaction with network
resonances
These points highlight the often mutually exclusive design challenges and explain why commercial
controllers tend to be highly specialised and contain multiple, often customised, control loops. It is
unlikely that a generic EMT model can match best commercial performance – some compromise will
probably be necessary. Where performance differences are unavoidable, knowledge of the connected
network is helpful in guiding an appropriate level of modelling conservatism rather than allowing the
production of optimistic results. For instance, if capacity (spinning) reserve is of primary concern a
slightly slower recovery response may present a less optimistic and more acceptable performance error.
3.2.1.5 Balance of Plant Model
Windfarms may include other assets that influence the overall dynamic response:
▪ High level SCADA systems and their control delays
▪ STATCOMS and/or synchronous condensers
▪ Capacitor banks and their switching logic
▪ Tap-changer controls for the main power transformer
The challenges in modelling these assets often relate to determining the control co-ordination and
effective communication delays.

3.3 Large-scale inverter-connected generators

A detailed EMT model of an IBR could comprise some or all of the following sub-systems:
▪ Aerodynamics
▪ Pitch control
▪ Mechanical drive train model
▪ Electrical generator
▪ Measurements
▪ Converter control and protection systems
▪ Semiconducting device switching scheme, e.g. pulse width modulation for IGBT switches
For example, as shown in Figure 3.3-1 a wind turbine model will account for all these sub-systems
whereas a solar inverter model will include the electrical energy source, measurements and converter
control. Different OEMs apply different practices when comes to modelling PWM switching, and a
detailed representation may be discarded without loss of accuracy.

71
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 3.3-1 Type 3 WTG model sub-systems

The inter-relationship between these sub-systems is shown in Figure 3.3-1 where subscripts s, r, el, and
LS denote stator, rotor, electrical and low-speed shaft.
3.3.1 Prime mover
The prime mover model is not an integral part of the overall IBR model. Several IBR models have been
found to exclude the prime mover model without loss of accuracy. For wide-area EMT studies the
aerodynamics and pitch controller may be neglected if they do not impact the accuracy of stability
studies.
Prime mover modelling for wide-area EMT studies is very much design dependent and judicious
decisions on whether or not to model it, is generally made by the respective OEM. Measured results
from the actual equipment response in practice would be a good benchmark to confirm the extent to
which the neglect of these aspects might impact the accuracy.
3.3.2 Power electronic switching
Explicit modelling of semiconducting switching devices such as insulated gate bipolar transistors
(IGBTs) within each IBR, will increase the computational burden significantly. This is particularly
important for wide-area EMT modelling where tens or hundreds of IBR models are used. To address
the speed of simulation issue whilst maintaining comparable accuracy, the ‘’average switch’’ approach
has been adopted by most vendors where detailed switching of the converters is represented by voltage
(or current) sources. Whether fully representing the IGBT switches or using an average switch model
the following will remain unchanged:
▪ IBR control structure
▪ DC-link voltage response
▪ Mechanical and aerodynamical components (if relevant)
Figure 3.3-2 shows responses of detailed and average switch EMT modelling for a wind turbine subject
to a network fault. It is shown that the simulation results obtained from the average model are
comparable to those of the detailed model in terms of accuracy while the simulation time is
approximately an order of magnitude faster [R45].

72
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Ptotal [MW]
0.85
2

0.8 0
UDC [kV]

5 5.2 5.4 5.6 5.8 6

0.75

0.7 2

Qtotal [MVAr]
0
0.65

-2
0.6
5 5.2 5.4 5.6 5.8 6 5 5.2 5.4 5.6 5.8 6
Time [sec.]
Ustatorrms [kV]

0.4

0.2

0
4.99 4.995 5 5.005 5.01 5.015 5.02

4
Istatorrms [kA]

2
1

0
4.99 4.995 5 5.005 5.01 5.015 5.02
Time [sec.]

Figure 3.3-2 Dynamic response of wind turbine with full (red) and average switch models (black)

3.3.3 Control systems

In any wide-area EMT studies focusing on plant stability, it is critical to correctly represent the plant
internal control systems in detail.
Figure 3.3-3 shows detailed functional block diagram representation of the WTG control system as
implemented in an EMT model of the turbine used for wide-area EMT studies.

Figure 3.3-3 Converter control model for type 3 turbine used for wide-area studies
In the figure:
▪ UL: stator line voltage
▪ UDC: dc-link voltage
▪ UTQ: q-axis voltage at tertiary winding of the turbine transformer
▪ IGDQ,UGDQ: d-q axis grid current and voltage
▪ IRDQ,URDQ: d-q axis rotor current and voltage

73
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ UGAB,URAB: grid- and rotor-side phase voltages

▪ SV-PWM: space-vector pulse width modulation
▪ Omega, Gamma: angular velocity and angular displacement of the rotor
Looking at Figure 3.3-4 the following sub-systems can be identified:
▪ Rotor-side converter control
 Rotor side current control
 Active/reactive power control
o Power control
o Slope limiters
o Filters
 Drive train torsional damping
▪ Grid-side converter control
 PLL
 DC-link voltage control
 DC-link current control
▪ Low voltage ride-through control logic (LVRT)
Additionally, correct representation of the turbine protection system including the dc choppers is
necessary for wide-area EMT studies.
As an example, a detailed control representation of the rotor-side converter control is shown in Figure
3.3-4. This includes the inner-loop rotor-current control, and outer-loop power controller which in turn
consists of active/reactive power controllers, drive train torsional damping, and the power filters. The
PQ filters assure that the power reference for the PQ controller will not have discrete jumps in case of
a sudden change in external power reference. The function of drive train damping is to damp out the
generator speed oscillations near resonance frequency of the drive train by injecting power into the grid.
This function is included by adding a component onto the active power control output as shown in Figure
3.3-4. As seen in Figure 3.3-4 the drive train damping has two blocks. The first block is a high-pass filter
acting on the speed difference between the turbine and the generator. The second block is the drive
train damper.

Figure 3.3-4 Model of the rotor-side converter control for a type 3 wind turbine including drive train
torsional damping

3.3.4 Model aggregation

Traditionally there has been a one-to-one correspondence between power system elements such as
generating units and the models of these elements in simulation software. Thus, each generating unit
has been represented individually in the power system model. This is practical with conventional
synchronous generator power plants, where the plant is a large power station comprising typically of

74
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

only a few large turbine-generators. However, IBR can include as many as several hundred individual
generating units. As these generating units are usually identical to one another, this has the effect of
multiplying the required computational effort and simulation run time for little benefit, compared to
representing these identical generating units as a smaller number of aggregates.
For such generating systems comprising tens to hundreds of small generating units, the general rule is
that the submitted plant model should be so arranged as to aggregate identical units into aggregated
generator models on aggregated collector network models. So for example, a plant that may have many
tens of units and is comprised of two different technologies of generating units, can typically be
represented by two equivalent aggregated units on separate aggregated feeder models, one for each
type of technology. In the simplest cases, a single aggregate may suffice to represent the entire plant
where the plant consists of a single technology (e.g. a plant with one hundred units of the nominally
identical type of WTG).
As stated above, aggregation should generally not be used to combine power system elements of
differing types. For example, if a plant consists of twenty full-converter WTGs from one manufacturer
and twenty type 3 WTGs from another, then two aggregated generator models are needed, one for each
technology.
In very simple terms, the aggregation takes the form shown below in Figure 3.3-5. In this example, a
simple plant with one type of inverter-based generation technology is assumed so that it can be
represented by a single aggregated unit and single aggregated equivalent feeder. This concept can
easily be expanded to represent multiple aggregated units where groups of feeders need to be
aggregated and/or there are multiple types of devices in a hybrid-plant. In essence the IBR unit is scaled
up in MVA to represent the total IBR MVA in the plant, this is the aggregated single unit scaled model.
Similarly, the unit transformers that typically steps up the low-voltage at the terminals of the IBR (e.g. in
the range of 500 to 1000 V) to the collector system medium-voltage level (e.g. in the range of 20 to 60
kV) are also scaled up by MVA into an equivalent aggregated unit.
Special attention must be given to the aggregated representation of the MV ‘collector system’ that
connects the MV terminals of the generating unit transformers and (usually) conveys the aggregate
generated or consumed power to an MV collector bus at the relevant substation.
In the simplest case, all identical generating units are combined into a single aggregate, and the model
specifies a single equivalent collector impedance connected between the MV collector bus and the MV
terminal of the aggregate equivalent generating unit transformer. In this case, the recommended
procedure for calculating the equivalent collector impedance is given in [R45] and [R46]. This procedure
is based on calculating the equivalent series resistance and reactance that yield the same active power
and reactive power consumption as the original MV collector system, where the units in that system are
assumed for simplicity to all operate at identical voltage. Other aggregation methods that consider the
potential variation in generator terminal voltage do exist, such as described in [R47].

230kV
0.6 kV 34.5 kV Aggregated
34.5 kV
To Point of Common
Coupling with Utility

Collector System Transmission

Aggregated Feeder Model Line Model
Generator
Model
Scaled Substation
Substation
Pad Mount Transfomer
Shunt(s)
Transformer

Figure 3.3-5 Aggregated Inverter-Based Resource (IBR) plant model

Presented below are the main characteristics that the aggregate models must include. A wind farm was
used as an example, but the same concept can be extended to photovoltaic parks.
▪ Representation of the medium voltage network
▪ Explicit modeling of the large substation power transformer(s)
▪ Control systems associated with wind turbines
▪ Control systems associated with the power plant controller
▪ Equivalent (scaled-up) model of the pad mount individual unit transformers

75
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ Harmonic filter, if any

▪ Explicit modeling of any shunt reactive devices at the substation
▪ Connection system (transmission lines, reactors, ...) with the utility at the point of common coupling
▪ Associated protections (wind turbines and wind plant level)

3.4 Network
3.4.1 Static network elements
3.4.1.1 Node-Breaker vs. Bus-branch modelling
The technique of maintaining a full substation topology representation (including circuit breakers) in a
simulation model is referred to as node-breaker modelling. Historically, this representation has not been
widely supported due to its increased computational requirements to explicitly represent the substantial
number of additional voltage nodes it introduces to the simulation.
Instead, in several commonly used power system modelling tools, network topologies for substations
may not reflect the actual layout of multiple bus-bars present and may neglect the presence of circuit
breakers and isolators entirely. Extremely low impedances between bus bars mean that for load-flow
and transient stability study purposes, multiple bus bars within a substation can effectively be considered
as a single node to simplify the simulation model without loss of fidelity. This results in an entire
substation being represented in simulation as a single bus bar, with transmission lines, transformers
and other plant connected all to this common node. This approach is known as bus-branch modelling.

Figure 3.4-1 Node-Breaker modelling (left) vs Bus-Branch modelling (right)

Bus-branch modelling is a valid approach whether PDT or EMT transient dynamic stability modelling is
being performed. In the case of EMT modelling, this approach may aid in reducing simulation time due
to a resultant simplified network admittance matrix that need not represent switchable elements that
may never change state in the simulation run, and hence result in more efficient simulation processing.
However, the simplified bus-branch representation and associated divergence from the system topology
can bring challenges when it is used in wide-area EMT studies.
3.4.1.1.1 Impossible operations
Many terminal stations may have configurations which result in singular circuit breaker operations
affecting a group of elements. When node-breaker modelling is used, it is clear and intuitive how the
operation of a circuit breaker will affect the in-service status of network components. When bus-branch
modelling is used, topology information is lost, and the user may inadvertently perform an operation that
simply cannot occur, invalidating the simulation results.
In the example shown in Figure 3.4-1, it is clear from the node-breaker representation on the left image
that the left-most line reactor can be switched out with the line itself remaining in service, however the
line cannot be taken out of service whilst keeping the reactor in service. When this substation is
simplified into a bus-branch representation on the right, such a limitation is not obvious. The end user
may inadvertently remove the line from service whilst keeping the reactor in service – an impossibility.
Whilst the example shown here is trivial, where complex inter-trip and control schemes exist, limitations
of system configurations are critical to be correctly represented in simulations.
Using a node-breaker representation to include models of circuit breakers and other switched elements
in the wide-area EMT simulation would indeed avoid such mistakes from occurring, but this comes at
the cost of significantly increased simulation computation time and increased resource usage to
represent the possibility of a step-change event being actuated. Simpler alternatives can be considered
to avoid incorrect operation by a user whilst still maintaining the speed advantage of bus-branch
modelling:

76
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ Where supported, the inclusion of an image of the substation topology alongside or underneath the
bus-branch representation may be sufficient to prompt users to correctly consider operation
outcomes.
▪ Some EMT simulation tools allow node-breaker representations, but with circuit breakers
configurable to disallow operation during runtime. Instead, circuit breakers with disallowed runtime
operation are replaced with a short or open circuit prior to model compilation and execution,
allowing a bus-branch representation in the mathematics, but node-breaker in appearance.
3.4.1.1.2 Inability to map between non-static asset identifiers
Reliably modifying a large EMT model to match a given scenario (typically from a load flow or SCADA
snapshot) is a key requirement of repeated wide-area EMT studies. When a case is imported into an
EMT environment from another modelling tool, there may not be a direct mapping of network assets
between the modelling platforms. This is especially an issue where the EMT tool uses a static layout,
but the external tool allows multiple elements to be collapsed into a single element (e.g. many-to-one
bus mapping, where the unique identifiers such as the bus number change depending on circuit breaker
configuration).
Failure to accommodate changing element identifiers will result in an inability to automatically re-initialize
the wide-area EMT case from the external modelling platform, leading to increased manual intervention
and time spent recreating a specific dispatch scenario. Although this might be acceptable for limited
sets of studies, if wide-area EMT models are to be used in earnest, a more robust solution must be
deployed, such as:
▪ Use of advanced modelling packages supporting a many-to-one topology identifier mapping
▪ Intermediate import scripting or translations
▪ Use of mating external cases to precisely correspond to the wide-area EMT model only
Superficially, the issues discussed above may seem trivial, however experience has shown that failure
to address these issues in a robust and consistent manner may result in large amounts of time wasted
or fundamentally incorrect EMT simulations being performed.
3.4.1.2 Transmission line modelling
Most EMT tools generate automatically frequency-dependent Bergeron line models from the
corresponding PI model when converting load flow model of the wide-area network from phasor-domain
to the EMT tool. Bergeron models generally provide sufficient accuracy for wide-area EMT studies when
the focus is on power system stability issues with a dominant frequency within the range of inverter
control bandwidth typically up to 200-300 Hz.
However, these models are not generally suitable when the inter-relationship between power system
stability and power quality is important, e.g. risk of exciting a network resonance frequency and
subsequent disconnection of an IBR due to excessive temporary over-voltages. Accurate modelling of
transmission lines or any other power system components, when the dominant frequency of a given
phenomenon exceeds the inverter control bandwidth mentioned above, is outside the scope of this TB.
See [R48] and [R49] for further details on modelling network components for frequencies beyond typical
inverter controls bandwidth.
Experience indicates that when modelling power systems with several IBR models the numerical
integration time step of those IBR models is typically the key determining factor in the overall simulation
time required. As such replacing Bergeron models with more detailed frequency dependent models such
as geometrical models will have minimal impact on the overall simulation time. Therefore, when there
is a desire to maintain one network model for both stability and harmonic resonance and interactions
studies, geometrical lines models can be used instead of Bergeron models for all studies. This is more
from a convenience perspective rather than providing added accuracy.
3.4.1.3 Cable modelling
The discussion above on transmission line modelling, and the use of more detailed (geometrical) vs less
detailed (Bergeron) models is generally valid for cable modelling except that different types of models
will be required to account for frequency dependency of cables compared to those commonly used for
transmission lines.
3.4.1.4 Transformer saturation characteristics
It has been a common practice for several decades to account for non-linear transformer saturation
characteristics when conducting conventional EMT studies involving dominant frequencies of several

77
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

hundred to several kilo Hertz. Such studies often include a significantly smaller part of the overall
network.
However, this level of modelling has not been applied nor has often been needed for wide-area EMT
studies. This is because the focus of wide-area studies is primarily on system stability rather verifying
equipment design from a high frequency response perspective.
These non-linear characteristics can sometimes have pronounced impact on IBR response during fault
conditions. This cannot always be accounted for with an PDT model. The impact of non-linearity can be
described as follows:
At the time of fault inception some residual flux remains in each main leg of the transformer core. The
maximum remnant flux occurs when the transformer current is interrupted during saturated operation.
The resulting residual flux will dissipate very little until fault clearance. Depending on the size of voltage
dip and point on voltage waveform, the normal flux can reach up to twice its value during steady-state
conditions. The combination of high normal flux and presence of residual flux could cause a large inrush
current upon fault clearance. The consequence is that while in reality sudden changes in voltage are
somewhat maintained by the transformer magnetisation characteristics, the inability of the PDT-type
model to account for this behaviour results in a sudden step change in IBR voltage upon fault clearance.
Criteria on whether or not to explicitly model transformer saturation is the MVA size of the transformer
under consideration compared to that of other nearby plant and to network fault level at that mode, and
whether an inrush mitigation logic such as point-on-wave switching or pre-insertion resistors are used
in the given transformer. Energising large transformers without any inrush mitigation logic could result
in severe voltage dips across the network with the potential to falsely invoke low-voltage ride-
through function of those IBR.

3.4.2 HVDC and FACTS

Whether voltage source or current source converters, HVDC links and FACTS controllers’ model share
many common features with those for inverter-based generators such as wind turbines, solar inverters
and battery energy storage systems. Vendor-specific site-specific models of HVDC links and FACTS
controllers are therefore equally important to those for inverter-based generators to ensure accurate
outcomes as far as the power system as a whole is concerned.
The following additional factors should be considered when dealing with models of LCC HVDC links.
▪ LCCs require large filters for harmonic mitigation and reactive power support. These filters have
ancillary control systems and all must be correctly modelled.
▪ LCCs are theoretically more susceptible to issues with low system strength; accurate modelling is
needed to determine their consequent performance limitations.
3.4.3 Protection systems
Protection relays are a fundamental part of the overall power system’s performance and should be
included in the dynamic model representation of complex plant. EMT-based simulations can support the
accurate representation of protection relays (including vendor-specific, algorithm-specific
representation) due to the necessary quantities often being inherently available (i.e., voltage and current
waveforms) and short simulation timesteps. Furthermore, the activation or failure to activate of a
protection mechanism can have material impacts on the overall outcome of a wide-area EMT simulation.
For some time, it has not been common practice to include the protection mechanisms of plant within
the vendor-specific EMT model unless a specialised study or particular failure mode is being
investigated. However, as systems are experiencing higher penetrations of IBR with more complex
interactions between devices, more modern wind turbine, solar and battery systems are increasingly
representing their firmware-based protection functions within vendor-specific models.
Discrete plant relays, network asset relays and synchronous generator protection relays may not be as
readily available, but depending on the study being completed, their inclusion in a wide-area EMT model
may be highly necessary, as described in the following sections.
3.4.3.1 Component-level protection
3.4.3.1.1 Synchronous generators
While it is not often considered essential to model synchronous generator pole slip protection in detail
in EMT studies, there are times when explicit representation of this protection function might be required.

78
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

For example, when relatively small, low inertia synchronous generators are likely to materially impact
the system strength available to inverter-connected generators, and their disconnection could have an
adverse impact on power system security and performance of EMT studies.
When assessing longer duration faults, especially with synchronous generators providing a material
system strength contribution having a critical clearing time less than the fault clearing time.
Modelling other protective functions for synchronous generators might become necessary, primarily
under conditions with relatively high steady-state voltages and relatively low steady-state frequency.
Operation under such conditions could result in abnormal behaviour of the following protection functions:
▪ Under-excitation (loss of excitation).
▪ Voltage/hertz (V/Hz).
Until now, it has not been a common practice to account for the response of these types of synchronous
generator relays. This partly stems from the presence of many synchronous generators in the network
meaning that each will have a smaller burden to control network voltages during low demand conditions,
and to withstand resultant RoCoF when a normally interconnected power system becomes islanded. A
substantial reduction in the number of online synchronous generators means that very few units could
be dispatched at any given time, and as such more exposed towards their boundaries of operation, or
even outside their design envelope. As such correct representation of these relay types will assist in
answering the question of whether or not a synchronous generator will remain connected under those
extreme operating conditions which will occur more frequently under scenarios with high IBR
penetration. Note that the same considerations will largely apply to a synchronous condenser as well.
3.4.3.1.2 Transmission lines and transformers
Distance and differential relays are the most widely used types of relays in transmission systems, and
frequently in the distribution systems. Over-current relays are typically used in the distribution system
but not so often in the transmission systems.
Of these three relays types over-current relays are most susceptible to scenarios with high IBR
penetration and very few synchronous generators online. However, these relays are also the simplest
to represent in power system stability analysis, and due to their diminishing use in the transmission
systems they are not the most critical relay type for the purpose of this Technical Brochure.
On the other hand, differential relays are being increasingly used in the transmission and distribution
systems, and sometimes even as main 1 and main 2, or main and back-up protection for the same
equipment. This stems from their lower susceptibility to changes in generation mix.
As such modelling impedance-based protection relays such as distance, and out-of-step protection
forms the most important part of modelling protective relays for system impact studies in power systems
with high IBR penetration. The focus shall be first on locations experiencing the highest changes in
available fault level and equivalent system impedances caused by changes in the generation mix.
Another important but less widely used relay type are those acting on negative-sequence current during
fault conditions, e.g. directional relays. IBR provide a very different and controlled negative-sequence
contribution as opposed to the inherent response of synchronous machines. In recent years, some
countries such as Germany have implemented mandatory requirements enforcing IBR to provide similar
negative-sequence current injection to that of a synchronous generator.
Key criteria on determining if a transformer differential protection including associated inrush control will
need to be modelled include:
▪ Presence of large levels of second and fifth harmonic during transformer energisation, especially
when they are not well damped.
▪ The MVA size of the transformer under consideration with respect to that of other IBRs and network
transformers
▪ MVA size of the transformer with respect to the network fault level in MVA
▪ Presence of several other nearby transformers and the possibility of sympathetic inrush
3.4.3.2 System-level protection
Appropriate modelling of network protection and control schemes that result in a power reduction
following a credible contingency is required. Examples of such schemes include:
▪ Fast or slow disconnection of some or all generating units.
▪ Fast or slow power runback across some or all generating units.

79
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ Topological schemes resulting in a significantly different response depending on the network

conditions, for example, anti-islanding schemes.
▪ Special protection schemes and emergency frequency control schemes, including under-frequency
or under-voltage load shedding, that might be activated under a scenario studied.
The modelling of these schemes provides a broader assessment of power system security, and will also
be used for determining power system security implications under prior outage conditions.
3.4.3.2.1 Criteria for inclusion in the models
All system-level control and protection schemes associated with generators need to be represented,
regardless of the level of power reduction they might cause.
Additionally, all schemes for network and other generation that are part of the power system being
modelled in EMT studies need to be represented if their operation causes the total reduction of
generation or load in the region to exceed the size of the single largest credible contingency.
Modelling network-wide control and protection schemes should not be limited to the network ownership
boundaries.
3.4.3.2.2 Modelling methodology
In general, network control and protection schemes are triggered in one of two ways:
▪ Measurement of analogue quantities (under-frequency load shedding, out-of-step protection); and
▪ Discrete events (circuit breaker (CB) status, plant trips).
If modelling of network control and protection schemes is required, the following approaches could
apply:
▪ For measurement-based schemes:
 Use generic dynamic modelling of the control or protection relay where available 4.
 Use an automated script that simulates the outcomes of the control or protection schemes.
 Use an automated script that assesses the outcomes of the control or protection schemes
post simulation.
 When automated scripts are used to assess the operation of a network control and protection
scheme, if it is determined that the scheme would have operated, the simulation should be re-
run, including all actions that would occur on operation of the scheme in that instance.
▪ For event-based schemes:
 Use an automated script to simulate the event and outcomes during the simulation.
 Manually determine the sequence of events and perform actions based on worse-case timing
delays (less preferred).

3.5 Loads and Distributed Energy Resources (DER) modelling

3.5.1 General load and DER modelling
Historically loads were represented as ZIP model (polynomial static load model) and DER were not
modelled for power system stability studies. This is because in the past, utility-scale plant constituted
the majority of generation with only a small contribution coming from DER. When DER levels were low,
the behaviour of DER had relatively little impact on power system stability and could be managed within
the range of normal uncertainty in the modelling process. Difficulties in characterising the DER response
to system disturbances and developing a sufficiently accurate aggregate model (as seen by the
upstream transmission network along with the presence of numerous makes and settings), have meant
that state-of-the-art dynamic models of the DER did not have the same level of maturity and accuracy
compared to generator dynamic models.
In recent years the dynamic load model “CMLD” was developed by the Western Electricity Coordinating
Council’s (WECC) Modeling and Validation Working Group (MVWG) [R50] [R51]. The CMLD
aggregates dynamic and static power system loads and lumps them at the distribution feeder level. The
model consists of six load components, whereby each of these are described in detail in NERC
documentation. Figure 3.5-1 shows this model aimed to represent the aggregated dynamic behaviour

4
Note that the proposed use of generic models as opposed to vendor specific models might be perceived as an inconsistent approach compared
to those discussed for primary power system equipment modelling where the use of vendor-specific models is a necessity. The proposed
approach stems from WG members’ experience that vendor-specific models of protective relays are not generally available from the respective
manufacturers even in the form of blackbox models.

80
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

of the DER and load in RMS stability studies (synonymous with PDT studies in the context of this
brochure). This model represents the combined (aggregated) dynamic behaviour of many tens to
hundreds of small distributed IBR on the distribution system, for example for commercial, industrial or
residential distributed photovoltaic generation. The same applies to the load model. However, in this
case four different types of motor loads, an electronic load and a static are provided to distinguish salient
differences between key load types. Some or all these components can be used as necessary. The
DER and load model interface with the wider distribution network model via aggregate impedances as
shown in the figure.

1:T
Rfdr + j Xfdr Motor A

jXxf Bf1 Bf2 Motor B

Bss
Motor C

Motor D

Electronic
DER
Load

Static
Load

Figure 3.5-1 DER and load model aggregation

3.5.2 Large variable speed motor drives

The motor A-D modelling are not strictly speaking intended for large variable speed motor drives in the
order of several tens of MW. These variable speed motor drives are increasingly connected through
voltage source converters, similar to those used in inverter-connected generators and, therefore, have
broadly the same limitations and capabilities.
However, their impact on grid level dynamic response will depend on the MVA rating of the plant. Unless
the plant is relatively large (tens of MW), the grid impact maybe minimal and may be represented inside
an aggregated load model. If the plant is relatively large, the power electronic inverters may have to be
represented with vendor models, specially, if interactions with other inverters in the local area is a
concern.
Explicit modelling of inverter-connected motor drives is generally required when:
▪ Connection of a sufficiently large variable speed motor drive which can be operated as both a
generator and a motor.
▪ Connection of a sufficiently large variable speed motor drive near existing assets, with potential
adverse interactions and further degradation of system strength if a detailed model of variable
speed motor drive is included.

3.6 The need for more detailed DER modelling

Further efforts are in progress worldwide to continuously improve the understanding of the DER
response and to improve currently available PDT models for the DER [R47], [R52]. The use of an EMT
tool allows more accurate representation of control systems used in distributed IBR, and correct three-
phase representation of power system components and response to unbalanced disturbances, as
shown in Figure 3.6-1.
A DER EMT [R53] model should represent the response of power electronic inverter interfaced devices
in a distribution feeder. Thus, the DER load model is developed to emulate the aggregate behavior of

81
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

many distributed generators in the distribution system and the collective controls, and protection
capabilities therein.
The control concept is based on decoupled (dq control) current control methodology, commonly adopted
in transmission and distribution level inverter applications. The control system generates the d and q
axis components of voltage reference signals. These are then converted to phase domain (A-B-C),
voltage reference signals with the inverter bus voltage angle (derived through a PLL module), used as
the reference. The reference voltage signals are appropriately scaled (considering the terminal AC
voltage) and the VSC is represented by a controlled voltage source.
The following protections and trip functions should be included in the model.
▪ Under voltage trip
▪ Overvoltage trip
▪ Rate of Change of Frequency (RoCoF) trip
▪ Staged frequency tripping for over frequency and under frequency
▪ Voltage phase angle jump trip

Figure 3.6-1 Comparison of DER model response between PDT and EMT simulation for a 100 ms fault
resulting in a residual voltage of 0.7 pu

82
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

3.7 Model visibility/transparency/portability

Although there are plethora of technical aspects to consider when developing wide-area EMT models
for areas with high penetration of IBRs, this is also an area that has important non-technical aspects to
be considered. This section discusses several such aspects.
3.7.1 Model visibility and confidentiality
Vendor specific EMT models are usually extremely detailed and may include full representations of the
Control and Protection software and, in some cases, embed a full copy of (part of) the proprietary control
software installed on the field.
Original equipment manufacturers (OEMs) are extremely concerned that their IP may be compromised
if the Intellectual Property (IP) information of their products could be obtained via the EMT model.
The EMT models are software and as such may be very easy to replicate and be subjected to possible
cyber security threats. Depending on model contents, the consequence of inadvertent model
mishandling may be extremely grave for the OEM, considering the highly competitive advantages that
are related to the IP created with decades of research and development.
It is understandable that manufacturers consider the protection of the IP contained within their EMT
models an extreme priority.
The priority and reluctance are even higher when the OEM is requested to share their detailed EMT
models and potential competitor should receive the models, even when an independent third party (e.g.
an ISO/TSO) may be owning both of the assets and require the model sharing to assess interoperability
issues.
Model visibility, confidentiality and respect of IP is an issue that must be given the same care and
consideration as the other technical matters considered in this Technical Brochure considering the
extremely significant impact that can have.
3.7.1.1 Black boxing and confidentiality
A common approach is to “black box” an EMT model along with providing the controlling software for
the model in a compiled format. This approach encrypts or compiles sensitive portions of a model
(typically the control system and/or parts of the electrical primary systems) such that it is not accessible
by unauthorized users without impact on the quality of the simulation model.
The approach of black-boxing is being increasingly supported and refined by common EMT software
developers, and in many cases, with a full support of such functionality that strongly simplifies the
process of encryption. A commonly seen approach from OEMs is to use the source code which is
wrapped in a translation layer that allows interfacing to the EMT model components. Confidentiality of
the source code is maintained by pre-compiling the controller and translation code into a binary library
file (with any common library dependencies) ahead of model release, which the EMT tool can call and
pass data to and from during runtime. Although this is not perfect protection of IP, it certainly makes
reverse engineering considerably more complex.
This approach has an added advantage of simplifying the number of code versions and types the OEM
needs to maintain, as updates to in-field firmware can be directly included into the EMT model, without
the need to translate the code into another, EMT tool-compatible, language. In this approach, the end
model user can understand the basic plant topology but will have little to no visibility (and hence,
understanding) of the EMT model’s control system.

83
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 3.7-1 Common approach for decoupling of control elements from EMT tool
Several manufactures have still found merits in black boxing the complete model, hardware included.
The merits are not only with IP protection, but also with modelling philosophy. In such cases, alongside
the controller source code, manufacturers may also have represented their plant/device hardware by
writing a hardware representation in a codebase compatible with the target EMT tool. Again, this
software and hardware representation is then pre-compiled into a non-human readable binary library
file ahead of model release, which the EMT tool can call and pass data to and from during runtime. This
approach may have further advantages to IP protection and ease of code versioning, in that running
hardware and software model representations within the same micro-environment could lead to model
runtime speed improvements. The end model user has even less visibility (and hence, less
understanding) of the EMT model’s representation of plant. This may be a positive or a negative
attribute, depending on the perspective and problem at hand.
Another approach to handle the decoupling of the control elements and the physical assets has been
proposed by ENTSOe for HVDC systems in line with the requirements of the European Network Codes
for HVDC [R54]. The aim of the work is to support the interoperability for a secure operation of the future
European energy system and the realisation of interaction studies for multi-vendor AC and multi-vendor
multi-terminal DC systems.
This approach includes the separation of the control system into different layers. The document
proposes an interface for Hardware in Loop (HIL) as well as for Software in Loop (SIL) simulation for
the different levels of the control as shown in Figure 3.7-2.

Figure 3.7-2 Control layers of an HVDC station [R54]

3.7.1.1.1 A question of trust
As black boxing methods obscure more and more the plant representation, the challenge of balancing
model visibility shifts to a question of integrity on behalf of the user of the model: at what level of black
boxing/obfuscation does the trust in the model’s representation of plant break down? That is, a
completely black-boxed model would certainly provide the best chance for OEM IP to be protected, but
the model end-user may find themselves in a position where they need to know more about the model's

84
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

internal behaviour to diagnose and understand the scenario they are analysing. For example, in a
scenario where low system strength is resulting in inter-plant oscillations, it would be important for the
end-user to have visibility of the internal control system references and the PLL output to understand
the source of instabilities or at least to have access to different pre-defined potential tuning of the control
system. Providing such detailed channels as an output to the end user needs to be considered as a risk-
based option, e.g., risk of IP being reverse-engineered versus risk of a customer’s plant being denied
connection or operation due to inadequate model visibility.
Combining these factors, the challenge for the industry becomes how to optimize the balance between
respecting the OEM’s right to protect their intellectual property versus the need for model users to be
able to trust and verify model behaviour. This is a vexing issue to address as it is both subjective and
often dictated by the legal frameworks in place in the region where the model is developed and used. In
the context of large-scale EMT studies, one potential solution to allow the secure use of EMT models
by multiple competing external parties is shown in Section 4.2.1.2.1 but it is not a ‘one-size-fits-all’
approach, and is still an area of ongoing evolution.
3.7.1.1.2 Black box model longevity and source code in escrow
Much of the above discussion considers the acquisition and use of vendor specific black boxed EMT
model at the time of connection, planning, or incident investigation. It assumes a computational and
commercial environment that remains static for the lifetime of an EMT model’s usage. However, this is
seldom the case.
It is important to recognise that software and simulation environments continually evolve at a rapid pace,
and that software versions (in particular, compilers) may only be supported and accessible for a few
years at a time before both support and sale is ceased. Furthermore, OEMs who provide the models
may become defunct, lose key staff and knowledge, or simply choose to cease operations in a region.
This leaves ISOs and generator owners in a precarious position when it comes to being able to provide
a current and stable EMT model to analyse system stability over the lifetime of the plant in question
(which may be in excess of 30 years). This is because as the software environment changes, the tool
used for EMT studies may no longer be able to compile, read or otherwise interface with the originally
supplied black-boxed EMT model, as backwards compatibility is not guaranteed over such long
timeframes.
Consider an example where a black-boxed EMT model was supplied for a new, large wind farm
connection in 2015. At the time, it was tested and confirmed to be both stable, and a sound
representation of the actual plant. As part of the model structure, it referenced several controller library
files on disk, which are included in the overall model executable at compilation and execution time of
the overarching simulation. The black-boxed model and its constituent components expect to be
compiled and executed in a specific version of a language / environment – a perfectly reasonable
requirement as it would be impossible to predict how the language and environment are to evolve in the
upcoming years.
It is now the year 2022, and major changes have occurred to the computational environment. The
compiler version originally specified to be able to compile and run the model in the EMT simulation
software is long out of support and can no longer be purchased. Attempts are made to compile and run
the model using the latest version of compiler and simulation environment software, but the model fails
to execute - and if it did execute, the difference in underlying environments may mean that its output is
unverified and untrustworthy. How can the situation be resolved without simply excluding the whole
model from wide-area EMT simulations (and potentially invalidating the simulation response)?
It is at this point where the OEM would be approached to recompile the model in the latest versions of
software, and test/verify that it remains an accurate representation of plant (and would likely need to be
compensated to do so). But what if the OEM becomes defunct or otherwise is unable to access or
provide the source code?
This is an evolving and increasing problem being seen by early adopters of wide-area EMT modelling
and users of UDM PDT models. The solution that is increasingly being considered is for the ISO or other
key party to request both the model and all its underlying source code at the time of submission.
However, given the potential for IP theft or dissemination, rather than the source code being held by the
ISO or plant owner, it is held in escrow by a trusted third party with a proven history of longevity and
cyber security, to only be released on the condition that the OEM no longer exists, has ceased
operations in the region, or is otherwise unable to provide the model source code on reasonable request.
This approach can at least allow the generator owner, ISO or their representative to independently
inspect and recompile the EMT model to continue to support system security into the future. However,

85
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

it is also important to recognize that the associated financial costs of designating a trusted third party
holding the source code may be a challenge for all parties involved.
3.7.1.1.3 OEM perspective
OEMs are extremely cautious on the visibility of black-boxed EMT models, especially if the OEM has
spent decades of R&D and investments on its technology. Requesting the OEM to expose internal
signals usually takes tremendous effort and involves evaluating risks, engineering, legal organizations,
and possibly more. It is not uncommon that the OEM could disagree with exposing the requested
signals. The risks of IP exposure may be determined high, and the distribution of models is not always
under OEM's jurisdiction. It should be acknowledged that different OEMs have different methodologies
on how and what to black-box in their EMT models. These varieties create limitations and features
among OEMs’ EMT models. For instance, some models may only work with a specific time step, while
others may allow flexibility to choose various time steps. A fully black-boxed EMT model that includes
control system and electrical components is built based on the IP protection perspective, as well as
considering the model accuracy.
An OEM with a relatively large span in the market usually requires providing models in various software
environments since each customer has a different simulation tool and requirements. Further, creating
models for the global market is a cumbersome task. Special care is taken when interfaces with the
specific EMT software, as each EMT software may have its unique feature that the model is requested
to provide. The model behaviour needs to be assessed as well across multiple platforms and validated
against site measurements. These black-boxed models are used for several purposes ranging from
planning to grid compliance studies to interaction studies. Further, the software environment is rapidly
changing, and newer versions of various tools no longer adequately support the black-boxed models
produced in earlier versions or vice versa. This results in additional efforts on parts of the OEMs to
deliver a compatible version of its models. However, there are ongoing initiatives to help the OEM to
provide models with a standard interface so they can be more easily interfaced in various software. It
aims at ensuring the interoperability over the software (and OEM equipment) lifecycle since the
proposed API will be supported by the software vendors.
The concept of creating generic DLLs compatible across various tool platforms has been gaining some
momentum. IEC 61400-27-2:2020 [R55] provides a brief description of the creation of generic software
interfaces which may allow the usage of models in different software environments. Although the
standard is for time domain RMS models (synonymous with PDT studies in the context of this brochure),
the interface proposed is independent of simulation domain. The interface description is based on C-
Code implementation since this is the most common programming language. However, there is no
restriction on the programming language to be used on which the interface can be implemented. An
essential part of DLL used for various simulation tasks is creating a flexible interface that can handle
the requirements of different simulation environments. The Extended Simulation Environment interface
(ESE-interface) claims to meet these requirements. This relatively new concept needs further
investigation and extensive testing in a complex model environment to validate its viability. Developing
such a generic DLL interface would ease backward compatibility, help interoperability between various
tools, and ensure a correct model response.

86
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

4. Large-scale EMT simulation

4.1 Network development
Broadly speaking, there are two approaches to develop a large-scale EMT model for simulation.
The first and most labour-intensive is to manually create each element of the model from detailed asset
datasheets to create a 1:1 match of the physical system in an EMT model. This is likely to yield the most
complete and appropriate models for the study purpose (e.g., if system restoration studies are the intent
of the model’s use, geometric line data and transformer saturation characteristics need to be captured),
but it may be a very labour intensive and error-prone data entry exercise. This option is only preferred
where there is no appropriate PDT model already available, or if the EMT model is to be used for highly
detailed studies such as system restoration, TOV, or harmonic studies.
The second method is to import an existing PDT model into an EMT package, allowing the EMT import
tool to substitute PDT components with an equivalent EMT equivalent as appropriate5. Once complete,
the user will replace (manually or through automatic replacement libraries) the EMT components that
will have material impacts on the outcome of a specific study with true EMT components (e.g., generator
dynamic representation, complex load models). This is likely the most common approach used by wide-
area EMT model users who already have appropriate and high-confidence PDT models available and
are intent on performing dynamic studies to evaluate general system and generator performance in
areas of high IBR penetration.
In either case, a wide-area EMT model that is to be used to evaluate the dynamic performance of a
system with high IBR penetration will most likely need to be set up to match a specific load-flow scenario
either from historical SCADA data or a future planning case. Various EMT tools handle this initialisation
differently. Some may allow the setup and solve the load-flow within the EMT tool itself, whilst others
will allow the importation of key settings from a native load-flow tool (e.g., load values, generator export
values, component in-service statuses and transformer tap settings) to allow a matching load-flow
condition to be reached after initialisation. However, the end goal remains the same: The post-
initialisation steady-state of the EMT simulation should match the desired load-flow conditions of a
reference load-flow case.
4.1.1 Load flow case creation
Load flow cases can be set up in PDT simulation tool, where a steady state load flow solution can be
calculated very quickly. When studying a real power system operation condition, a load flow case can
be setup by extracting data from state estimators in SCADA systems. When studying a power system
planning scenario, the following factors must be examined in setting up a load flow case which will later
be considered in EMT simulation:
▪ All devices must operate in the designated operating zone. For example, generator active and
reactive power must be within its capability and controller limits. Network support devices, such as
SVCs and STATCOMs, must operate within the reactive power range. It can be challenging to
setup voltage control devices with droop control, if the selected PDT load flow platform is incapable
to represent droop in its load flow engine. In this case, manual calculation might be required to
force a fixed voltage and reactive power operating point for such voltage control devices.
▪ All measurement must be taken from the same location between PDT load flow models and EMT
models. For generators controlling power factor, especially wind farms and solar farms with large
reticulation systems, the generators should be set to operate with designated P and Q output so
the power factor setpoint is achieved at the connection point, and the same equivalent reticulation
system must be used in both PDT load flow models and EMT models.
▪ Negative loads sometimes are used load flow cases, which could be representing embedded
generation. It is important to note that negative loads in PDT may be translated into voltage sources
in EMT, which may cause erroneous results for dynamic studies. All fictitious voltage sources
should be replaced with their generator representation, or if that is impossible, removed (noting this
will negatively impact the match between PDT and EMT simulations).

5
It is important to recognise that the EMT import tool cannot create more detailed representations of PDT data, as more detailed representations
would naturally require additional data not captured in the PDT model. That is, the EMT model import tool cannot “make up” the missing
information. Rather, the EMT import tool will use EMT-native components populated with the simplified PDT data (e.g., a transmission line
imported from a PDT model will likely be replaced with some form of a PI-section model).

87
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

4.1.2 Dynamic model initialisation

The EMT simulation can retrieve load-flow solution data from a separate software (such as PDT) and
use it for initialization of its network to achieve fast steady-state in time-domain. The EMT software may
also and ideally, contain its own load-flow solution method, in which case a multiphase and unbalanced
load-flow solution is used.
Network initialization is an essential factor for solving large networks in time-domain. With the size of
the studied networks and the large number of synchronous machines and control functions, the network
model should be started as close as possible to its operating steady-state conditions.
Some simulation time must be allowed for EMT models to be initialised to the designated load flow
condition.
4.1.2.1 Synchronous machines initialisation
In the load-flow solution, the synchronous machines are typically specified by a PV (power and voltage)
constraint. The following automatic steps can be applied for the synchronous machines in an EMT
simulation:
▪ The load-flow solution provides all synchronous machines with the terminal voltage phasors.
▪ The steady-state solution uses internal steady-state model to initialise all mechanical and electrical
variables from the terminal voltage phasors. Unbalanced conditions are also considered by
injecting equivalent negative sequence voltages into the network.
▪ The steady-state solution field voltage Efss signal becomes available as a bundled observable
signal and used for automatically initialising the machine control diagram.
The last step is illustrated in Figure 4.1-1, where Efd_ic is actually Efss, Vm_rms is found from the
steady-state voltages. The initial stabilizer output signal Vaux is normally zero. The hold function is used
to fix the reference voltage Vref for the following time-point solutions. A similar approach is used to
initialise the governor/turbine model where the initial conditions are established from the computation of
the steady-state mechanical power or torque and steady-state speed.

Figure 4.1-1 Initialisation and dynamic blocks in an EXST1 exciter

The steady-state solution automatically initializes all state variables, including propagation history
buffers on transmission line models, and starts the time-domain simulation.
4.1.2.2 IBR initialisation
It is possible that certain models for inverter-connection resources, such as wind farm, BESS and solar
farm models, require very long initialisation time due to power electronics and related controls. This
could increase the overall simulation time significantly if not checked, and will distort the final steady
state operating point of these models, and the EMT model can settle at a different steady state condition
other than the one setup previously.
One solution to this is to switch out inverter-connected resource models from the power system model,
and only represent them with voltage sources, which will produce exactly the same active power and
reactive power from these inverter-connected resource models as in the load flow condition. This will
help the EMT power system model to reach a steady state quickly. In the meantime, the real inverter-

88
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

connected resource EMT models will be initialised separately by their own individual voltage sources,
and the voltage magnitude and phase angle of these individual voltage sources are taken from the
connection points of these inverter-connected resources in the power system model. Once the EMT
power system model has reached steady state, the real inverter-connected resource model will be
switched back into the power system model, and because it has been initialised at the same voltage
and phase angle, the transition will be very smooth with only negligible disturbance to the steady state
of the EMT power system model. This is shown in Figure 4.1-2.

Figure 4.1-2 Isolated EMT generator initialisation technique. Left: Prior to steady state. Right: Post steady
state.
In summary, the following steps can be followed, for an IBR, a wind-turbine for example:
1. Specify the load-flow constraint of IBR based on type of control, PV constraint for example.
2. Find the load-flow solution which gives the voltage phasor at the point of interconnection (POI)
of IBR.
3. Connect an ideal voltage source which is given the magnitude and phase found in the above
step at the POI for the time-domain solution and let it connected for a specified interval (may
be adjusted manually). This allows to have the correct load-flow solution on the network side
and to allow to the IBR to initialize its control system and converters.
4. Switch off the ideal voltage source after the specified interval and leave the IBR connected to
the grid. This step will create a small transient, but will eventually establish the correct steady-
state.
The above automatic steps can be applied for fast initialization of IBRs, but much better results, such
as initialization timings below 200 ms can be achieved if initialization rules (initial conditions) and also
added into the IBR controls and components. The control-blocks can be encapsulated and initialized
backwards from the computed steady-state solution. Parameters, such as wind-turbine slip and
mechanical input can be also calculated automatically.
4.1.3 Network equivalencing
The network model utilised for EMT studies should include representation of a sufficiently large part of
the network, comprising components such as transformers, transmission lines, HVDC links, and network
reactive power support devices such as reactors, capacitors, and FACTS controllers within a defined
study area. The size of the study area also depends on available computation capability, which can be
a major limiting factor.
Judgement is, therefore, exercised to determine the extent of the network that can have an appreciable
impact on the EMT study results. This section discusses key considerations that need to be made when
deciding how much of the wider network to represent in an EMT simulation, and how to present the rest
of the power system through network equivalencing.
A wide-area model of the transmission network, in which the EMT study is to be conducted, is generally
recommended. In most cases, extending the network by a few busbars into the adjoining AC-connected
networks is necessary to obtain sufficiently accurate results. Extension of modelling into a neighbouring
network is essential for EMT simulation focusing on power systems in the vicinity of two or more
boundaries, including direct current (DC) links.
In any case, dynamic models of all synchronous machines or grid forming inverters providing
appreciable system strength to the study area should be included, and different combinations and

89
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

dispatch patterns of such devices need to be considered. Where other inverter-connected resources
receive system strength support from those synchronous machines or grid forming inverters, they should
be included in the wide-area power system model.
A trial-and-error approach can be used to determine the location of boundary buses. The location of a
boundary bus is considered sufficiently accurate if expanding the network further by explicitly
representing associated dynamic plant does not materially change the results of the analysis.
Consideration should also be given to the types of stability phenomenon of interest, when determining
appropriate location of the boundary buses.
For EMT studies, the following outlines non-exhaustive possible network equivalent options for
modelling the wider network.
4.1.4 Static voltage sources
The network equivalents include voltage sources that maintain the power flow from the boundary
locations. The internal impedance of the voltage sources is fixed and is derived based on network
equivalencing methods. Typically, the voltage source magnitude and the internal phase angle are also
fixed, calculated based on a specific network operating condition. Thus, if changes are made in the
study area of the model (such as tripping of a line to clear a fault), the voltage sources (network
equivalent) will not maintain their pre-disturbance power flow levels. This can impact the line flows in
the study area of the model. The impact is generally negligible, when the network boundaries are located
a few buses away from the location of the disturbance. However, this can be a concern in specific
situations. One approach to ensure that the voltage sources re-adjust magnitudes and angles is to
represent them as ‘controlled’ voltages sources. The voltage magnitude and the phase are dynamically
adjusted to maintain the pre-set power flow level.
A static equivalent can be represented by a voltage source with an appropriate impedance, but it does
not capture dynamics associated with the wider network.
As noted above, depending on the determination of a static network equivalent location, material impacts
on study outcomes can occur. Inappropriately located static network equivalents can mask interaction
issues and result in unrealistic study outcomes. Conversely, excluding synchronous machine dynamics
that provide appreciable system strength support could give rise to unrealistically pessimistic results.
Static network equivalents are currently by far the most widely used and simplest network equivalencing
method.
Lastly care must be taken when using fixed voltage sources as even a single fixed voltage source with
internal impedance could generate fictitious active power during frequency disturbances, and reactive
power during voltage swings and erroneously accentuate voltage excursions.
4.1.5 Dynamic voltage sources
The underlying assumption for the use of static voltage sources is the network boundaries are electrically
far from the study area and the disturbances applied to the study area will not cause appreciable
changes to the boundary bus voltage (magnitude and angle). It is assumed that the response of dynamic
devices in the ‘external network’ will have negligible impact on the dynamic response of devices in the
study region.
However, there may be situations where the above assumptions may not hold. If the voltage at a
boundary bus changes appreciably (magnitude and angle), the post contingency power and reactive
power injection from the voltage source may change from its pre-event value (as the voltage source
internal magnitude and angle are set to fixed values). In such cases, moving static voltage sources
further away until when no changes are observed in the voltages of boundary buses or the use of a
dynamic equivalent voltage source would be required.
A dynamic equivalent consists of an electrical source that aims to represent the aggregate dynamic
behaviour of the power system behind it.
Two approaches may be adopted if deemed necessary under particular circumstances.
▪ Adopt a controlled voltage source: In this approach, the internal voltage magnitude and phase angle
are changed (using a simple PI control approach or more detailed approach) to maintain the voltage
source power and reactive power injection to the pre-event value (or to a defined set value). This
will ensure that the post event power flows in all parts of the system are accurate as per the
generation dispatch levels at the time of the event.

90
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ Represent the external network using a synchronous machine with generic AVR and turbine-
governor: In this approach, the machine data and the AVR and turbine-governor data should be
selected with due care. The machine parameters should fall in the range of typical data. The
reactances should be adopted to reflect the expected short circuit contribution. The control
parameters maybe adjusted to match known expected dynamic responses. The model tuning in
this case is predominantly a trial-and-error process when the study engineer should use sound
engineering judgement.
4.1.6 Frequency dependent network equivalent (FDNE)
The network equivalent source impedances can be derived through a matrix reduction technique where
the full system nodal admittance matrix is reduced with only the buses in the study area and the
boundary buses explicitly retained. Typically, the impedances are calculated at the system fundamental
frequency. This is a sufficient representation for most wide-area stability studies. While standard
software tools allow the user to derive ‘frequency dependent’ network equivalents from the available
network data, the frequency dependent network equivalent representation in not often an important
detail for large scale system studies. The passivity concern described in Section 2.2.2 will be more of
an issue in the presence of improperly conditioned network data.
It should be noted that the model network boundaries are generally located at least a few buses away
from the main study area. The principles of deciding where to put boundary buses are similar to those
discussed in Sections 4.1.4 and 4.1.5. whereby boundary buses are extended until the point where no
appreciable difference can be observed in frequency domain characteristics with the extended boundary
bus and the boundary bus chosen prior to that.
While FDNE is not often a pre-requisite for wide-area stability studies, it provides an opportunity for
streaming data and models for different types of studies. It will also be a useful tool when higher
frequency super-synchronous interactions are the point of interest. These interactions could be
occasionally caused by control systems of multiple nearby IBR, but more often the interaction between
control system and system resonances.
As such some TSOs/ISOs apply FDNE as a standard option in their wide-area EMT model. For example,
the frequency response tool is used extensively in some networks such as Hydro Quebec for
determining network equivalences and the extent to which the network can be reduced without loss of
accuracy.
Frequency responses can be used to evaluate network reduction. The results are presented in Figure
4.1-3 for 735 kV and 315 kV substations electrically far away from each other. The L and R networks
are those presented in section 1.3.7. It is apparent that the similarity of responses indicates acceptable
behavior considering the elimination of low and medium voltage networks. The worst case is occurring
at Substation 4 which is electrically closest to the location of the lower voltage equivalent.

91
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 4.1-3 Frequency response (positive sequence impedances) for L-Network (orange) R-Network
(blue)
Note that the combined use of FDNE and dynamic equivalent voltage sources is the most accurate but
also the most involved modelling approach. As such this approach has not been applied to date for any
practical power system studies under high IBR penetration scenarios. However, the need for such an
approach cannot be ruled out as power systems worldwide are getting closer to 100% IBR operation
scenarios over the coming years.
4.1.7 Comparison of different static voltage source network equivalencing
points
EMT simulations have been conducted using models with different static voltage source network
equivalencing points, to observe the difference in model response to the same disturbance. The
simulation attempted to replicate a real generator loss event.
The EMT models used in these simulations were from the Australian NEM network, and include:
▪ A 2-state model, which includes Victoria and New South Wales power systems in detail, with South
Australia and Queensland represented as static voltage source equivalent.
▪ A 3-state model, which includes Victoria, New South Wales and South Australia power systems in
details, with Queensland represented as static voltage source equivalent
▪ A 4-state model, which consists of all four mainland states power systems in East Coast Australia
Figure 4.1-4 shows the overlay of a synchronous generator MW output in response to the same
generator trip event in three different models. It can be seen this synchronous generator model in 2-
state and 3-state models did not show any post contingency increase in MW output, while the 4-state
model showed the generator output was increased by around 30 MW, showing its governor response
to the frequency deviation caused by a generator trip. The generator response in 4-state model matches

92
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

with the high speed monitor (HSM) measured response quite well, indicating the 4-state model has
achieved the highest accuracy in replicating the real event.

Figure 4.1-4 Comparison between synchronous generator response in different network models
Figure 4.1-5 shows that the simulated frequency in both 2-state and 3-state models was restored to 50
Hz after the generator trip, while the HSM data showed the frequency was only maintained at around
49.8 Hz after the event, which is also demonstrated in the 4-state model.
When analysing the results, it was observed the constant voltage source equivalent in both 2-state and
3-state model provided large amount of MW in response to initial frequency deviation following the
generator tripping event, which not only restored the network frequency, but also prevented real
generator model to raise MW to compensate for the MW loss, while in the real event the network
frequency was not restored to 50 Hz in the same period as the simulation showed, so other synchronous
generators have to provide additional MW based on their governor settings.

Figure 4.1-5 Comparison between transmission bus frequency in different network models
These simulations suggest that modelling the whole power systems would bring the highest accuracy
in simulating power system events, especially for a frequency disturbance, as full network dynamics are
completely captured in such models. While it is still possible to achieve reasonable accuracy using a
reduced-order model, using an appropriate network equivalent is key to achieving the required accuracy.
Static voltage source equivalents are therefore not an ideal option to be used in simulating frequency
disturbance. It is noted that a properly tuned dynamic voltage source equivalent would also likely provide
adequate results for this example, but the challenge would be to balance the efforts required to tune

93
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

such a model (and the confidence on its applicability for all other operating conditions) versus the
computational burden of running a full network model in EMT.

4.2 Co-Simulation techniques

Co-simulation is a modular simulation approach that has been widely used in many engineering fields.
It consists of decomposing a complex simulation problem into various sub-systems that are solved
individually. To compute the overall solution, the sub-systems are then coupled via the exchange of
simulation results at discrete time intervals.
The advantages of co-simulation arise from the individual solution of each sub-system and are primarily
twofold. Firstly, it allows for the parallel simulation of the sub-systems and therefore, better overall
computational performance (particularly in large-scale systems, such as those under consideration in
this Technical Brochure). Secondly, independent solvers with different degrees of modelling detail and
different simulation step sizes can be used for each sub-system. For instance, Figure 4.2-1 shows the
combination of a detailed vendor-specific EMT-model of an HVDC system, typically accounting for the
internal dynamics in a time range of a few microseconds to milliseconds, with an PDT model of the
transmission network, which accounts for the system-wide dynamics in the time range from a few
milliseconds to several seconds. Hence co-simulation can play an important role as bridging technology
towards a pure large-scale EMT simulation.
Cross-platform co-simulation, i.e. the use of different software packages for each sub-system, is not yet
widely used in the electric power industry. This is partly due to a lack of standardization in the
communication and simulation control protocols, causing cross-platform co-simulation to require a non-
negligible effort to configure, as opposed to a “plug and play” solution. Dedicated, high-performance
protocols do exist in some modelling packages. For the integration of ‘real code’, interfaces are under
development to support the integration of digital controllers to existing dynamical models [R56]. Further
standardization effort is expected to address the increasing demand for cross-platform co-simulation,
not least to support co-simulation between interdisciplinary tools.

Figure 4.2-1 Application case of a hybrid co-simulation in power system

Convergence of the solution is guaranteed if no algebraic loops exist between both sub-systems, i.e. in
the coupling equations [R57]. In this case, the accuracy of the co-simulation is equal to that obtained
via the solution of the individual sub-systems. This is the case if the system is split at boundaries with
inherent time delays, such as long transmission lines. If the wave travelling time between both ends of
the line (or penstock) belonging to different sub-systems is longer than the co-simulation synchronization
time, there is no algebraic coupling equation because of the dead time, and convergence of the solution
is guaranteed. In other words, the inherent system delays between the sub-systems impose an upper
bound on the synchronization time of the co-simulation with the consequent reduction of the simulation
step size, unwanted accumulation of numerical error and last but not least, negative impact on
computational performance. Another application case for a system split at an inherent time delay are
long penstocks in hydraulic power plants, when coupling the electric and mechanical system in the
turbine model.

94
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

In the more general application case however, the system cannot be split at boundaries with inherent
time delays. In such cases, stability of the co-simulation can be only guaranteed by iterative coupling,
which can be computationally intensive or not supported by the coupling scheme (i.e. by the simulation
control options in the software package). Nevertheless, acceptable accuracy can be still achieved
without iterative coupling through network equivalencing, or in a more generic formulation, through
numerical treatment of the inputs and outputs by the coupling scheme. For power system co-simulation
this translates into single- or multi-port network equivalents used to represent the coupled sub-systems.
In the most simplistic approach, the network equivalent consists of an equivalent impedance of the
coupled subsystem at power frequency, as illustrated in Figure 4.2-2. In the hybrid EMT-PDT co-
simulation, the accuracy can be further enhanced by a frequency-dependent network equivalent that
approximates the behaviour of the coupled sub-system in the range of frequencies of interest. Other
coupling schemes make use of recursive substitution of the equivalent impedance [R58] with very good
results. This is a field of continuous research and more sophisticated dynamic system equivalents can
be expected in future. The challenge is to increase accuracy of the co-simulation whilst avoiding the
introduction of additional (artificial and non-existent) resonances by the network equivalents in the
coupling scheme.

Figure 4.2-2 Co-simulation of two sub-systems. The diagram on the left shows the signal exchange at the
co-simulation boundaries. The diagram on the right illustrates the synchronization of both sub-systems
at discrete times

4.2.1 Hybrid simulation

The use of hybrid dynamic simulation combining the advantages of each of the PDT and EMT
simulations has been recently implemented in some commercial power system analysis tools. These
include hybrid PDT and offline EMT, or PDT and real-time EMT simulation. Such an approach aims at
representing region(s) of interest in EMT simulation and the wider power system in PDT with the two
simulations running in parallel.
This approach provides the opportunity to use existing and encrypted PDT models for some of the
legacy plant where possibly no information is available on their transfer function block diagram
representation or associated parameters. Such a method assists in increasing the overall simulation
speed compared to pure EMT simulation if the numerical integration time step of each of the IBR is
relatively high, e.g. several tens of microseconds. However, in wide-area power system models
comprising several models (requiring time steps of 1-2 µs) this approach may not provide a significant
gain in time-efficiency.
4.2.1.1 PDT-EMT simulation
Hybrid co-simulation requires the conversion of data exchanged between different simulation tools. For
EMT–PDT co-simulation, phasors have to be extracted from the data sent from the EMT domain to the
PDT domain. This is usually based on frequency analysis methods, such as the Fast Fourier Transform.
Conversely, the data transferred from the PDT domain to the EMT domain has to be time interpolated.
For illustration purposes, this section reports on the results of a hybrid co-simulation obtained with a
model of the Texas transmission system, which has been built from publicly available information. The
test case consists of 2000 buses and 432 running generators. The original model was altered to
represent wind park equivalents in the northern area of the grid as detailed fully rated converter wind
turbine models.
For the setup of the co-simulation, the grid model has been split into two areas. The so-called Panhandle
of Texas is a relatively weak area of the grid with relatively high penetrations of PE-based generation in
comparison to conventional generation. For the given test case, no conventional units are in service,
while 627 MW are generated by wind turbines within the Panhandle region. In order to analyze the effect
of these wind turbines in detail, this region is simulated in the EMT domain, while the rest of the grid is

95
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

simulated in the PDT domain (balanced). The four long lines (57 mi, 122 mi, 167 mi and 259 mi)
connecting the Panhandle to the rest of the grid are used to split the network, as shown in Figure 4.2-3.

Figure 4.2-3 Geographical representation of the synthetic network model of Texas, showing the split of
the network for the co-simulation (green area as EMT, grey area as PDT-balanced).
The test case investigates a three-phase short-circuit at a 161 kV busbar, which is located close to the
main wind generation of the Panhandle area. Switching actions to clear the busbar fault after 100ms
result also in the disconnection of one of the major 500 kV transmission line and load shedding in the
area. The fault case has been analyzed in three different time domains: PDT (balanced, 0.10ms step
size), EMT (0.05 ms step size) and PDT (balanced) – EMT co-simulation (0.1 ms step size for PDT and
0.01ms step size for EMT). Simulation results are shown in Figure 4.2-4.
The EMT simulation shows an oscillatory instability with high voltage peaks for the observed busbar.
This oscillatory behaviour is recognized as being caused by the interaction between power electronic
based equipment and the network impedance, resulting in unstable behaviour at a frequency below 1
kHz. These oscillations could lead to grid or unit protection tripping and thus result in a cascading effect.
Though the co-simulation method considers the rest of the network in the PDT domain, the same effects
can be seen in this simulation. Small differences occur within the peak value and time, but the general
behaviour is equally good. The result indicates that the pure PDT simulation cannot capture the
instability and could lead to an underestimation of the problem. It also highlights the importance of
accurate simulation models in the EMT domain, as small differences in the modelling approach may
lead to significant behaviour changes.
However, simulating the whole network in the EMT domain would have a huge impact on the overall
simulation time. With respect to the balanced PDT simulation, the computation time of the EMT-
simulation is 148 times higher. In comparison, the computation time of the PDT-EMT co-simulation is
only 18 times higher. Hence, in the test case, co-simulation can increase the efficiency of the EMT
simulation by a factor of 8, and yield equally good results.

96
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 4.2-4 Transient voltage behaviour for a busbar fault, showing PDT simulation (dashed blue), EMT
(green) and PDT-EMT co-simulation (red). The diagram on the right shows a zoomed scope of the diagram
on the left.

4.2.1.2 EMT-EMT simulation

EMT-EMT co-simulations may be required for a variety of applications. Some are technical in nature
while others are due to requirements of having to incorporate vendor provided models that does not
provide sufficient flexibility to be used in specific simulation environments. Examples include:
▪ Improve simulation speed: This can be achieved by placing different parts of the simulation circuit
in different processors and running each sub system in time step synchronism.
▪ Use multiple time steps: Inverters typically require smaller timesteps for accurate results. The
transmission network on the other hand maybe simulated with a larger timestep. This can easily be
accommodated (with faster simulations) using EMT-EMT co-simulations
 In some cases, the vendor provided models are defined to run only at specific time steps
▪ Models of dynamic devices provided by equipment vendors may not support the software
environment used for the main simulation. In such cases, EMT-EMT co-simulations across different
virtual environments are required to perform a complete system level simulation.
▪ Provide a platform for intellectual property sensitive models to be places in secure environments
with model access restricted to selected users, as discussed in the following sections.
An example of this type of co-simulation can be seen from AEMO who recently has started the
implementation of an Electromagnetic Transient Simulator for reproducing the response of its power
system with large penetration of IBR based on EMT models. The simulator will allow AEMO to perform
decision support, training and testing as fast as possible using RTS [R59] [R60].
One of the most interesting features of this tool is that it allows to include in the co-simulation controllers
components from different vendors using dynamic link libraries (DLL) to preserve the IP of each
manufacturer (Figure 4.2-5 and Figure 4.2-6). This way, is possible to perform quickly analyze network
disturbances an ensure network security. The key performance goal is to reach nearly 3 times real time
to simulate the whole Australian network.

Figure 4.2-5 Different components that can be imported as DLL files in an IBR generation plant

97
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 4.2-6 Scheme used to perform EMT-EMT co-simulations with multiple targets
4.2.1.2.1 Application for maintaining confidentiality
EMT-EMT simulations can be used to overcome the paradox of maintaining confidentiality yet allowing
users to interact with confidential models (i.e. use but not have). This method can be used to address
the “trust” issue raised in section 3.7.
4.2.1.2.1.1 Example 1 – Single connection point exposure
In this approach shown in Figure 4.2-7, two mating simulations are run side-by-side in separate
computing environments, where one environment contains and runs confidential EMT models but is
secured from external access, and the other environment contains only an external user’s model and a
time-series data streaming network interface to the instance running the confidential models. In this way,
the arrangement acts much like a SMIB study, however the “infinite bus” is actually a detailed
representation of the full network. The image below illustrates this arrangement.

Figure 4.2-7 EMT-EMT simulations to address confidentiality concerns

This approach still has ample room for refinement; however, it is simple, powerful, and easy to
implement on various infrastructure styles (i.e. physical machines, across LANs or, on the cloud etc.).
In the Australian region, this approach is being tested to allow newly connecting generators to better
tune their EMT models and physical installations with a level of detail and fidelity that could otherwise
not be achieved without sharing full, confidential models. Experiences to date have been largely positive

98
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

with no methods to breach confidentiality identified. However, there remains room for improvement on
overall usability and workflow.
It is also noted that the engineering model management resources required to setup, monitor and adjust
the mating network model in this approach is significant, as the primary user cannot make network-level
changes independently. If a redispatch of machines, a network configuration change, or a new fault type
needs to be studied, the model management resource will need to manually do this on behalf of the
external user, as the external user only has access to modelling information up to their connection point.
4.2.1.2.1.2 Example 2 – Multiple connection point exposure
The previous EMT-EMT modelling approach can altered to reduce the managing organization’s
personnel requirement, by changing where the split occurs between computing instances (and hence
model visibility).

Figure 4.2-8 Modified EMT-EMT simulations to reduce model management resource requirements
As shown in Figure 4.2-8, rather than bundling the network model in with the confidential models in the
secure computing instance, the network model is placed on the non-confidential computing instance
side, allowing the external user to see, use and adjust the network information (this is subject to local
laws on what portion of power system models are confidential/protected information). The confidential
plant models remain on the secure computing instance as in the first example, however, rather than a
single network connection between computing instances, multiple (potentially dozens) of network
connections are required, one for each confidential plant model and its mating connection point in the
network model.
This simple but powerful change allows the external user to evaluate their own custom scenarios without
the need for a model management resource to make changes on their behalf, as is the case in the first
example. This approach is best applied when confidentiality laws allow it, and when the confidential
EMT plant models are able to receive run-time settable P/Q/V dispatch targets that do not require a
recompilation of the confidential EMT plant model (a difficult - but not impossible - problem to overcome
with some older or poorer quality EMT models).
4.2.1.2.1.3 OEM feedback
As part of the proof-of-concept AEMO ran for this EMT-EMT co-simulation tool, AEMO received
feedback from some OEMs that the second “multiple connection point exposure” approach may still
leave their models vulnerable to reverse engineering. With each confidential plant’s network connection-
point exposed to an external user, a rudimentary equipment profile and dynamic performance could be
reverse-engineered by a malicious user. This would be achieved through applying an exhaustive set of
disturbances and grid-side parameter sweeps in an attempt to characterize the equipment behind the
connection point. However, it remains unclear how much useful information could be gathered by this
approach and hence how likely intellectual property theft could be in this situation. This reverse-
engineering approach would likely be further obstructed if the plant has multiple plant active behind the
connection point (e.g. supporting STATCOMs, synch-cons, SVCs alongside primary IBR plant), as

99
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

individual plant responses would be difficult to disaggregate from the combined response at the network
connection point.
4.2.1.2.1.4 Licensing considerations
The approaches demonstrated above constitute a software as a service (SaaS) functionality. When
establishing such a system, the managing organization must consider that the provision of this service
may be eroding the ability of a software vendor to otherwise sell their tool to an external party. It is
therefore important to ensure that appropriate licensing discussions have occurred, contracts are in
place, and that the software vendors are appropriately compensated for their potential lost revenue
before any such SaaS is offered. This not only includes the EMT tool, but any dependency the tool may
have, such as compilers, or even the operating system itself.

4.3 Real-time EMT simulation

Real-time simulation can be considered as a particular case of simulation where the computation time
of the simulated system must be constrained within the simulation time-step. In other words, the process
of sampling the inputs, solving the system equations and updating the outputs must be performed in a
time-frame shorter than or equal to the wall-clock duration of the time-step. If the computation time
exceeds the time-step at any point of the simulation, inaccurate results or divergence of the simulation
may occur. This is illustrated in Figure 4.3-1.

Figure 4.3-1 Simplified process of RTS solution

In order to perform real-time simulation, real-time simulators (RTS), combining parallel multi-core
processors, real-time operating systems and specialized modelling and simulation software, are used.
With the continuous increase in processor performance, larger systems can be simulated with the same
time-step and number of cores. Furthermore, specialized software has made it easier to port simulation
models on field-programmable gate arrays (FPGAs), allowing the use of time-steps in the range of
hundreds of nanoseconds. RTS provides various tools and solvers to support most of the previously
mentioned simulation types, namely phasor, PDT, EMT or a combination of them.
Typically, for EMT type simulations of large-scale systems, comprising thousands of electrical nodes,
decoupling of the model is necessary in order to divide it into sub-tasks and parallelize the simulation
on multiple CPU cores to comply with the real-time constraint. On transmission systems, where long
transmission lines are employed, decoupling techniques like the use of Bergeron line model are
commonly used. However, for distribution systems or microgrids, where transmission lines propagation
delays are smaller than the simulation time-step, other decoupling techniques can be employed. The
same propagation delay principle can be applied to decouple at the location of power transformers, for
example, by replacing their leakage impedance with stub lines, resulting in a trade-off between
simulation performance and accuracy. Indeed, the additional capacitance required to obtain the desired
propagation delay at the transformer leakage impedance obviously impacts the simulation results. More
adequate options exist for large scale systems, like the use of State-Space Nodal (SSN) solver [R56]
[R61].
4.3.1.1 Prototyping
One of the main advantages of using RTS is the ability to perform rapid control prototyping (RCP) along
with Hardware-In-the-Loop (HIL), where a portion of the system is simulated and actual equipment is
interfaced to the RTS using standard industrial communication protocols, low-level signals, a power
amplifier or a combination of these. HIL test beds expand the test coverage and possibilities. For
example, it can be used to validate power electronics controllers and converters response to faults, test
FACTS and HVDC control interactions, study generator controllers, protection system coordination and,
more recently, resilience to cybersecurity contingencies [R62][R63].
Most commercial simulation tools are capable of automatically generating a controller source code from
the model, which greatly facilitates the process of implementation of the controller in RTS. This controller

100
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

prototype can be used for real-time model testing, verification, and integration. The combination of RCP
and HIL techniques in a project allows the following advantages [R64]:
▪ Design issues can be discovered earlier in the process, enabling required trade-offs to be
determined and applied, thereby reducing development costs.
▪ The development cycle duration is reduced, due to parallelization in the workflow.
▪ Testing cost can be reduced in the mid to long term since HIL test setups often cost less than
physical setups, and the RTS can be used for multiple applications and projects.
▪ Tests results are more repeatable because RTS dynamics do not change through time the way
physical systems do.
▪ It can replace risky or expensive tests that use physical tests benches.
It is also important to realize that the same control system prototype code can be used for software-in-
the-loop (SIL) for accelerated simulation of EMT models.

101
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

5. Acceptance testing and validation of EMT models

This section outlines several recommended practices when accepting and testing EMT models from
suppliers and building and validating wide-area EMT models with high penetration of IBRs. Examples
of both successful and unsuccessful wide-area validations are provided for the reader’s information.

5.1 Pre-commissioning model acceptance on individual models

Model acceptance tests are necessary to provide confidence that the model is usable and numerically
robust, and represents the installed plant under reasonably expected operating conditions 6. These tests
can be performed on an individual plant model basis in a SMIB environment.
The objectives of the acceptance tests described in this section are to determine the following:
▪ Robustness of the model for defined test conditions specified by upper and lower boundaries of
System Strength.
▪ Consistency and accuracy of EMT modelled performance from the manufacturer/Generator
provided validation, and consistency of EMT modelled performance reflective of equivalent system
strength conditions at the point of connection (PoC).
▪ If the model information:
 Is fit for the purpose in progressing with the power system connection studies.
 Is acceptable for application of models for operational, planning, and power system
assessment needs.
 Meets relevant modelling requirements.
 Has documentation and structure that meets relevant requirements, including for provision of
data, source information, settings, and control diagrams.
The following sections include a series of model acceptance tests which should be considered in pre-
commissioning stage, with each test described in further details.
5.1.1 Pre-requisite test – SMIB flat run
This test is to run EMT models in SMIB environment without disturbance for extended period and various
system strength conditions to identify:
▪ whether there is memory leak issue with the EMT model
▪ whether the EMT model can be initialised sufficiently quickly at various output levels
▪ whether the EMT model can be initialised at various system strength levels
The test should be run a few times, and the results from each flat run for the same test conditions should
be identical.
5.1.2 Balanced fault – large disturbance test
This test is to run EMT models in SMIB environment with balanced disturbance of various duration and
various system strength and X/R ratios to investigate:
▪ whether the EMT model demonstrate reasonable response to balanced disturbance
▪ whether the EMT model can reproduce the active and reactive current performance requirements
described in the Grid Code
▪ potential SCR withstand capability, beyond which the EMT model will not function properly
5.1.3 Unbalanced fault – large disturbance test
Similar to Section 5.1.2, this test is to be conducted with unbalanced faults, including:
▪ Single phase to ground faults
▪ Phase to phase faults
▪ Phase to phase to ground faults

6
A comprehensive set of pre-commissioning PDT and EMT model acceptance tests is published by the Australian Energy Market Operator, and is
available here: https://aemo.com.au/-/media/files/electricity/nem/network_connections/model-acceptance-test-guideline-feb-2021.pdf

102
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

5.1.4 Sequential fault – large disturbance test

Where requirements are applicable in relevant Grid Codes or Technical Standards in capability to ride
through multiple sequential faults, acceptable tests should be carried out to examine the model’s
capability to ride through sequential faults.
It should be noted that such tests should focus on whether the EMT model demonstrate reasonable
response to a series of faults, rather than whether such response would meet relevant performance
standards.
5.1.5 Temporary overvoltage test
This test is to run EMT models in SMIB environment with a capacitive shunt device switching during the
simulation, resulting in transient overvoltage at the connection point. This test should be carried out with
the EMT model running at full active power output, and maximum reactive power production and
absorption respectively. Various levels of system strength and X/R conditions should be examined as
well.
This test can identify whether the EMT model can demonstrate reasonable response, particularly
reactive power absorption capability, under transient over voltage condition.
5.1.6 Voltage reference step change test
This test is to change the voltage reference setpoint during EMT simulation, and observe the change of
reactive power output from the EMT models, and determine whether voltage controller is functional. This
test should be carried out at full active power output and low active power output of the EMT model.
Different system strength and X/R conditions should be examined.
In this test, both voltage reference step up and down should be investigated.
Similar tests can be carried out for power factor control, where power factor setpoint can be
changed instead of voltage.
This test can be carried out in SMIB environment.
5.1.7 Active power controller reference test
This test is to change the active power setpoint of the EMT model, and observe whether the active
power of the EMT model would follow the setpoint. Sufficient time should be allowed following the
setpoint change to consider reasonable ramping rate.
This test is particularly important when the real equipment is fitted with runback scheme.
This test can be carried out in SMIB environment.
5.1.8 Grid frequency – controller test
This test is to change the grid frequency of the infinite bus in SMIB case, where the EMT model is
being tested, and monitor whether the active power output of the EMT model will respond in a way that
mitigates such frequency change. A suitable rate of change of frequency should be applied in the test,
instead of frequency shift which cannot physically occur.
Both frequency increase and decrease should be tested. During the test of frequency increase, the EMT
model should be initialised at full active power. During the test of frequency decrease, the EMT model
should be initialised at minimum active power.
5.1.9 Inertia – frequency control test
Plants/models with inertia controllers would be tested on case-by-case basis, taking into consideration,
for example:
▪ Stored energy.
▪ Inertia period/ underfrequency.
▪ The speed of the response.
▪ Activation deadband.
▪ Plant settings (for example, droop).
▪ Recovery characteristics.

103
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

5.1.10 Grid voltage change – response test

This test is to change the voltage magnitude of the infinite bus of SMIB case where the EMT model is
being tested. This test complements the test described in Section 5.1.6 and investigates the
performance of the voltage controller in response to a grid voltage change.
No tap changer is considered in this test. The model is expected to maintain its active power, if the grid
voltage change is within normal operation range. The FRT response of the model may be examined if
the grid voltage change triggers FRT response of the model.
The test should be carried out for various system strength and X/R conditions, for both full active power
output and minimum active power output.
Voltage step changes and ramped change can be applied.
5.1.11 Oscillatory rejection test
This is a control and response sensitivity test. It is expected that plant models maintain stable operation
for all voltage modulated frequencies and for measured responses to be consistent with changes in
current injection references. The test is primarily focused on asynchronous generators by monitoring
active and reactive current references together with the resulting active and reactive current responses,
however, the test is applied to all plant models.
An example of modulated voltage is presented in Figure 5.1-1. The modulated oscillation frequency
ranges from 1 Hz to 10 Hz, with 1 Hz increase per step.

1 Hz increments

Figure 5.1-1 Oscillatory rejection tests on PDT grid voltage (1-10Hz, 1Hz per step)

5.1.12 Grid phase angle change – response test

Phase angle jumps can occur, one example being when a heavily loaded line trips, with multiple lines
going out of a substation. Plant must not lose control for such phase angle jumps. For example, in
some weak networks, plants may be required to tolerate a 30 to 40-degree phase angle change within
a cycle or two. Control robustness to higher phase angle changes (if any) may apply as a requirement
in consideration of connection point characteristics and contingencies in some cases. These aspects
may not be known during the model acceptance testing and may need to be revisited once information
from network contingencies becomes available. Manufacturers should try to provide evidence of the
biggest phase angle change that their equipment can sustain.
5.1.13 Extremely weak network tests
In extremely weak network connection scenarios (e.g., near SCR = 1.0) additional EMT model checks
should be undertaken to test the robustness of the model at operational extremes. In such scenarios, a
successful test may show plant instability, as a plant that is stable for extreme conditions outside its
design envelope may be indicative of an incorrect or “impossibly good” model.
The follow examples are recommended as a minimum:
▪ Increase active power reference in gradual steps until the plant reaches its rated output. It is
possible that the plant is unable to maintain stable operation at 100% output level. Active power
ramp durations may be extended to meet the equipment maximum slew rate limitation.
▪ Fault-clearance system strength changes. This test considers system strength change on
clearance of the fault where pre-disturbance SCR conditions are lowered to SCR = 1.0. It is possible
that the plant/model performance would not be able to sustain operation at SCR = 1.0. In cases

104
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

this is actually possible, evidence (other than modelled results) would be required to substantiate
model ride through capability at SCR=1.0.
The tests above should be repeated for an expected range of SCR and X/R at the point of connection
and for a range of different active and reactive power operating points.

5.2 Cross-platform and long-run checks on wide-area models

To ensure no major errors have been introduced when constructing a new wide-area EMT network, one
possibility is to compare its initial simulation results to an existing PDT-type network. This type of
verification is for establishing network coherency and validating network performance for
electromechanical transients where IBR energy sources are simplified. This assumes that the PDT-type
network data and performance was previously verified.
Coherent model performance can be verified using different signals at different network locations. The
responses across packages are not expected to return identical results given the simplification of IBR
plant models, but should provide general agreement, especially for slower transients resulting from
electromechanical interactions.
For example, a full (L-network) and simplified (R-network) EMT models are compared against a full PDT
model. The L and R networks are described in section 1.3.7. Figure 5.2-1 shows a sample of active
and reactive powers. Figure 5.2-2 presents voltage comparisons. It is noted that the phasors are
extracted from the EMT time-domain solution waveforms. It is apparent that EMT waveforms contain
higher frequencies, but the overall waveform shapes with both solution methods are in agreement.

Figure 5.2-1 Active and reactive power comparisons for dynamic behaviour

105
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.2-2 Bus voltage phasor module comparisons, dynamic performance

5.2.1 Long-run stability checks

As an EMT-type simulation includes a significant number of details for the representation of the power
system, its performance is not immune to various mathematical complications in the solution process.
It is also affected by various modeling details.
A simple technique used to assess numerical stability and accuracy is the verification of simulation
results over a very long simulation interval. A good indicator is the measurement of synchronous
machine speed. Its steadiness in the absence of perturbations, confirms the performance of control
systems, the modeling options and the lack of numerical drift.
In the example below, the wide area EMT models from Section 1.3.7 are simulated for 1000 s. The L
and R networks are the full and reduced networks, respectively. The simulation results for a given
generator are presented in Figure 5.2-3 using a numerical integration time-step of 100 µs. The tests are
performed with and without magnetization branch modeling in transformers. It can be observed that the
numerical noise in the speed signal is suppressed when taking into account transformer magnetization.
This observation confirms the need to account for the actual physical behaviour of components through
their models.

106
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.2-3 Long term simulation: synchronous generator speed signal with and without transformer
magnetisation branches

5.3 Examples of validation against staged field test results

A step in gaining full confidence in the model is the comparison of the models to any on-site tests. This
will ensure that the model is representative of the final delivered project. It also allows one to ensure
compliance to the specified requirements which are often at ac system conditions that are not
considered normal operation (i.e. N-1-1, weak systems). By having confidence in the models, one can
compare the “operational system” tests to the models by configuring the models to match this condition.
If there is good correlation, one can then simulate the specified conditions and accept based on this.
The following section describes three staged tests where the ISO, transmission owners and asset
owners carried out between 2019 and 2020. These tests are:
▪ November 2019 - Voltage oscillation in Northwest Victoria region with five solar farms contributing
to the oscillation
▪ March 2020 - Voltage oscillation in Southwest New South Wales caused by interaction between
solar farm and SVC
▪ April 2020 - Commissioning test in Northwest Victoria region with five solar farms retuned to
eliminate previously identified voltage oscillation.
5.3.1 November 2019 staged test – voltage oscillation in Northwest Victoria,
Australia
EMT simulation results indicated 8 Hz sub-synchronous voltage oscillations, caused by some newly-
connected solar farms in a very weak part of the power system, which resulted in post fault voltage
profile exceeding the maximum flicker level required by network asset owners. Such voltage oscillation
was not identified during the solar farm PDT grid connection studies. Historically EMT models and wide
area EMT studies were not part of the requirement for generator grid connection process, and only PDT
simulation tool was used when these solar farms were connected to the grid between 2015 and 2017,
and EMT models of these solar farms only became available to the ISO after the solar farms were

107
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

connected. Since 2018, EMT models and studies are mandatory requirement for all generator grid
connection process, particularly IBR.
EMT simulation results indicated the sub-synchronous oscillation is mainly triggered by a particular
transmission circuit outage in the region, and the results indicated that such voltage oscillation can be
triggered by switching off this particular transmission line, without the presence of a power system fault.
To confirm that these findings are true behaviour of the actual solar farms, the ISO, with the local
transmission asset owner, performed real time transmission line switching tests in November 2019,
which triggered the sub-synchronous voltage oscillations, similar to what was observed in the EMT
simulation.
Voltage measurements were obtained from this test, then the ISO simulated the exact test condition
using EMT simulation. Each component in the EMT model used in the simulation was previously
validated using the playback method. In this simulation, the exact power system operating condition
was replicated in the EMT simulation, followed by the same transmission line switching event. The ISO
also worked with the solar farm owners and OEM to obtain the actual inverter settings used exactly on
the field for these solar farms, and updated the existing solar farm model with these actual settings.
The overlay results are presented below. Figure 5.3-1 and Figure 5.3-2 show the comparison of real-
time measurements against simulation using the full-scale integrated approach. Both figures only show
the overlay at the connection point of a single solar farm, but the rest of the solar farms showed a similar
match between simulation and real-time measurements.

Figure 5.3-1 Voltage oscillation overlay between simulation and actual measurement (overview)

Figure 5.3-2 Voltage oscillation overlay between simulation and actual measurment (zoomed view)

108
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Both plots show congruent results, as the simulated voltage oscillation matches with the actual
measured voltage oscillation, not only at the magnitude, but also at the frequency of oscillation (6 ~
8 Hz). This oscillation mode was also observed in other solar farm connection point results as well,
further proving the validity of the EMT model, and previous findings.
The above validation results show that the EMT model can capture fast control system interactions,
which traditional PDT models fail to replicate. It also shows that the EMT model is capable of capturing,
not only power system behaviour at fundamental frequency, but also power system oscillations at sub-
synchronous level.
5.3.2 March 2020 test – Southwest New South Wales, Australia
The relevant network owner and system operator conducted another staged test to verify the existence
of another voltage oscillation in the same region, but caused by a different transmission line outage.
EMT simulation results indicated that with two SVCs in service together with a nearby solar farm, voltage
oscillations can be observed following the transmission line outage. However, with only one SVC in
service along with the solar farm, such voltage oscillations will not occur. This suggested that this voltage
oscillation is caused by interaction between these two SVCs and the solar farm. During the staged test,
one SVC is already switched off for maintenance, so the staged test was carried out with only one SVC
in service to firstly verify that no oscillations are present in this configuration.
Figure 5.3-3 and Figure 5.3-4 present an overlay between a measured voltage at a transmission bus
and the simulated voltage at the same location using EMT studies. The same line switching was
simulated in EMT studies, and both the simulated voltage and the measured voltage indeed showed no
post-contingency voltage oscillation.

Figure 5.3-3 Transmission bus voltage profile - March 2020 test

109
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.3-4 Transmission bus voltage profile - March 2020 test (zoomed view)

5.3.3 April 2020 test – Northwest Victoria Australia, with revised inverter
settings
After the voltage oscillation described in Section 5.3.1 was identified and confirmed by system tests,
AEMO worked with industry participants and the solar inverter OEM to retune the inverter settings for
solar farms contributing to the voltage oscillation in Northwest Victoria region. This tuning was
conducted using EMT simulation, and the outcome was quite promising, as the voltage oscillation was
dramatically reduced and the post contingency voltage profile can meet the flicker requirement specified
by the network asset owners.
Following the inverter tuning, all five solar farms which were originally contributing to the voltage
oscillation had their inverter settings updated according to the revised settings. Under the requirement
of the Australian National Electricity Rules, the solar farms must go through commissioning tests
following firmware updates. In April 2020, AEMO together with relevant network asset owners and
generator owners, facilitated these commissioning tests of the solar farms with the revised settings.
The settings tests were conducted in 4 days continuously in a week, where the solar farms were tested
with various numbers of inverters connected and MW output levels, from 25% of maximum output to full
output. Measured data were obtained from sites, and were subsequently used for model validation.
AEMO had setup EMT models using the exact same test condition 5 minutes prior to the actual test,
where the same transmission line was switched off which would have caused voltage oscillation at solar
farm connection points with the original solar inverter settings. The following plots presented the overlay
between EMT simulation and actual site measured data for voltage at one solar farm connection point.
Figure 5.3-5 and Figure 5.3-6 showed the voltage profile at one solar farm connection point during the
April 2020 test, with revised solar inverter settings. From the measured data, it can be seen that the
8 Hz voltage oscillation identified in previous tests are eliminated. The EMT simulation results are slightly
more pessimistic compared with the measured data, but still showed the post fault voltage profile at the
solar farm connection point has met the flicker requirements specified by the network asset owners.

110
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.3-5 Voltage profile at solar farm connection point

Figure 5.3-6 Voltage profile at solar farm connection point (zoomed view)

5.4 Validation of wide-area EMT models against actual system

disturbances
Extending the ideas of validation of EMT models against field data shown in Section 5.3, this section
provides examples of a wide-area EMT model being validated against actual system disturbances
(faults) using high speed measurement data.

111
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

5.4.1 Australian examples

5.4.1.1 Example indicating acceptable correlation
The following figures present examples of acceptable correlation between EMT simulation results and
measured system disturbances.

Figure 5.4-1 Generator 1 active power response comparison

Figure 5.4-2 Generator 1 reactive power response comparison

Figure 5.4-1 and Figure 5.4-2 show comparison of active power and reactive power response of a
synchronous generator between EMT simulation and actual measured response. In the EMT simulation,
measured three-phase voltages and frequency were fed back into the model and the response of the
model was observed and compared with the real-time measurements. These figures show that the plant
model performance is well within the accuracy range, set out in the AEMO’s Power System Model
Guidelines7, as highlighted by green band. This accuracy band is calculated as ±10% of the maximum
deviation from steady state level, for each measured quantity.

7
Available at: https://aemo.com.au/-/media/Files/Electricity/NEM/Security_and_Reliability/System-Security-Market-Frameworks-
Review/2018/Power_Systems_Model_Guidelines_PUBLISHED.pdf

112
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.4-3 Generator 2 active power response comparison

Figure 5.4-4 Generator 2 reactive power response comparison

Figure 5.4-3 and Figure 5.4-4 show comparison of active power and reactive power response of an IBR
generator between EMT simulation and actual measured response. A wide area EMT model was setup,
where the actual disturbance was simulated. The simulated active power and reactive power were
overlaid with the actual measured response. These figures show that the plant model performance is
well within the accuracy range, set out in the AEMO’s Power System Model Guidelines, as highlighted
by green band. This accuracy band is calculated as ±10% of the maximum deviation from steady state
level, for each measured quantity.
5.4.1.2 Example indicating unacceptable correlation
The following figures present examples of unacceptable correlation between EMT simulation results
and measured system disturbances. It should be noted the main reason for unsatisfactory results is due
to insufficient or incorrect modelling, rather than genuine deficiency in capability of EMT modelling tools.
Figure 5.4-5 and Figure 5.4-6 show comparison of active power and reactive power response of a
synchronous generator between EMT simulation and actual measured response. The comparison
shows that the simulated response is more oscillatory than the measured response. Further
investigation suggested such different performance could be related to different hardware
implementation on the equipment, than the generic information used to develop this model.

113
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.4-5 Generator 3 active power response comparison

Figure 5.4-6 Generator 3 reactive power response comparison

5.4.2 Gaspésie Peninsula Wind turbine model validation

The Hydro-Québec Type 3 WG model was validated with field measurements. This validation is
presented below. From 2007 to 2009, various events consisting essentially of voltage sags resulting
from remote faults on the power system were recorded. Among these, six were selected to validate the
Hydro-Quebec Type-3 WG model. The selection was based on the fault severity, the type of fault and
the WG pre-fault operating conditions. These events cover a wide range of actual operating conditions,
from low power at sub-synchronous speed to full power at super-synchronous speed.

Figure 5.4-7 Measurement points of the Type 3 Wind Turbine

The model validation process used is based on playback techniques, where the model is fed with
recorded voltages from the actual WG, and validation is confirmed when the model produces the same
currents as those recorded during the disturbance. Following the same approach of waveform playback,

114
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

the entire WPP model has also been validated, using recorded voltages and currents at the POI level.
The wind farm control system was also modeled and validated in the meantime.
Figure 5.4-8 shows a comparison of simulation results and field measurements for event 6, the most
significant event in terms of diversity and severity. It can be seen that the conformity of the model with
field measurements is very good for this event. Such good correspondence for the other recorded events
has helped increase confidence in the validity of the model. Fine-tuning of the model is a relatively
straightforward process for small disturbances, but it becomes more complex with large and/or
unbalanced disturbances due to various nonlinearities. Regardless of the disturbance severity, this
approach requires a good understanding of the internal dynamics and control strategies of the WG to
be modeled.
Based on the validation performed, the performance of the model seems satisfactory. But it was
validated using only six events and using only one WG. Furthermore, the model does not include any
protection systems since no information was available to model them. Therefore, to increase confidence
in the validity of the model, it was decided to continue the validation of the model by adding on-line
monitoring equipment’s at other WPPs.
Table 5-1 Recorded events used to validate type 3 wind turbine model
WG pre-fault Fault type (duration)
operating conditions
Positive and negative
sequences of stator voltage
(% of the positive sequence
pre-fault voltage)

Event1 Speed: super- Unbalanced fault (0.05 s):

synchronous
(2007) positive sequence: -4%
negative sequence: +3%
Event 2 Speed: sub- Unbalanced fault (0.2 s):
synchronous
(2007) positive sequence: -4%
Generated power: 360
negative sequence: +5%
kW
Event 3 Speed: super- Unbalanced fault (0.11 s):
synchronous
(2008) positive sequence: -28%
Generated power: Full
negative sequence: +21%
power
Event 4 Speed: sub- Unbalanced fault (0.15 s):
synchronous
(2009) positive sequence: -9%
Generated power: 230
negative sequence: +9%
kW
Event 5 Speed: approx. Unbalanced fault (0.2 s):
synchronous
(2009) positive sequence: -13%
Generated power: 440
negative sequence: +13%
kW
Event 6 Speed: super- Unbalanced evolving fault:
synchronous
(2009) For the first 0.42 s:
Generated power: 1.3
positive sequence: -17%
MW
negative sequence: +14%
For the next 0.8 s:
positive sequence: -28%
negative sequence: +4%

115
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Stator voltages, Blue: positive-sequence, Green: negative-sequence

500

400

Voltage (V)
300

200

100

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)

Stator currents
3000

2000

1000
Current (A)

-1000

-2000

-3000
1 1.02 1.04 1.06 1.08 1.1
Time (s)

Rotor currents

1000

500
Current (A)

-500

-1000

0.2 0.25 0.3 0.35 0.4 0.45 0.5

Time (s)

Figure 5.4-8 Recorded positive- and negative-sequence voltages at stator level; comparisons of currents
for different time frames, for recorded and simulated waveforms at the Hydro-Québec Type 3 WG level for
event 6. The simulation results and measurements are overlayed.

5.5 Hardware-in-loop testing

5.5.1 Software-based validation
TasNetworks’ RTS has been operational since mid-2019. Before its commissioning, the models of the
Tasmanian power system were primarily validated against TasNetworks’ existing power system models;
both PDT and EMT, with only limited opportunity for modelling against actual network events. Since
April 2020, it has become feasible to validate more of the model, especially the generic power electronic
based models, against actual network events following the commissioning of new windfarms and the
addition of new phasor measurement units near to existing windfarms.
Therefore, this section records the performance of the RTS model of the full Tasmanian power system
(as dispatched) against several network events from year 2020. The power system dispatch is
downloaded from TasNetworks’ SCADA measurements to build a PDT load flow case which is then
converted to an equivalent steady state case for the RTS.
5.5.1.1 Event of 07 January 2020 – HVDC Interconnector Validation
The RTS model of an HVDC Interconnector was validated against a detailed offline EMT model, which
itself has been shown to give a faithful representation of actual system events. However, the offline
EMT model is an extremely detailed, proprietary model and could not be copied into RTS EMT.
Therefore, the RTS model is based on a generic PDT model with a phase-locked loop added for the
EMT environment of the RTS.

116
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Since the HVDC Interconnector uses line-commutated converter technology most disturbances at its
inverter terminals will cause a commutation failure and these common events provide a good opportunity
to validate model performance. The event of 07.01.2020 was a distribution fault near to the converter
station, which while deep enough to trigger a commutation failure, did not cause other major impacts
across the power system. The grid voltages during the disturbance are shown in Figure 5.5-1 and while
the initial disturbance is not great the subsequent commutation failure causes a deeper voltage dip; this
aligns with the high pulses of current drawn by the converter as shown in Figure 5.5-2.
It is not possible to perfectly recreate the converter currents during commutation failure with the RTS
model since its HVDC transformer model is fixed and different to the actual transformer winding
arrangement. Nevertheless, a reasonably close comparison during the fault is achieved. Importantly,
the active and reactive power recovery profiles are very close, refer to Figure 5.5-3 and Figure 5.5-4. It
was also noticed that just prior to the fault the actual voltage and current waveforms were experiencing
noticeable unbalance which was not present post fault but it is unknown whether the change occurred
due to the fault clearance. It must be emphasised that actual fault events can be multifaceted.

Figure 5.5-1 Overlay of actual and RTS simulated L-G voltages at 220 kV busbar during the fault and its
recovery period

Figure 5.5-2 Overlay of actual and RTS simulated HVDC AC currents during the fault and its recovery
period

Figure 5.5-3 Overlay of actual and RTS simulated HVDC active power during the fault and its recovery
period

117
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.5-4 Overlay of actual and RTS simulated HVDC reactive power during the fault and its recovery
period

5.5.1.2 Event of 27 August 2020

This event was a line-line fault on a 110 kV transmission line but due to the very high penetration of
asynchronous generation at this time, the impact of the event was felt across the power system. This
event is interesting to compare because the windfarms and the HVDC Link (as inverter) were generating
high output. Figure 5.5-5 illustrates the challenge in trying to accurately replicate an actual network
event in simulation since power systems never truly reside in a “steady-state”. This is clearly magnified
when “light” power systems, such as Tasmania, are exposed to the stochastic power flows that can
emanate from wind and HVDC as occurred at the time of this event. The event to be analysed occurred
at 19:02 and is seen as the spikes on the traces of HVDC Power and AC Frequency. Just prior to the
event, the AC frequency was below nominal but on an increasing trajectory, whereas the RTS case is
built from a solved PDT case with a stable 50.0 Hz steady-state condition.

Figure 5.5-5 Variation of HVDC Power and AC Frequency within ± half hour of the fault
Figure 5.5-6 highlights that the RTS closely replicated the large phase shifts that occurred at a remote
110 kV busbar during the L-L fault. It is often these types of faults that pose the greatest challenges to
the protection and control of power electronic systems and is the reason why the performance of such
systems should be tested in real time if their impact on the power grid can be critical.
The overlays of HVDC power flow and AC frequency are shown in Figure 5.5-7 and Figure 5.5-8
indicating a very good correlation. The HVDC recovery trajectories are very close, while the post
recovery divergence can be explained by the non-steady state nature of conditions prior to the actual
event: due to the depressed initial frequency in the actual system, the synchronous generation fleet
would already be moving to raise frequency prior to the event, thereby reducing the requirement from
the HVDC system’s frequency controller post event. To allow a better AC frequency comparison a
0.07 Hz offset has been subtracted from the displayed RTS frequency. Both HVDC power traces clearly
show the breakpoint in the response characteristic of HVDC System’s Frequency Control; once the AC
frequency returns within the Normal Operating Frequency Excursion Band (50.0 Hz. ± 0.25 Hz.) the
controller’s proportional gain is greatly reduced thereby slowing the response.

118
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.5-6 Overlay of actual and RTS simulated L-G voltages at a remote 110 kV busbar during the fault

Figure 5.5-7 Overlay of actual and RTS simulated HVDC power flows during and for 15s after the fault

Figure 5.5-8 Overlay of actual and RTS simulated AC Frequency during and for 15s after the fault
Refer to Figure 5.5-9 for a comparison of the actual and simulated active power flows from a Type 3
windfarm. Even though this windfarm was very remote from the fault location its fault ride through (FRT)
logic triggered for this event. The correlation is very good with the model accurately capturing the
recovery profile.

Figure 5.5-9 Overlay of actual and RTS simulated Type 3 windfarm power flows during and for 15s after
the fault
Refer to Figure 5.5-10 for a comparison of the actual and simulated active power flows from a Type 4
windfarm. This windfarm was also quite remote from the fault location but its FRT logic has a more
moderate response and the windfarm maintains its “sent” active power during the fault. This continuity
is also captured by the RTS model but with minor glitches in current flow at fault inception and clearance.
These glitches are more visible in the sinusoidal traces of Figure 5.5-11 and Figure 5.5-12 but do not
materially affect the overall windfarm response.

119
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.5-10 Overlay of actual and RTS simulated Type 4 windfarm power flows during and for 15s after
the fault

Figure 5.5-11 Overlay of actual and RTS simulated voltages at Type 4 windfarm during the fault

Figure 5.5-12 Overlay of actual and RTS simulated Type 4 windfarm currents during the fault

5.5.1.3 Event of 30 August 2020 – Legacy wind farm model validation

This event started with a solid single-phase fault at a remote 110 kV substation terminal although its
impact was felt across the power system. It has been selected to validate the performance of a legacy
windfarm, which has various turbine types, since the windfarm was operating at high output at this time.
Due to a difference in recovery voltage between simulations and actual measurements over a period of
seconds, this case was also repeated in PDT tool. Refer to Figure 5.5-13 where the PDT model exhibits
a 5 % divergence in voltage between 1 s and 2 s after fault clearance.

Figure 5.5-13 Overlay of actual, RTS and PDT simulated voltages at 110 kV PCC during and for 10 s after
the fault
Refer to Figure 5.5-14, which shows that although the active power flows are very close, there is a
difference in reactive power flow which corresponds with the 5 % voltage dip during recovery. The

120
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

control systems for the legacy windfarm are not fully documented and the reason for the difference is
unclear. However, additional tuning of the RTS model by resetting and pausing the reactive power
recovery has allowed the RTS to more closely match actual performance.

Figure 5.5-14 Overlay of actual, RTS and PDT Active & Reactive powers at 110 kV PCC during and for 10
s after the fault

5.5.2 Hybrid use of software and control replica

In this section validation of the INELFE link using real-time simulation is illustrated. Comparisons
between real-time simulation including replicas and on-site measurements are presented [R65].
The INELFE link is an interconnection between France and Spain, power rated 2 GW and is composed
of two symmetrical monopolar configurations with a rated power of 1,000 MW/each. This HVDC is based
on the Modular Multilevel Converter (MMC) and, up to now, is the largest VSC based HVDC project in
the world.
For the considered system validation, the equivalent AC network, illustrated in Figure 5.5-15, was
modelled in a real time simulator. This AC network model has been considered sufficient to validate the
considered transient dynamics. More details on the hardware-in-the-loop setup and replicas
commissioning can be found in [R66].
The RTE AC side is composed of a Thévenin source representing the 400 kV AC station of Gaudière
that is linked via an AC overhead lines to the 400 kV AC station of Baixas. The Gaudière substation
represent the major SCL (Short Circuit Level) for the HVDC station. Two other Thévenin sources
representing the 225 kV and 63 kV are also linked to the 400 kV Baixas station via power transformers.
Near the INELFE link, an AC interconnection connects Baixas to the Spanish 400 kV AC station Vic.
The Spanish grid is based on the design given by ENTSOE at 02/13/2019 and REE information provided
to RTE. We can notice that Vic is linked to the AC side of the station Santa Llogaia trough the 400 kV
AC station Bescano.

121
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.5-15 a) Network representation of the INELFE link, b) HIL of INELFE Replicas with Real Time
simulator
Active power steps change of 20% were performed onsite to validate the correct performances of the
HVDC system. During this commissioning test, the ramps associated to the active power set points have
been disabled to be able to do real active power steps. This test is interesting to validate the HVDC
system control and the AC grid small-signal dynamics model.

On-site
Real-time

a) b)
Figure 5.5-16 Active power step – Active power (a), DC current (b) [R65]
The active power flow in the HVDC link is presented in Figure 5.5-16. The dynamics that follow the step
up and step down set-point are quite similar between measurement and simulation with replicas;
frequencies and amplitudes of oscillations are very close. They are mainly driven by the station control
system and dynamic behaviour of the grid surrounding the HVDC link. This shows that small perturbation
studies carried out on the replicas will reflect the good behaviour of the actual installation.
A second test validation is a block/deblock sequence. This onsite test is used to test the converter
performances to recover from a possible AC fault or miss-operation in the control cubicles. In some
specific cases, the converter might block during AC faults and it shall deblock during fault recovery. The
active power flow shall be restored as fast as possible. Comparisons with on-site measurements are
presented in Figure 5.5-17 and in Figure 5.5-18 (zoom).

122
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 5.5-17 Converter block/deblock – Comparison with on-site measurements [R65]

Figure 5.5-18 Converter deblock – Comparison with on-site measurements (zoom)

The differences observed in the voltage and power reactive power are related to the presence of the
second HVDC link during the field test. Indeed, during this on-site test, both links were in operation, the
active and reactive power loss due the converter blocking of one link is compensated by the second
link. This provides a voltage support from the second link. This behaviour related to the presence of the
second link is not reproduced on the replicas because only one link was represented for this validation
purposes.
Except this difference, the general behaviour and maximum peak values of all variables are in very good
concordance between field measurements and replica results.

123
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

6. Simulation case studies

6.1 System strength studies
6.1.1 Impact of a proposed inverter connected generator on system strength
The performance of an IBR plant is highly dependent on the network that it is connected to. If connected
to a strong system, it is expected that an IBR would have less stability and control interaction issues
than if it is connected to a weak system. A system is considered stronger if a higher number of
synchronous machines and less IBR (grid following) are connected in the vicinity whereas a system with
less synchronous machines and a high penetration of IBR (grid following) would be considered as a
weaker system.
To represent the accurate dynamic performance of an IBR, a plant specific EMT model of the IBR is
required. The EMT model must have both inverter level and plant level controls modelled. Any filter or
capacitor bank used for harmonics mitigation or for reactive power capability requirements must also be
included in the model.
When a new IBR is connected to the network, it effectively reduces the system strength for the nearby
existing IBRs. Therefore, addition of a new IBR can have impact on the performance and stability of the
existing IBRs in the network. System strength studies are performed to gain insight into the dynamic
performance and stability of the new IBR and its impact on the existing plants.
To properly evaluate system strength conditions, the system must be modelled in EMT domain. The
attributes of the system model should be such that
▪ System conditions are set up with the minimum allowable dispatch of the synchronous generators
and synchronous condensers.
▪ All the IBRs should be considered at maximum output.
▪ All the dynamic reactive power plants should be part of the system model.
▪ Any existing intertrip or special protection scheme should be included in the model.
▪ Loads should be modelled with the best available load models.
▪ Different loading condition in the system (summer high, low, winter high and low) should be
considered.
In this example, system strength studies for an IBR connecting into Queensland (Australia) transmission
network is presented. Figure 6.1-1 shows the Queensland transmission system and an IBR with yellow
circle is a new proposed connection to the network. To perform the system strength studies for this
plant, entire Queensland transmission network was modelled in EMT with all the dynamic plants
(synchronous and asynchronous) represented with the manufacturer specific models.
Before integrating the EMT model of the plant under study to the system model, following conditions
should be met.
▪ Model of the plant under study is tested as per section 5.1.
▪ Performance of the system model is acceptable without the new IBR (under study)

124
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.1-1 Map of the system under consideration for the system strength studies for a new IBR in
Queensland, Australia
The performance of the system was tested with the simulation of the credible contingencies (two-
phase fault) at different locations in the network. All the plants connected to the network showed
satisfactory recovery from the fault and system returned to a satisfactory state after the fault.
Figure 6.1-2 shows the system voltage response at two different key locations before the IBR was
considered in the network.

125
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.1-2 System performance before adding the new IBR – Voltage at a key node in the network
Once system performance was tested, the IBR under study was connected to the system model. Due
to a new plant connection, system will result in a different load flow solution compared to the conditions
prior to adding the new plant. It is important to make sure that system model solves for the new load
flow after connecting the additional IBR and all the dynamic plants are within their operating range.
Updated system model with the new IBR was tested for the credible contingencies (two-phase faults) to
understand the dynamic performance and stability of the new IBR and the system. Depending on the
grid code requirements, different fault conditions e.g. CB fail, long shallow faults or three-phase faults
should also be simulated to understand the dynamic performance of the plant.

Figure 6.1-3 System performance after adding the new IBR – Voltage at a key node in the network

126
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.1-4 System performance after adding the new IBR – Voltage at a key node in the network

Figure 6.1-5 Output of the new IBR

Figure 6.1-3 and Figure 6.1-4 show the system voltage response at two different key locations after the
IBR was considered in the network. Figure 6.1-5 shows the output of the plant. These figures show that
addition of the new IBR adversely impact the system performance for the following reasons.
▪ New plant causes undamped 8-10 Hz voltage oscillations in the network.
▪ Exceeds the Power Quality (flicker) planning levels.
6.1.2 Remediation for System Strength Adverse Impact
Low system strength could cause a variety of issues in a power system. Malfunction of IBR control
systems, control interactions among multiple IBR and negative impact on the stability of synchronous
plants are the most common issues due to low system strength. Experience from the Queensland
transmission system suggests that solutions to system strength issues cannot be generalised and have
to be evaluated on a case-by-case basis. In some cases, there can be multiple solutions to solve the
same issues. However, detailed EMT studies have been proven to be valuable in finding a cost-effective
solutions.
6.1.2.1 Synchronous condensers
There could be multiple solutions to the adverse impact identified in this example. A synchronous
condenser has been assessed as a solution in this case. Impact of a synchronous condenser connected
at the PoC of the new IBR is shown in Figure 6.1-6 to Figure 6.1-8.

127
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.1-6 System performance after adding a Synchronous Condenser with the new IBR – Voltage at a
key node in the network

Figure 6.1-7 System performance after adding a Synchronous Condenser with the new IBR – Voltage at a
key node in the network

128
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.1-8 Output of the new IBR after adding a Synchronous Condenser

6.1.2.2 IBR Control Modification

Modification in IBR control [R67] was also studies as a solution to the issues presented in 6.1.1. Impact
of modified control in the new IBR is shown in Figure 6-9 and Figure 6-10.

Figure 6.1-9 System performance with modified control of the IBR - Voltage at a key node in the network

129
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.1-10 Output of the new IBR with modified control

6.1.2.3 Control System Retuning

In a separate case, tuning of the IBR control parameters without changing the control structure was
found to be most cost-effective solution to the control interactions due to the low system strength. Impact
of change in control parameters is shown in Figure 6.1-11 to Figure 6.1-13.

Figure 6.1-11 System performance with old and new control parameters of the IBR – Voltage at a key node
in the network

130
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.1-12 Output of the IBR with old control parameters

Figure 6.1-13 Output of the IBR with new control parameters

6.1.3 Quantity and combination of synchronous generators/condensers

The Australian Energy Market Operator (AEMO) conducted EMT studies to determine a minimum
number of required synchronous generators/condensers for each region in National Electricity Market
power systems. An EMT simulation platform was used to properly represent power system dynamics
with fast-acting control systems used in wind turbines and solar inverters, which cannot be accurately
accounted for in PDT type simulation.
The next important step is to determine the most critical power system contingency which is likely to set
the security boundary. Given EMT studies are time consuming, it is not practical to study every single
possible contingency, and AEMO’s studies focus on the most critical contingency which causes the most
widespread impact for the whole power system. Previous knowledge on power system limitation or
weakness will help to determine such contingencies. The following types of contingencies were
analysed by AEMO (two-phase faults were applied):
▪ Loss of the largest synchronous generator in the power system

131
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ Loss of a critical transmission line, which will significantly reduce the level of system strength for
areas with high penetration of IBR
It is important to assess the contingency with system intact condition and N-1 prior outage as well. It is
also important to note the most critical contingency might be different for different operation conditions,
such as different synchronous generator dispatch.
The EMT studies started by investigating typical synchronous generator dispatch patterns which are
results of natural dispatch cycles. These synchronous generator dispatch patterns consist of
synchronous generators which are dispatched most often – mostly coal-fired synchronous generators
in Australia’s case. Then the number of dispatched synchronous generators are reduced to a point
where the simulation will show that the power system cannot reliably ride through the most critical
contingency and/or results in unstable system operation post contingency. After this, the number of
dispatched synchronous generators is increased one by one, until the EMT simulation shows the power
system can maintain stable operation post contingency. Such a synchronous generator dispatch pattern
will set the reference point of base system strength level which must be present in the power system to
ensure system security.
The next step is to determine whether there can be alternative synchronous generator dispatch patterns
to provide the same level of system strength to the power system. This is achieved by substituting a
large synchronous generator which is usually dispatched, with a few smaller synchronous generators
which are dispatch less frequently. The same contingency is studied, and a synchronous generator
dispatch combination is considered acceptable if the simulation demonstrates power system stable
operation post contingency with such a synchronous generator dispatch.
The treatment for synchronous condensers is the same for synchronous generators, with the only
difference that synchronous condensers cannot provide frequency control capability.
6.1.4 Developing operational constraints
Secure power system operation requires sufficient system strength in the network. System strength
available to the power system reduces with fewer online synchronous machines, or with increased
electrical distance between synchronous machines and IBR. Transmission network asset owners
disconnect transmission lines for routine or emergency maintenance or upgrade of insulators or
communication infrastructure. The timing and duration of an outage would vary depending on the work
carried out. During a network outage, the system strength support available for IBR further declines.
This could result in undesired behaviour of IBR.
In many instances there are large distances between the areas of IBR concentration and where most
large capable synchronous generators are located. Therefore, when there is an outage of any
transmission line between synchronous plants and IBR, system wide EMT studies are carried out to
determine the secure operating envelop of the power system. Often it is found that the system that is
capable of hosting full output of the IBR under normal conditions cannot host the full output of all the
IBR during the outage of a transmission line. Therefore, constraints on the MW output of the impacted
IBR are imposed while transmission line is out of service. In some parts of the network, limiting only MW
of the IBR without limiting number of inverters/turbines online has been enough to find a secure
operating envelop of the network. However, in other parts it was necessary to turn the inverters/turbines
off to find a secure operating envelop of the network.
Similar to the system intact conditions, the allowable output of an IBR is dependent on the number of
synchronous machines online. However, a prior outage of a critical transmission line might mean that
a higher number of synchronous generators are required to allow IBR operation at full capacity. In
circumstances when this cannot be achieved by natural dispatch of synchronous generators, applying
a constraint on the IBR output might be necessary. System wide EMT studies are conducted with the
best forecast of the availability of synchronous generators. Often multiple sets of generators are studied
to cover the unexpected system conditions. As a result of these studies, constraints that contain
percentage reduction in IBR MW capacity or number of inverters during outage are developed to inform
the system operator so that system can be set to a secure state before taking the outage of a
transmission line. These constraints are also used for unplanned outage to bring system into secure
state within the defined time as per the grid code requirements. In case of Australian National Electricity
Market, this time is 30 minutes.
In Queensland network, there are some SVCs that positively contribute to minimise the control
interactions. Constraints are also developed to manage the network in case of the planned outage of
those plants.

132
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

6.2 Adverse control interactions and harmonic instabilities

6.2.1 Sub-synchronous oscillations
Northwest Victoria and Southwest New South Wales have been attracting attentions of renewable
projects developers, with abundant solar resources in this region. Several solar farms have been
connected to this region since 2016, and a few more already planned or already under construction.
AEMO conducted large-scale EMT studies to assess the performance of newly connected solar farms
in this region, and observed sustained voltage oscillation at the connection points of five solar farms in
this region.
Further investigations were conducted to understand the major contributing factors to such sustained
voltage oscillation, and whether such oscillation can be observed under various operating conditions.
The following factors were considered and varied in EMT simulation:
▪ Voltage control setpoint of nearby inverter-connected devices, such as HVDC link and SVC
▪ Reduced active power output levels, but with all inverters connected, from the five solar farms,
where voltage oscillation are observed
▪ Reduced active power output levels, with reduced inverters connected, from the five solar farms
▪ Different solar farm dispatch patterns, with one or two solar farms switched off in some scenarios
▪ Various transmission line contingencies in the region
Depending on the scenario studied, the oscillations could be of constant or varying magnitude.
Irrespective of their shape, the magnitude and duration of the oscillations exceed allowable stability
limits under National Electricity Rules, power system stability guidelines, and network standards for
quality of supply. Figure 6.2-1 demonstrates an example of such sustained voltage oscillation observed
in Northwest Victoria region from high speed monitoring in the field.

Figure 6.2-1 Observed sustained voltage in Northwest Victoria, Australia from field monitors
EMT study results indicated that reducing solar farm active power output, without reducing the number
of inverters connected to the power system, will not reduce the level of voltage oscillation observed in
this area. Every solar farm is contributing to the sustained voltage oscillation, as reduced voltage
oscillation is observed from the studies only when any solar farm is disconnected from the simulation.
Study results also indicated such voltage oscillation is observed following one specific contingency.
AEMO and related transmission asset owners performed a series of power system switching tests in
Northwest Victoria region. Real time measurement obtained from these tests confirmed the existence
of such voltage oscillation, and Figure 6.2-2 represents a voltage measurement at the connection point
of one solar farm in this region.

133
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.2-2 Voltage oscillation measurement obtained from tests in Northwest Victoria, Australia
AEMO used the measured data from these tests to validate the EMT models for power systems in this
region. An EMT model was setup which is then initialised to the exact load flow condition as in the tests.
The same transmission line switching as in the test condition was performed in EMT simulation, and the
simulated voltage at solar farm connection points were overlaid with the actual measured voltage at the
same location. The results of this model validation can be viewed in Section 5.3.
It was noticed in simulation that by constraining the IBR, such voltage oscillation can be eliminated. But
the simulation showed such constraint should be applied to the total number of inverters connected
online, rather than the total MW output. Figure 6.2-3 shows that when only the total MW output of IBRs
are constrained, severe voltage oscillations can still be observed at the solar farm’s connection. But if
the constraint is applied to the total number of connected inverters, the voltage oscillation is greatly
reduced and is negligible.

Figure 6.2-3 Comparison between constraining MW output and total number of connected inverters

6.3 Designing system-wide control and protection schemes

The recent fast-paced integration of renewable energy has pushed the development of high
concentration of production and transmission assets based on IBR also in the ELIA network that has in
general sufficiently strong short circuit level to assure stability of the power electronics devices in normal
operation and incidents.
Some specific operational situations and very rare incidents (i.e. sudden N-2 or N-1 while one of the
other elements is in maintenance) may bring some part of the network from a very strong short circuit

134
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

level (i.e. SCR>10) to a very weak situation (i.e. SCR<0.6), above the maximum transmissible as shown
in the figure below.

Figure 6.3-1 Synthetic sketch of the situation analysed

Being a high impact low probability event, the effective risk and consequences have to be analyzed and
solutions proposed to guarantee the containment of the incident.
While for classical synchronous units, the consequences of such incidents and the type of instability is
relatively easy to foresee using a sufficiently detailed PDT model, for power-electronic based devices
the estimation of the consequences is very uncertain as it mostly depends upon the characteristics and
tuning of the control and protection (C&P) system that can be only represented with a highly detailed
EMT model.
The risk of instability has been detected several years before the commissioning of the assets and the
construction of the renewable units. The stability risk was detected via load flow simulations and short
circuit ratio assessment.
At that time, the detailed EMT blackbox models of the manufacturers of the power electronic parks were
not available to assess the consequences as the technology or manufacturer were not yet known. The
study approach used is shown in Figure 6.3-2 below.

Detailed design
Long term and validation Full validation
planning study of the solution via EMT study
on (SPS)
•PDT study triggered •Using Generic EMT •Full scale study
by steady state KPI models •Vendor models
•Power electronic •HIL simulation on a •Large scale model
PDT models specific reduced model
behavior validated
against EMT
simulations

Figure 6.3-2 Study approach for planning

A long-term planning study has been performed to define the solutions to the instability.

135
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

This study has been performed in a PDT simulation environment by tuning generic power electronic
models against EMT simulations of typical and generic assets, validating the behavior on specific
incidents and equivalent networks.
The PDT study has shown instability of the proposed solution. The zone in analysis has a very high
concentration of IBR. At the moment of these studies, neither the exact size and topology of the IBR,
nor the type and models of the IBR were known. For this reason, the model used could not completely
represent reality, more specifically in terms of how fast the instability was arising and the secondary
consequences (e.g. possible overvoltages or further consequences of the instability towards the
remaining of the system).
Further EMT simulations performed using the vendor models after commissioning of the different
injectors have confirmed the instability.
The study proposed a solution to reduce the risk of instability as a combination of a Special Protection
Scheme (SPS) to cover sudden incidents and a series of operational guidelines in case of foreseen
maintenances or detected higher risk of incident.
The SPS has been designed and validated in laboratory via HIL simulation using the best EMT models
available, being generic ones.
As soon as the detailed vendor specific models of the power electronic injectors were available, a
detailed EMT study has been performed to validate the solutions by extensive simulation to confirm the
results obtained during the long-term planning study.

Figure 6.3-3 High Level Comparison between assessment of stability of classical units vs IBR (PE)
based in the case of analysis
The availability of detailed EMT models at an early stage of development of new assets based on power
electronics is of fundamental importance to allow a correct assessment of their behavior when subject
to extreme network conditions and incidents. For the latter generic models or PDT simulation may not
represent their behavior with the needed level of conservatism necessary at this stage of development,
mostly as generic models do not usually represent with sufficient level of detail several protection
functions and detailed control functions (e.g. PLL) that are embedded in a detailed EMT vendor model.
This is very impacting in the studies performed due to the very high concentration of IBR in a zone
electrically far from synchronous generators or loads. In particular, models shall be usable during all the
project development, also when HIL/SIL simulations are needed.
For future projects, a generic vendor EMT model will be required at an early stage of project
development (for example during tendering) o to allow sufficient time for performing the full validation
EMT study.

136
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

6.4 Low Frequency Inter-area oscillations studies

6.4.1 Introduction
On 1 Dec 2016, the Iberian Peninsula and Turkey oscillated against the central Europe such as
Germany and Poland.

Figure 6.4-1 Measured frequencies in different locations of central Europe during the 1st Dec 2016 event
[R68]

INELFE HVDC link

Figure 6.4-2 AC and DC interconnections between France and Spain Transmission grids and mode shape
To analyse this event and to find mitigation methods, RTE and REE have conducted several studies
using both PDT and real-time simulations. PDT simulations have been conducted with the entire
European network and with a simplified HVDC link. In the following section, the real-time simulation
studies are reported to highlight the utility of such tool for inter-area oscillation event studies.
6.4.2 Event Study in Real-Time Simulations Using Replica
In order to reproduce this event in Real-Time simulations, a test bench depicted in Figure 6.4-3 was
used. It consists of two system equivalents linked through the two INELFE HVDC links with his AC
emulation control, and 3 parallel AC lines. Further details are reported in [R69].
The AC emulation control of the HVDC link, emulates the behaviour of an AC line increasing/decreasing
active power flow as a function of the angle differences [R70]. The AC emulation function is:
tot setpoint
𝑃ref = 𝑃ref + 𝐾 HMI (𝛿station1 − 𝛿station2 ) Equation 11
tot setpoint
where Pref is the total active power reference, Pref refers to the active power reference constant
setpoint, K HMI is the static gain set by the operator (MW/degree) and ( station1 −  station 2 ) is the
differential angle between the two stations. For the considered test, the K HMI is set to 360 and
setpoint
Pref =0

137
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

This real-time setup is not intended to model a detailed and realistic European network but to reproduce
the phenomenon; that is the inter-area oscillation around 0.155 Hz, damping ratio and the sequence of
event. Therefore, in this specific case, the Replicas were used in order to implement and test C&P
software modifications, whose purpose is the damping of possible inter-area oscillations, before
deployment on-site.

Figure 6.4-3 Simplified Model used in Real-Time Studies with Replica [R69]
Here is the timeline of the simulation:
▪ For t < 5 s, all AC circuit breakers are closed except for the BRK3. This allows for a 2 GW
exchange between PCC1 and PCC2.
▪ At t = 5 s, opening of the circuit breaker BRK1 (tripping of the AC line). This results in an
oscillation around 0.155 Hz between PCC1 and PCC2.
▪ At t = 55 s, opening of BRK2 and closing of BRK3 to simulate a load carryover, hence greatly
damping the inter-area oscillations.
Different simulations were played and are presented in Figure 6.4-4. Relevant situations were studied,
taking into consideration faults, outages and exchange scenarios, to verify the impact and test new
solutions.

138
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.4-4 Results of 3 simulations regarding the time constant of the P-mode-3 filter [R69]
Figure 6.4-4 shows the delta angle difference between PCC1 and PCC2, the active power transmitted
by the HVDC interconnection (MW), and the total power exchanged (MW) between PCC1 and PCC2.
Three different simulations were run:
▪ Simulation 1 (green graphs) corresponds to the case where the AC emulation is deactivated
and the HVDC interconnection is operated in fixed active power mode. The HVDC
interconnection is not impacted by the event (loss of a parallel AC line) and does not excite the
inter-area oscillation mode, which is positively damped.
▪ Simulation 2 (blue graphs) corresponds to the case where the AC emulation is activated with its
initial setting (750ms time constant). The trip of the parallel AC lines generates un-damped
oscillations with a frequency of f = 0.155 Hz. This simulation corresponds to the event occurred
on 1st December 2016.
▪ Simulation 3 (red graphs) corresponds to the case where the AC emulation is modified in the
C&P systems (time constant of 50s instead of initial value). Similar to simulation 1, the
oscillations appearing after the parallel AC line trip are positively damped, but this time the AC
emulation slowly adjust the transmitted power through the HVDC interconnection. Such results
have been cross-checked with PDT simulations in order to validate both real-time and offline
studies.
Comparison between PDT model and Real-time simulation are depicted in Table 6-1. The PDT model
includes the full ENTSOE grid model where the real-time simulation includes the representative test
bench. In Table 6-1 two different AC emulation control tunings are compared.

139
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Table 6-1 Comparison between PDT and Real-time simulation results

PDT simulations with the entire ENSTOE network Real-time simulation using replicas

1000
Tuning #2

P HVDC(MW)
800
600
400
Tuning #1
200
0
0 20 40 60 80 100
time (s)
2

 angle (degree)
1.5

0.5
0 20 40 60 80 100
time (s)

For this event study, the scope of real-time simulation including replicas are:
▪ Validate and improve the HVDC time domain models used in PDT software.
▪ Validate control behaviour before on-site modification
▪ Investigate potential solution in the control system
▪ Validate future C&P software updates
6.4.3 On-Site Modification of the C&P by RTE and REE
Once the C&P modification approach was tested and validated on the replica, RTE and REE wrote a
step-by-step procedure involving both operation centres and maintenance teams. The procedure was
to describe the course of operations for the on-site C&P modifications for both the single-link and the
multi-link control modes of the two HVDC links.
This procedure’s purpose was to ensure coordination between all the teams concerned by the on-site
modification, and provide a time-stamped step-by-step validation and verification protocol. The C&P
modification was successfully implemented on January 16th, 2019.

6.5 Protection studies

6.5.1 Introduction
Over the past decades IBRs have been increasingly integrated in power systems around the world.
Such IBRs include solar plants, Type 4 WTGs (also referred to as Full-Size Converter (FSC)), and Type
3 WTGs (also referred to as Doubly-Fed Induction Generator (DFIG)) whose schematics are shown in
Figure 6.5-1, Figure 6.5-2 and Figure 6.5-3, respectively.
A challenge faced by protection engineers is the anticipated impact of IBRs on conventional protection
functions [R71],[R72]. These functions have traditionally been designed and set with the expectation of
fault current signatures of conventional SGs. Such an assumption may no longer hold when IBRs
displace a large amount of SGs, thus leading to potential protection mis-operation issues. Examples of
such mis-operation have been reported for distance protection [R72]-[R75], negative-sequence
quantities-based protection [R76]-[R79], directional protective relay elements [R80], [R81], Rate-Of-
Change-Of-Frequency (ROCOF) [R82], and power swing protection [R83]-[R86]. Protection engineers
need to study these impacts, identify potential protection mis-operation problems, and develop solutions
to circumvent the mis-operation problems and ensure effective protection under high shares of IBRs.
An aspect of this study is to identify IBR modelling requirements to conduct the required protection
studies. This is necessary to ensure that protection engineers have confidence in their simulation models

140
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

and results. These modelling requirements are well-established for conventional SGs [R87]-[R90].
However, as these classical SG models cannot be used to represent an IBR, there is a need to establish
IBR modelling requirements for protection studies, which is preliminarily addressed in this report. The
focus is specifically on modelling requirements for power swing protection studies, and the modelling
requirements of other protection functions have been considered for future work.
Both an Electromagnetic Transient (EMT) IBR models and a phasor-domain IBR models can be
considered for use in power swing protection studies. The EMT model provides a detailed representation
of the IBR which represents IBR dynamics in a wide range of frequencies including both fast
(electromagnetic) and slow (electromechanical) dynamics. As such, the EMT model offers the highest
accuracy. However, the use of EMT models and simulation tools for power swing studies may be
challenging for a number of reasons. First, power swing is essentially an electromechanical transient
phenomenon which has traditionally been studied using transient stability-type simulations. Only the
slower IBR dynamics are expected to impact the power swing characteristics, and the representation of
faster (electromagnetic) dynamics, as offered by the EMT models, may not be necessary. Secondly,
practical power swing simulation studies often involve extensive simulation of a large-scale grid covering
many possible operating conditions, and for such applications EMT simulations may become
computationally demanding. Thirdly, protection coordination studies are typically carried out first in a
long-term planning assessment wherein detailed user defined models of IBRs may not exist resulting in
the use of generic models. In contrast to the EMT models, the phasor-domain IBR models provide a
quasi-steady state fundamental frequency representation which does not include the same level of
modelling details as that of the EMT models, but is less computationally demanding.
In conclusion, two types of IBR models can be used for power swing studies: EMT IBR models offer the
highest accuracy, however, their use for power swing protection studies may be challenging due to high
computational requirements and/or unavailability of required modelling details during protection
coordination studies. Phasor-domain models represent IBR dynamics in the typical range of stability
studies and do not accurately represent the minute details, but are less computationally demanding and
compatible with transient-stability-type tools traditionally used for power swing studies.
The objective of this section is to evaluate the accuracy of EMT and phasor-domain IBR models for
power swing protection studies. It cross-examines the power swing simulation results of the two models
with two specific objectives: (i) to study whether the key power swing characteristics calculated by the
two models are consistent and (ii) to study whether the results of the EMT and phasor-domain models
provide consistent conclusions regarding the impact of IBRs on power swing protection functions.
The simulation case studies have been carried out on the IEEE PSRC D29 test system [R89] including
IBRs. Two simulation models of this system have been developed: an EMT implementation which
includes a generic IBR model and an implementation within a positive-sequence program which includes
a generic phasor-domain IBR model. The two models have been subjected to the same power swing,
and the key power swing characteristics are compared.

Figure 6.5-1 PV Solar Farm

Figure 6.5-2 FSC WTG

141
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.5-3 DFIG WTG

6.5.2 Power Swing Protection

Power swing is a temporal variation in power flow caused by power system disturbances such as faults,
line switching, loss of infeed, and the loss or connection of large blocks of load [R88]. During a power
swing, the rotor angles of machines swing or oscillate relative to each other to achieve a new equilibrium
state. In a stable swing, the system regains a new state of equilibrium where the generator machine
rotor angle differences are within stable operating range to generate power that is balanced with the
load. In an unstable swing, the generation and load do not find a balance, the machines rotor angle
difference continues to increase, and the terminal voltage angle of a generator, or group of generators,
go past 180° with respect to the rest of the connected power system. This phenomenon is called pole
slip, and the condition is termed Out-Of-Step (OOS).
During a power swing, line distance protection mis-operation on an unfaulted line may occur, when the
voltage and current variations due to the swing cause the load impedance trajectory to enter the trip
zones of the protected line distance relay, misidentifying that as a fault on that line and causing
unintentional tripping of the line, which may further aggravate system stability. Power swing protection
prevents such unintentional tripping by supervising the operation of the line distance relay. Under a
stable swing, the power swing protection blocks the operation of the line distance relay to allow the
system to return to a new equilibrium. Under an unstable swing, the power swing protection prevents a
possible widespread outage and equipment damage through a controlled separation of the system at
predetermined locations. The idea is that the isolated portions of the system are most likely to survive
when network partitioning occurs at locations preserving a close balance between generation and load.
Power swing protection accomplishes these objectives by means of two main functions:
▪ Power Swing Blocking (PSB) which detects a power swing by differentiating it from a fault on the
protected line. This is done by measuring the rate of change of swing impedance; in a rotating
machine dominated system, this rate is slow during a power swing since additional current is
autonomously injected by rotating machines, whereas it is very fast during a fault on the protected
line as the large reduction in voltages negates the impact of injection of current.
▪ Out-of-Step Tripping (OST) which determines whether the swing is stable/unstable and initiates
system partitioning in case of an unstable swing. To determine stability/instability, this function uses
a point along the swing trajectory where the power system cannot regain stability; the swing is
declared unstable if the trajectory passes this point and stable otherwise. The optimal location for
the implementation of the OST protection is near the Electrical Center (EC) of the power system
which is the point or points in the system where the voltage becomes zero during an unstable power
swing. Typically, the location of OST relay systems determines the location where system
partitioning occurs during OOS.
Power swing detection elements have been available in protection relays for many years. A traditional
method of power swing detection has been to use a double blinder impedance zone characteristic to
measure the time taken for the positive sequence impedance to move between the outer and inner
blinder. This allows the relay to distinguish between fast moving impedances associated with short
circuit faults from slower moving impedance associated with conventional power swings. A typical timer
with a time delay setting is used: if the impedance takes longer than the time delay to cross between
the blinders then the disturbance is interpreted as a power swing. If faster than the time delay, it is
assumed to be a fault.
6.5.3 Expected misoperation issues due to IBRs
With the increasing share of IBRs, the power swing characteristics of a power system may considerably
change. This changed swing characteristic may have a detrimental impact on the performance of the

142
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

PSB and OST function and further impact the location of the EC for the implementation of the OST
protection [R82],[R83].
6.5.3.1 PSB misoperation
Reduced system inertia under high IBR penetrations results in faster power swings and increased rate
of change of swing impedance. The PSB function may mistakenly categorize such a fast power swing
as a fault on the protected line since its settings are selected with the expectation of slower swings under
high inertia operation. In such a case, the PSB function fails to block the designated line distance zone
elements during stable or unstable power swings, thus causing undesired line trip.
Figure 6.5-4 shows a case of PSB mis-operation due to IBRs. A power swing has been simulated under
SG (blue) and under 25% IBR integration (dashed black). The PSB function uses two concentric
impedance elements denoted by “outer” and “middle” and a timer with a time delay setting of 3 cycles
(i.e., the PSB time delay) to measure the rate of change of swing impedance. If the impedance takes
longer than 3 cycles to cross the outer and middle elements, the PSB declares a power swing and
blocks zones 1, 2, and 3 of the distance relay marked by Z1, Z2, and Z3, respectively. However, if the
impedance takes less than 3 cycles to cross the outer and middle elements, the event is classified as a
system fault and the distance zones are enabled.
The power swing begins at t0 = 5s. As shown, under SG the impedance crosses the outer and middle
elements in 3.5 cycles which is larger than the PSB time delay, and hence the PSB successfully declares
a power swing and blocks the distance zones. By contrast, under 25% IBR the swing impedance moves
faster due to reduced system inertia and crosses the outer and middle elements in 2.7 cycles. This time
delay is less than the PSB setting of 3 cycles, and consequently the PSB function fails to detect the
swing and mistakenly classifies it as a fault on the protected line, resulting in undesired trip the line
which was unfaulted.

Figure 6.5-4 PSB mis-operation due to a faster swing under IBR

6.5.3.2 OST malfunction

Reduced system inertia together with the different dynamic characteristics of IBRs may significantly
change the swing impedance trajectory when IBRs displace SGs. The trajectories of stable/unstable
swings under IBRs may be substantially different than those under SGs. Due to this changed swing
trajectory, the OST function may mistakenly classify a stable swing as unstable and cause unnecessary
partitioning of the system.
Figure 6.5-5 to Figure 6.5-6 illustrates an example of an OST malfunction due to IBRs. A stable power
swing has been simulated under SG (blue) and under 50% IBR integration (dashed black). The OST
function uses a concentric impedance element denoted by “inner” to differentiate between a stable and
unstable swings. This inner element is designed such that the most severe stable swing does not cross
it. Hence, if a given swing trajectory does/does not cross this inner element, the swing is declared
unstable/stable. The settings of this element are selected such that only an unstable swing cross it. As
shown, under SG the swing impedance trajectory initially moves toward the inner element, but reverses
direction just before crossing it. Hence, the OST function successfully classifies the swing as stable,
and system partitioning is not initiated to allows the system to reach a new state of equilibrium. By

143
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

contrast, under 50% IBR the changed swing trajectory crosses the inner element before reversing
direction, thus resulting in incorrect classification of the swing as unstable. This incorrect classification
may lead to unnecessary partitioning of the system by the OOS protection.

Figure 6.5-5 OST Malfunction due to changed swing trajectory under IBRs

6.5.3.3 Impact on EC and the optimal location of OST protection:

It can be shown that the location of EC of a power system depends on the equivalent impedance of
generators [R89]. This equivalent impedance becomes different when the generator is an IBR instead
of a SG, which may change the location of EC. Hence, when IBRs displace a substantial amount of
SGs, it may be necessary to recalculate the location of EC for the optimal placement of the OST
protection relays.
Figure 6.5-6 provides an example of the impact of IBRs on EC by comparing the swing trajectory of an
OOS condition under SGs, 25% IBR, and 50% IBR. As shown, under SG the EC is at 0.22+j4.18 Ω, and
the swing locus passes through a transmission line denoted by L3. Hence, line L3 is the optimal location
to implement the OST function. However, under 25% and 50% IBR, the EC moves to 0.20+j3.76 Ω and
0.14+j2.59 Ω, respectively, and the locus passes through a different transmission line denoted by L1,
thus changing the optimal location of OST relays to line L1. This case study shows that it is necessary
to recalculate the location of EC under IBRs to ensure that the OST function is implemented in the
optimal location.

Figure 6.5-6 Impact of IBR on EC

144
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Given the above impacts, it is necessary to identify such mis-operation issues and develop solutions to
ensure effective power swing and OOS protection under high shares of IBRs. Due to the complexity of
the power system, analytical calculation of power swing characteristics is challenging, and power swing
relay calculations are done using transient simulation studies. The required modelling details and
simulation tools for such studies are well-established for traditional power systems energised by SGs.
However, these traditional models may be inadequate for dynamic studies involving IBRs, due to the
fundamental differences in dynamic behaviour of IBRs and SGs. Hence, it is necessary to identify the
modelling requirements for power swing studies under IBRs by comparing power swing simulation
results from EMT models with results from phasor-domain models.
Implementations of the IEEE PSRC D29 test system have been developed within both an EMT program
(containing EMT models) and a positive-sequence program (containing phasor models). The first step
is to ensure that these two implementations are consistent when there is no IBR in the simulation model
and all generators are SG. The cross-examination of simulation results, suggests that the two
implementations are consistent when all generators are SGs. Next, IBRs are added to the test system
and the results are cross-examined.
6.5.4 Inverter modelling requirements
This section compares the power swing results of an EMT and a phasor-domain simulation (PDT) model
of IBRs. There are two main objectives: (i) to study whether the key power swing characteristics
calculated by the two models are consistent; and (ii) to study whether the results of the two models
provide the same conclusions regarding the impact of IBRs on power swing protection.
The IEEE PSRC D29 test system has been subjected to different power swing events under various
IBR integration levels:
▪ 0% IBR in which all generators of the test system are SGs;
▪ 25% IBR realized by replacing the SG on Bus16 by a Type 3-based IBR with consistent MW
capacity as that of the replaced SG; and
▪ 50% IBR realized by replacing the SGs on Bus14 and Bus16 by Type 3-based IBRs with
consistent MW capacity as that of the replaced SG.
The three IBR integration levels have been realized by adding Type 3-based IBR models to the PDT
positive-sequence and EMT implementations. The EMT implementation uses a generic EMT Type 3-
based model which has been field-validated in [R91]. The model represents the WTGs, feeders, and
collector grid of the Wind Park (WP) by their aggregated model while preserving the overall control
structure of the WP. The modelling details include wind turbine aerodynamics, mechanical system,
induction machine, back-to-back converter, line inductor (choke filter) and ac harmonic filters, WTG
transformer, equivalent collector grid, and WP transformer. In terms of control scheme, the EMT model
includes WP controller, Grid-Side Converter (GSC) current and voltage controllers, Rotor-Side
Converter (RSC) current controller, maximum power point tracking (MPPT) algorithm, converter current
limiter and PQ priority schemes, and FRT functionality. The GSC and RSC current control loops have
been represented by a vector current control scheme with a PI controller. The GSC current control mode
is assumed to be Coupled Sequence Control (CSC) (no control on negative sequence current) for all
WPs. The WP control mode is assumed to be Q-control, the crowbar protection has been deactivated,
and FRT has been activated.
The PDT implementation uses the 2nd generation generic positive sequence model [R92], however the
mechanical and aerodynamic modules were omitted. Other modelling differences with respect to the
EMT model include no representation of the GSC/RSC inner current control loop, an approximate PLL
model [R92], no representation of the electrical or mechanical dynamics of the induction machine, and
no separate modelling of GSC and RSC.
Table 6-2 summarizes the key modelling differences between the PDT and the EMT models. These
generic models are intended to provide a reasonable representation of the trend of dynamic behaviour
of an IBR and do not accurately represent the minute details. While the representation of this trend may
be sufficient from the perspective of dynamic planning studies, it is possible that they could result in
erroneous operation of relay models.

145
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Table 6-2 Key modelling difference between cross-examined PDT and EMT models of a Type 3 WTG
model
Phasor-domain model
EMT model
Component (2nd generation positive-
(generic EMT model of [R92])
sequence model)
Represented by first order time Represented by a vector current
GSC inner current control loop
constant. control scheme with a PI controller.
Represented by a vector current
RSC inner current control loop Not represented.
control scheme with a PI controller.
PLL Approximate PLL model. Detailed circuit-based PLL model.
Detailed representation of electrical
Induction machine Not represented.
and mechanical dynamics.
Detailed aspects of the converters Detailed aspects of the converters
RSC and GSC are not represented. RSC and GSC are represented. RSC and GSC are
are not separately modeled. separately modeled.
Mechanical and aerodynamic
Not represented. Represented.
models

A power swing has been simulated by applying a 300-ms fault on Bus12 followed by the outage of the
line connecting Bus12 to Bus15.
Figure 6.5-7 shows the results under 0% IBR. The two power swing impedance trajectories exhibit a
close match; for illustration purposes only the part where the impedance crosses the outer and middle
elements and the corresponding time stamps have been shown in the figure. As shown, the impedance
takes 8.1 cycles to cross the two elements in the positive-sequence program compared to 8.5 cycles in
EMT.
Figure 6.5-8 shows the simulation of the same swing under 25% IBR, suggesting the consistency of the
key swing characteristics calculated by the two tools. Specifically, the starting point of the swing is -
26.89-j4.29 Ω in the PDT program which is close to -18.63-j17.63 Ω in the EMT program. Further, the
impedance takes 6.4 cycles to cross the two elements in PDT which is close to 6.1 cycles in EMT.
Additionally, the calculated location of EC is 0.96+j6.26 Ω in PDT which is close to 1.04+j6.83 Ω in EMT,
and in both cases the calculated swing trajectory passes through the line connecting Bus16 to Bus8.
The results further exhibit some inconsistencies, particularly in the initial cycles of the swing. These
initial differences are expected due to the substantially different modelling of faster dynamics including
induction generator model, RSC and GSC models and the associated inner current controls. As the
swing continues, the trajectories become more consistent.
Figure 6.5-9 shows the results under 50% IBR. With the increased IBR level, the modelling differences
between the EMT and PDT Type 3 models lead to more significant inconsistencies between the results.
Specifically, the time delay between the outer and middle elements calculated in PDT and EMT become
6.4 cycles and 4.3 cycles, respectively. The location of EC still shows an acceptable match as it is
calculated as 0.97+j6.82 Ω in positive-sequence and 0.98+j6.51 Ω in EMT. The pattern of the impedance
trajectories also shows more inconsistencies compared to the previous cases due to the more
significantly pronounced modelling differences.
The rate of change of swing impedance trajectory tends to increase with increasing IBR shares, and an
objective of this section is to show that the EMT and phasor-domain IBR models show this effect. Table
6-3 shows this by comparing the PSB time delay calculated in positive-sequence and in EMT domain
under the three IBR integration levels above. As shown, both models show the reduction of the PSB
time delay with increasing IBR level. At 50% IBR level, the PDT representation shows an inconsistent
PSB time delay to that of the EMT representation, which is due to modelling and parameterisation
differences.

146
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.5-7 Comparison of power swing simulation results of the IEEE PSRC D29 test system under 0%
IBR in EMT and PDT simulations.

Figure 6.5-8 Comparison of power swing simulation results of the IEEE PSRC D29 test system under 25%
IBR in EMT and PDT simulations.

147
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure 6.5-9 Comparison of power swing simulation results of the IEEE PSRC D29 test system under 50%
IBR in EMT and PDT simulations.
Table 6-3 Calculated PSB time delay from the PDT and EMT simulations as a function of IBR integration
level
IBR integration level (%) PSB time delay (cycles)
Positive-sequence program EMT
(PDT)
0 8.1 8.5
25 6.4 6.1
50 6.4 4.3

The above case studies and simulation results suggest that the phasor-domain IBR model can provide
consistent power swing simulation results compared with those of the more detailed EMT model for
conventional power systems; however, as IBR level increases, the modelling differences may lead to
less consistent results, thus highlighting the need for further improvement of the consistency between
the EMT and phasor-domain (PDT) IBR models and values of parameters for power swing protection
studies.
6.5.5 Conclusion
This section has presented a study as a first step towards establishing IBR modelling requirements for
power swing protection simulation studies under IBRs. Specifically, two Type 3 WTG-based IBR models
have been considered, namely a generic EMT and a 2nd generation generic positive sequence model
(PDT). The EMT model uses a detailed representation of the WTG including detailed aspects of the
converters, control schemes, and mechanical and electrical dynamics of the induction machine. As
such, the EMT model offers the highest accuracy by representing both very fast (electromagnetic) and
slow (electromechanical) dynamics of the WTG. The PDT 2nd generation generic positive-sequence
model only represents IBR dynamics in the typical range of stability studies and does not accurately
represent the minute details. The cross-examination results suggest that the PDT positive-sequence
IBR model can provide consistent power swing simulation results compared with those of the more
detailed EMT model for conventional power systems; however, as IBR level increases, the modelling
differences may lead to less consistent results, thus highlighting the need for further improvement of the
consistency between the EMT and phasor-domain IBR models for power swing protection studies.

6.6 System separation study

To analyze the performance of Hydro-Quebec’s 735 kV grid protection system during extreme
disturbances and network separation, it is necessary to simulate various scenarios using detailed
electromagnetic models including nonlinearities, such as transformer saturation, and metal-oxide
varistors. With modern computational capabilities in EMT solvers, it is possible to perform such detailed
simulations directly. The R-Network (reduced-network described in section 1.3.7) is used here with
complete detailed synchronous machine (including saturation) models. The network solution starts

148
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

directly from the load-flow solution and moves into the steady-state solution before the occurrence of
the first event. The scenario studied here is referring to the Hydro-Quebec series-compensated network
bus names shown in [R12]. A similar scenario is discussed in more detail in [R12]. The sequence of
events is the following:
▪ t = 0.1s : occurrence of a three-phase fault on the Chibougamau-Chamouchouan corridor
▪ between t = 0.2 s and t = 0.25 s: the fault is cleared resulting into the loss of the three Chibougamau-
Chamouchouan lines and Albanel-Chibougamau tie
▪ t = 0.45 s: LG 4 power plant rejection
▪ t = 0.53 s: remote shedding of 2500 MW of load in the Montreal area
▪ t = 0.78 s: the protection system detects the 62 Hz threshold for switching on the sacrificial metal-
oxide surge arresters at five substations for the potential network separation scenario
The simulation results are presented in Figure 6.6-1. These results cannot be obtained using phasor-
domain simulations and the network instability conditions cannot be studied. Further analysis is available
in [R12] and [R14].

Figure 6.6-1 James Bay system voltage oscillations due to an extreme disturbance

149
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

7. Conclusions and further work

This section summaries key conclusions of this TB, and provides a few suggestions for future work.

7.1 Key conclusions

▪ Inadequacy of PDT models in power systems with high IBR penetration
 Power system modelling engineers around the world are experiencing growing instances
where PDT models fall short in accurately predicting power system dynamic performance
under high IBR scenarios. This is becoming more prevalent as the IBR penetration increases
in each system. The use of PDT modelling alone, cannot ensure secure, reliable and efficient
planning and operation of such power systems.
▪ Wide-area EMT modelling is a reality.
 Wide-area EMT models comprising up to several hundred dynamic models and several
thousand busbars have been successfully developed and tested. These are currently in use
for a range of applications including generator connection studies, and power system planning
and operation. Whilst these models are currently in use in a few countries only, it is expected
that several other countries especially those experiencing a rapid IBR uptake will develop and
use these models in the next 2-5 years.
▪ Key applications to date
 The focus has so far been on generator connection studies and operational decision making.
For these applications the use of vendor-specific site-specific models is a necessity.
Application for long-term power system planning studies has been limited due to additional
challenges in developing power system dynamic models for the future network and future
power system equipment’s capabilities and limitations. This is recognising that the existing
vendor-specific black-box models will not be suitable for representing future grid-forming and
grid-following IBR performance.
▪ The need for clear and upfront modelling requirements
 Systematic methods for developing detailed IBR EMT models from the actual control code of
the IBR is widely used across most OEMs. This approach has reduced the time required to
develop and validate such models. However, even so clear and upfront modelling
requirements are essential such that OEMs will have sufficient time to develop, test and
validate these models, and can provide the level of details sought without compromising their
IP. This is becoming more important as the blackbox models are becoming less insightful in
scenarios where adverse control system interactions between several IBRs is the main point
of interest.
▪ Vendor-specific vs generic-models
 The use of generic models for long-term planning studies is more prudent than vendor-specific
models. Considering that rapid technology changes can occur within the shortest planning
timeframe of five years, there is often uncertainty regarding the precise type and make of the
generation to be connected, and changes to network topology may occur in different stages,
the use of site-specific, vendor-specific dynamic models is not practical for long-term planning
studies. Generic models are, therefore, the most realistic approach. Increased uptake of IBRs
and the types of new phenomena already being observed means that EMT modelling will need
to play a greater role in long-term planning studies going forward.
▪ Best practices for wide-area network model development
 The most common approach is to import an existing PDT network load flow model into an EMT
tool. Once complete, the user will populate true EMT models (manually or through automatic
replacement libraries) that will likely have material impacts on the outcome of a specific study.
The systematic and automated creation of EMT load flow snapshots from the resultant PDT
load flow snapshots is becoming the common practice rather developing and using a handful
of EMT base cases for minimum and maximum demand and generation.
▪ The need for model testing and validation
 EMT models provide the opportunity and flexibility to include as much details as sought
necessary for accurate IBR modelling. However, they cannot be considered inherently
accurate unless tested and proven. Model acceptance tests are necessary to provide the
confidence that the model is usable and numerically robust, and represents the installed plant
under reasonably expected operating conditions. This will be conducted once the model is

150
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

submitted for the first time. Once the plant is connected and commissioned, model validation
tests on the installed plant will provide further confidence in the accuracy of plant EMT models
compared to the actual system measurements. The use of data associated with natural system
disturbances that occur during the lifetime of the plant is also helpful.
▪ Wide-area EMT modelling challenges and considerations
 Role and responsibilities
o Generally system operators and network owners are the only entities permitted
to have access to the entire wide-area EMT models. However, these
organisations are not generally power system equipment design and control
system experts, and in any case the use of black-box models provides limited
opportunity to make judicious changes in the plant model without unduly
compromising other aspects of the performance anyway. On the other hand,
OEMs have expert knowledge of their products, however, it often happens that
an optimal control system design and tuning cannot be developed without
access to other IBR models, which is not often permitted by power system
modelling requirements in various countries.
 Control system tuning based on black-box models
o As discussed in the previous point
 Impact on project timeframe
o Despite producing more accurate results, the use of wide-area EMT modelling
can increase the overall studies timeframe by up to two times.
 Engineering resources required
o Traditionally EMT studies have been considered as niche power system studies
where the expertise primarily resides in a few specialised consultancy
companies. The use of wide-area EMT modelling for day-to-day power system
studies has challenged this status, resulting in the need for several more power
system modelling engineers with EMT modelling and analysis capabilities.
 Computing resources required
o Wide-area EMT studies in large-scale power system could take between 1 and
10 hours depending on the size of the system and computing resources used.
The use of highly capable on-site computing resources will reduce the
computational time, still in the order of one or more hours, however it would
typically cost several hundred thousand dollars. Cloud computing has been
emerging as a more cost-effective approach

7.2 Suggestions for future work

▪ Faster simulation speed
 Much faster simulation speeds are required to allow a broader integration of EMT models in
planning and in particular operational decision-making tools in the same fashion PDT models
have been used for decades. For example, integration into dynamic security assessment tools
would require the completion of tens of simulation studies over a few minutes which is not
currently possible. Near-real-time EMT simulation will therefore be required to facilitate a
broader use of wide-area EMT models.
▪ More systematic development of boundary buses
 The combined use of FDNE and dynamic equivalent voltage sources is the most accurate but
also the most involved modelling approach. As such this approach has not been applied to
date for any practical power system studies under high IBR penetration scenarios. However,
the need for such an approach will become more important as power systems worldwide are
getting closer to 100% IBR operation scenarios over the coming years.
▪ Robust data management
 The move to extremely detailed models that may be using actual source-code from units in
the field increases the challenge of ensuring EMT models are up to date. This is particularly
challenging when OEMs are capable of remote upgrades of plant, and where there is no
standardised form of EMT model framework. Further work regarding how to best manage the
influx of modelling data and updates in standard database packages would be beneficial to
the modelling community.

151
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

▪ Developing more suitable screening methods

 As discussed EMT tools are costly to run, both in terms of the computational burden and the
infrastructures required. Therefore, they cannot be conveniently used when sensitivity to
hundreds of different factors is the point of interest. Screening methods have been sometimes
used by system operators and network owners to provide an indication of when EMT studies
are a necessity as opposed to the PDT studies. However, their use in power systems with high
IBR penetration is gradually moving towards answering the question of how to limit the number
of EMT studies required recognising that sole reliance on PDT studies will not be prudent. This
TB has consolidated and compared a few screening methods already available in the public
domain. However, dramatically changing generation mix and power systems would result in
the emergence of new and emerging power system phenomena which may not be predicted
by existing screening methods, which are primarily based on fault level and short circuit ratio.
▪ Integration of protection system models into wide-area EMT models for dynamic studies
 As discussed in this TB, the response of the protective relays can differ under high IBR
penetration scenarios. Wide-area EMT studies are generally conducted with detailed
modelling of various generation technologies, and the transmission and distribution network
assets. However, these studies do not often include a dynamic model for protection relays.
Further work is recommended to develop sufficiently detailed EMT models of relevant
protection relays from various OEMs, and to assess the impact on conclusions made about
overall system performance under high IBR scenarios.
▪ More detailed modelling of inverter controls for distributed energy resources.
 The model discussed in this TB will be sufficient to account for inadvertent disconnection of
the DER during the credible and non-credible contingencies. However, it is not suitable for
analysing the impact of high DER penetration on bulk power system services such as system
strength or investigating any potential adverse interactions within the distribution network, or
between the transmission and distribution networks. These could occur due to the residual
instability impact of the DER not presently accounted for.
▪ The use of these models to answer some of the following key questions:
 The extent to which grid-forming inverters can substitute large centralised synchronous
generators and how system dynamics will change is not currently known, in particular as often
grid forming inverters are not installed in the same locations as the synchronous generators
 The necessary mix of grid-forming and grid-following IBRs as the power system evolves, and
to determine what is the minimum synchronous characteristics required
 A better understanding of new and emerging forms of system stability and other phenomena,
i.e., power quality and power system protection, with a potential adverse impact on system
security
 A better understanding of new and emerging technology capabilities and limitations
 A better understanding of system operation with few synchronous generators and many
synchronous condensers

152
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

References
[R1] Odessa Disturbance. Texas Events: May 9, 2021 and June 26, 2021. Joint NERC and Texas
RE Staff Report, September 2021,
https://www.nerc.com/pa/rrm/ea/Documents/Odessa_Disturbance_Report.pdf
[R2] Opening a Proceeding to Review Hawaiian Electric’s Interconnection Process and Transition
Plans for Retirement of Fossil Fuel Power Plants, Response to Commission’s information
requests, 2021
https://dms.puc.hawaii.gov/dms/DocumentViewer?pid=A1001001A21F14B62327F00172
[R3] B. Ahern, A Rawat, R. Somayajulu, B. Marszalkowski, L. Unruh, A. Isaacs, EMT Evaluation
of Transmission Impact from Large Scale DER Integration, CIGRE Science & Engineering
Volume No.21, June 2021, pp. 132-140, https://e-cigre.org/publication/cse021-cse-021
[R4] System Security Market Frameworks Review, Final Report, AEMC, 27 June 2017: System-
Security-Market-Frameworks-Review-Final-Report.pdf (aemc.gov.au)
[R5] System Security Market Frameworks Review. AEMO, 1 July 2018:
https://aemo.com.au/en/energy-systems/electricity/national-electricity-market-nem/system-
operations/system-security-market-frameworks-review
[R6] Renewable Energy, EirGrid, http://www.eirgridgroup.com/how-the-grid-works/renewables/
[R7] EirGrid Group Strategy 2020-25 “Transform the Power System for Future Generations”,
EirGrid, 2019, http://www.eirgridgroup.com/about/strategy-2025/EirGrid-Group-Strategy-
2025-DOWNLOAD.pdf
[R8] M. Val Escudero, A. Martin, L. Fisher and I. Dudurych, Technical challenges associated with
the integration of long HVAC cables and inverter based renewable generation in weak
transmission networks: the Irish experience, C4-110 paper, CIGRE Paris Session 2018,
https://e-cigre.org/publication/SESSION2018-2018-cigre-session
[R9] A. Mansoldo, M. Norton, M. Val Escudero, The Electromagnetic Transient Analysis in the
decision making process at planning stage, EEUG meeting, September 2013.
[R10] S. Dennetiere, H. Saad, Y. Vernay, P. Rault, C. Martin, B. Clerc, Supporting energy transition
in transmission systems: An operator's experience using electromagnetic transient
simulation, 2019, IEEE Power and Energy Magazine, Volume 17, Issue 3, pp. 48-60.
[R11] U. Karaagac, H. Gras, J. Mahseredjian, A. El-Akoum, X. Legrand, Synchronous Machine
Exciter Circuit Model In A Simultaneous Field Winding Interface, IEEE PES General meeting
2013, Vancouver
[R12] Q. Bui-Van and M. Rousseau, Control of Over-voltages on Hydro-Québec 735-kV Series-
Compensated System During a Major Electro-mechanical Transient Disturbance,
Proceedings of International Conference on Power Systems Transients, IPST 2001, Rio De
Janeiro
[R13] J. Mahseredjian, S. Dennetière, L. Dubé, B. Khodabakhchian and L. Gérin-Lajoie, On a new
approach for the simulation of transients in power systems, Electric Power Systems
Research, 2007, Volume 77, Issue 11, pp. 1514-1520
[R14] L. Gérin-Lajoie, J. Mahseredjian, Simulation of an extra-large network in EMTP: from
electromagnetic to electromechanical transients, International Conference on Power
Systems Transients, 2009, Kyoto, Japan
[R15] A. Abusalah, O. Saad, J. Mahseredjian, U. Karaagac, I. Kocar, Computation of
Electromagnetic Transients Through Parallel Solution of Sparse Matrices, IEEE Power and
Energy Technology Systems Journal, 2019, Volume 7, pp. 13-21
[R16] I. Kocar, J. Mahseredjian, R.E. Uosef, B. Cetindag, Simulation of transients in very large
scale distribution networks by combining input text files with graphical user interface, Electric
Power Systems Research, 2016, Volume 138, pp. 146-154

153
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

[R17] O. Saad et. al., Simulation of Large-Scale Grids in EMT: The Hydro-Quebec Experience,
CIGRE Paris e-Session 48, Paris, 2020
[R18] PSS® power system simulation and modeling software, Siemens,
https://new.siemens.com/global/en/products/energy/energy-automation-and-smart-
grid/pss-software.html
[R19] Panhandle System Strength Study Feb 23 2016, ERCOT, 2016,
http://www.ercot.com/content/news/presentations/2016/Panhandle%20System%20Strengt
h%20Study%20Feb%2023%202016%20(Public).pdf
[R20] Panhandle and South Texas Stability and System Strength Assessment, Electranix, 2018,
https://www.ercot.com/files/docs/2018/04/19/Panhandle_and_South_Texas_Stability_and_
System_Strength_Assessment_March....pdf
[R21] 2019 Panhandle Regional Stability Study, ERCOT, 2019
https://www.ercot.com/files/docs/2020/11/27/2019_PanhandleStudy_public_V1_final.pdf
[R22] J. Zhou, H. Ding, S. Fan, Y. Zhang and A. Gole, Impact of Short- Circuit Ratio and Phase-
Locked-Loop Parameters on the Small- Signal Behavior of a VSC-HVDC Converter, IEEE
Transactions on Power Delivery, 2014, Volume 29, Issue 5, pp. 2287-2296
[R23] Y. Zhang, J. Schmall, S. Huang, J. Conto, J. Billo, and E. Rehman, Evaluating System
Strength for Large-Scale Wind Plant Integration Proceedings of 2014 IEEE PES General
Meeting, July 27-31, 2014, National Harbor, MD.
[R24] B. Badrzadeh et. al., The need for enhanced power system modelling techniques and
simulation tools, CIGRE Science & Engineering Journal, 2020, Volume 17
[R25] K. Samarawickrama, A. Kariyawasam, S. Kariyawasam, Investigation of Stability Issues
introduced by Network Reduction in EMT Simulations, IPST 2019
[R26] IEEE STD 1110-2019, IEEE Guide for Synchronous Generator Modeling Practices and
Parameter Verification with Applications in Power System Stability Analyses,
https://ieeexplore.ieee.org/document/9020274
[R27] P.S. Kundur, Power System Stability and Control, McGraw-Hill, New York, ISBN:
007035958X
[R28] P. Pourbeik, B. Agrawal, S. Patterson and R. Rhinier et al., Modeling of synchronous
generators in power system studies, CIGRE Science & Engineering Journal, 2016, Volume
6 October 2016
[R29] Use of GENTPJ Generator Model, North American Electric Reliability Corporation (NERC),
https://www.nerc.com/comm/PC/NERCModelingNotifications/Use%20of%20GENTPJ%20
Generator%20Model.pdf
[R30] Dynamic Model Improvements, ERCOT, 2021,
http://www.ercot.com/mktrules/issues/PGRR085
[R31] Dynamic Model Acceptance Test Guidelines, AEMO, 2021, https://aemo.com.au/-
/media/files/electricity/nem/network_connections/model-acceptance-test-guideline-feb-
2021.pdf
[R32] Technical Brochure 671 (2016) Connection of Wind Farms to Weak AC Networks, WG
B4.62, www.e-cigre.org
[R33] Technical Brochure 068 (1992) Guide for Planning DC Links Terminating at AC Locations
Having Low Short-Circuit Capacities, Part 1: AC/DC Interaction Phenomena, WG 14.07,
www.e-cigre.org
[R34] Technical Brochure 364 (2008) Systems with multiple DC Infeed, WG B4.41, www.e-
cigre.org
[R35] H. Saad, S. Dennetière and B. Clerc, Interaction investigations between power electronic
devices embedded in HVAC network, 13th IET International Conference on AC and DC
Power Transmission (ACDC 2017), 2017, Manchester, pp. 1-7.

154
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

[R36] Y. Vernay, B. Gustavsen, Application of Frequency-Dependent Network Equivalents for

EMTP Simulation of Transformer Inrush Current in Large Networks, 2013, International
Conference on Power Systems Transients, Vancouver, Canada, July 18-20, 2013
[R37] W. Wang, G. M. Huang, D. Ramasubramanian, E. Farantatos, Transient stability analysis
and stability margin evaluation of phase-locked loop synchronised converter-based
generators, IET Generation, Transmission & Distribution, 2020, Volume 14 Issue 22, pp.
5000-5010. Special Issue: Challenges and New Solutions for Enhancing Ancillary Services
and Grid Resiliency in Low Inertia Power Systems.
[R38] W. W. Baker, D. Ramasubramanian, M. Val Escudero, E. Farantatos and A. Gaikwad,
Application of an Advanced Short Circuit Strength Metric to Evaluate Ireland's High
Renewable Penetration Scenarios," The 9th Renewable Power Generation Conference
(RPG Dublin Online 2021), 2021, pp. 167-172.
[R39] W. Wang, D. Ramasubramanian, E. Farantatos, A. Gaikwad, D. Bowman, H. Scribner, J.
Tanner, C. Cathey, C. Cates, J. Caspary. Evaluation of Inverter Based Resources Transient
Stability Performance in Weak Areas in Southwest Power Pool’s System Footprint. Paper
C4-116. Cigre Paris e-session 2020.
[R40] B. Badrzadeh et. al. “Large-scale electromagnetic transient model validation based on
measured system disturbances”, CIGRE e-Session 48, Paris, 2020
[R41] L. Shuai et. al., Eigenvalue-based Stability Analysis of Sub-Synchronous Oscillation in an
Offshore Wind Power Plant, 17th International Wind Integration Workshop, Stockholm,
Sweden, Oct. 2018
[R42] Articles 38 and 39 of Commission Regulation (EU) 2017/1485 of 2 August 2017 establishing
a guideline on electricity transmission system operation, https://eur-
lex.europa.eu/eli/reg/2017/1485/oj
[R43] The Migrate Project, https://www.h2020-migrate.eu/
[R44] EU-sysflex, https://eu-sysflex.com/
[R45] B. Badrzadeh, T. Yip, Wind Turbine Modeling – the Vestas Perspective, 10th International
Workshop on Large-Scale Integration of Wind Power into Power Systems, Århus, Denmark,
2011
[R46] National Renewable Energy Laboratory (NREL) report NREL/CP-500-42886, “Method of
Equivalencing for a Large Wind Power Plant with Multiple Turbine Representation”, July
2008, https://www.nrel.gov/docs/fy08osti/42886.pdf
[R47] P. Pourbeik et al, “An aggregate dynamic model for distributed energy resources for power
system stability studies’’, CIGRE Science and Engineering Journal, N°20, Feb 2021
[R48] Technical Brochure 556 (2013) Power System Technical Performance Issues Related to the
Application of Long AC cables, WG C4.502, www.e-cigre.org
[R49] Technical Brochure 766 (2019) Network Modelling for Harmonic Studies, JWG C4/B4.38,
www.e-cigre.org
[R50] Technical Reference Document - Dynamic Load Modeling, NERC, 2016
https://www.nerc.com/comm/PC/LoadModelingTaskForceDL/Dynamic%20Load%20Modeli
ng%20Tech%20Ref%202016-11-14%20-%20FINAL.PDF.
[R51] Composite Load Model Specifications, WECC, 2015
http://home.engineering.iastate.edu/~jdm/ee554/WECC%20Composite%20Load%20Model
%20Specifications%2001-27-2015.pdf
[R52] P. Pourbeik et al, “An aggregate dynamic model for distributed energy resources for power
system stability studies’’, CIGRE Science and Engineering Journal, N°14, June 2019.
[R53] Distributed Energy Resource Modeling Guideline, NERC, 2017,
https://www.nerc.com/comm/PC_Reliability_Guidelines_DL/Reliability_Guideline_-
_DER_Modeling_Parameters_-_2017-08-18_-_FINAL.pdf
[R54] ENTSO-E Standardized control interface for HVDC SIL/HIL conformity tests, 21 April 2020
https://www.entsoe.eu/2020/04/24/entso-e-standardized-control-interface-for-hvdc-sil-hil-
conformity-tests

155
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

[R55] IEC 61400-27-2, IEC Standard, Wind energy generation systems – Part 27-2: Electrical
simulation models – Model validation, Edition 1.0 2020-07
[R56] C. Dufour et. al., A Combined State-Space Nodal Method for the Simulation of Power System
Transients, IEEE Transactions on Power Delivery, Volume: 26, Issue: 2, April 2011
[R57] R. K. a. W. Schiehlen, Two methods of simulator coupling, in Math. Comput. Model. Dyn.
Syst., vol. 6, no. 2, pp. 93–113, 2000
[R58] P. A. C. G. a. T. V. C. F. Plumier, Co-simulation of electromagnetic transients and phasor
models: A relaxation approach, in IEEE Transaction on Power Delivery, Vol 31, N° 5, pp.
2360-2369, 2016
[R59] AEMO Real-time Simulator Project Update, AEMO, 2020,
https://aemo.com.au/en/newsroom/news-updates/aemo-real-time-simulator-project-update
[R60] AEMO power system simulator moves to implementation phase, AEMO, 2021,
https://aemo.com.au/en/newsroom/news-updates/aemo-power-system-simulator-moves-
to-implementation
[R61] L. Grégoire et. al., Real-Time Simulation-Based Multisolver Decoupling Technique for
Complex Power-Electronics Circuits”, IEEE Transactions on Power Delivery, Volume: 31,
Issue: 5, Oct. 2016
[R62] Y. Dong et al, Hardware-in-the-loop simulation and test of a control and protection system
for MMC-based UPFC, 2016 IEEE Power and Energy Society General Meeting (PESGM)
[R63] C. Dufour et. al., Hardware-in-the-loop closed-loop experiments with an FPGA-based
permanent magnet synchronous motor drive system and a rapidly prototyped controller,
2008 IEEE International Symposium on Industrial Electronics
[R64] J. Martinez-Velasco, Transient Analysis of Power Systems: Solution Techniques, Tools and
Applications, Wiley-IEEE Press, 2015
[R65] S. Dennetière, H. Saad, Validation of MMC station real-time models with field tests CIGRE
B4 Conference, Winnipeg, Canada, September 30 – October 6, 2017
[R66] H. Saad, S. Dennetiere, C. Lallemand, B. Clerc, Y. Vernay, Commissioning of the France
Spain HVDC VSC control system replicas. InProc. CIGRE-IEC Colloquium 2016 May 9 (pp.
1-8).
[R67] C. Hardt et. al., Practical experience with mitigation of sub-synchronous control interaction
in power systems with low system strength, CIGRE Science & Engineering, Vol 21, June
2021
[R68] “Analysis of CE Inter-Area Oscillations of 1st December 2016”, ENTSO-E SG SPD Report,
13.07.2017
[R69] H. Saad, A. Schwob, Y. Vernay, Study of Resonance Issues between HVDC link and Power
System Components using EMT Simulations, 2018 Power Systems Computation
Conference (PSCC), Dublin, 2018, pp. 1-7
[R70] A. Dominguez Ferrer, J. M. Argüelles Enjuanes, G. D. Castejón, J. Loncle, et al, Feedback
on INELFE France Spain HVDC Project, CIGRE Conference, Paris, France, Aug. 2016
[R71] Impact of inverter-based generation on bulk power system dynamics and short-circuit
performance, PES-TR68, prepared by the IEEE/NERC Task Force on Short-Circuit and
System Performance Impact of Inverter Based Generation, Jul. 2018.
[R72] Advanced Short-Circuit Modeling, Analysis, and Protection Schemes Design for Systems
with Renewables – TVA Case Study. EPRI, Palo Alto, CA: 2017. 3002010950.
[R73] A. Hooshyar, M. A. Azzouz, and E. F. El-Saadany, Distance protection of lines connected to
induction generator-based windfarms during balance faults, IEEE Trans. Sustain. Energy,
vol. 5, no. 4, pp. 1193–1203, Oct. 2014.
[R74] A. Hooshyar, M. A. Azzouz, and E. F. El-Saadany, Distance protection of lines emanating
from full-scale converter-interfaced renewable energy power plants–Part I: Problem
statement, IEEE Trans. Power Del., vol. 30, no. 4, pp. 1770–1780, Aug. 2015.

156
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

[R75] A. Hooshyar, M. A. Azzouz, and E. F. El-Saadany, Distance protection of lines emanating

from full-scale converter-interfaced renewable energy power plants – Part II: Solution
description and evaluation, IEEE Trans. Power Del., vol. 30, no. 4, pp. 1781–1791, Aug.
2015.
[R76] System Protection Guidelines for Systems with High Levels of Renewables: Impact of Wind
& Solar Generation on Negative-Sequence and Power Swing Protection, EPRI, Palo Alto,
CA: 2017. 3002010937.
[R77] Impact of Inverter-Based Resources on Protection Schemes Based on Negative Sequence
Components, EPRI, Palo Alto, CA: 2019. 3002016197.
[R78] System Protection Guidelines for Systems with Inverter Based Resources: Performance of
Line Current Differential, Phase Comparison, Negative Sequence, Communication-Assisted,
and Frequency Protection Schemes Under Inverter-Based Resources and Impact of
German Grid Code. EPRI, Palo Alto, CA: 2019. 3002016196.
[R79] A. Haddadi, M. Zhao, I. Kocar, U. Karaagac, K. W. Chan and E. Farantatos, Impact of
Inverter-Based Resources on Negative Sequence Quantities-Based Protection Elements,
IEEE Trans. Power Del., 10.1109/TPWRD.2020.2978075, Mar. 2020.
[R80] M. Nagpal and C. Henville, Impact of power-electronic sources on transmission line ground
fault protection, IEEE Trans. Power Del., vol. 33, no. 1, pp. 62–70, Feb. 2018.
[R81] I. Erlich, T. Neumann, F. Shewarega, P. Schegner, and J. Meyer, Wind turbine negative
sequence current control and its effect on power system protection, IEEE Power Energy
Soc. Gen. Meeting, Jul. 2013.
[R82] Impact of Inverter-Based Resources on Power Swing and Rate of Change of Frequency
Protection. EPRI, Palo Alto, CA: 2020. 3002016198.
[R83] A. Haddadi, I. Kocar, U. Karaagac, H. Gras and E. Farantatos, Impact of wind generation on
power swing protection, IEEE Trans. Power Del., vol. 34, no. 3, pp. 1118-1128, Jun. 2019.
[R84] System Protection Guidelines for Systems with High Levels of Renewables: Impact of Wind
& Solar Generation on Negative-Sequence and Power Swing Protection. EPRI, Palo Alto,
CA: 2017. 3002010937. [8] Protection Guidelines for Systems with High Levels of Inverter
Based Resources. EPRI, Palo Alto, CA: 2018. 3002013635.
[R85] System Protection Guidelines for Systems with High Levels of Renewables: Impact of Wind
& Solar Generation on Negative-Sequence and Power Swing Protection, EPRI, Palo Alto,
CA: 2017. 3002010937.
[R86] Protection Guidelines for Systems with High Levels of Inverter Based Resources, EPRI, Palo
Alto, CA: 2018. 3002013635.
[R87] Protection system response to power swings, NERC System Protection and Control
Subcommittee Report, North Amer. Elect. Rel. Corp., Atlanta, GA, USA, Aug. 2013. Figure
14. Schematic diagram of the IEEE PSRC D6 test system
[R88] IEEE PSRC WG-D6, Power swing and out-of-step considerations on transmission lines,
IEEE Power System Relaying and Control Committee (PSRC) Working Group Rep. WG-D6,
Jul. 2005.
[R89] Tutorial on setting impedance-based power swing blocking and out-of-step tripping functions
on transmission lines, IEEE PSRCWG- D29 Draft Report, 2004.
[R90] N. Fischer, G. Benmouyal, S. Samineni, Tutorial on the impact of the synchronous generator
model on protection studies, the 35th Annual Western Protective Relay Conference,
Spokane, WA, Oct. 2008.
[R91] A. Haddadi, I. Kocar, T. Kauffmann, U. Karaagac, E. Farantatos and J. Mahseredjian, Field
Validation of Generic Wind Park Models using Fault Records, J. Mod. Power Syst. Clean
Energy, vol. 7, no. 4, pp. 826– 836, Jul. 2019.
[R92] P. Pourbeik, J. J. Sanchez-Gasca, J. Senthil, J. Weber, A. Ellis, S. Williams, S. Seman, K.
Bolton, N. Miller, R. J. Nelson, K. Nayebi, K. Clark, S. Tacke, and S. Lu, Value and
Limitations of the Positive Sequence Generic Models of Renewable Energy Systems, WECC
Modeling and Validation Working Group, Dec. 2015.

157
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

[R93] J. Mahseredjian, S. Dennetière, L. Dubé, B. Khodabakhchian and L. Gerin-Lajoie, On a new

approach for a simulation of transients in power systems, Elect. Power Syst. Res., vol. 77,
no. 11, pp. 1514-1520, Sep. 2007.
[R94] L. O. Chua and P.-M. Lin, Computer-Aided Analysis Of Electronic Circuits: Algorithms And
Computational Techniques, New Jersey: Prentice-hall, INC., 1975.
[R95] S. Henschel, Analysis of electromagnetic and electromechanical power system transients
with dynamic phasors, Ph.D. dissertation, University of British Colombia, 1999.
[R96] M. Stankovic, S. R. Sanders, and T. Aydin, Dynamic phasors in modeling and analysis of
unbalanced polyphase AC machines, IEEE Transactions on Energy Conversion, vol. 17, no.
1, pp. 107-113, 2002.
[R97] “Special reliability assessment interim report: Effects of geomagnetic disturbances on the
bulk power system, NERC, Feb. 2012.
[R98] J. Kappenman, Geomagnetic storms and their impacts on the U.S. power grid, Meta–R-319
Report, CA, U.S., 2010.
[R99] NERC Geomagnetic Disturbance Planning Guide, NERC, Dec. 2013
https://www.nerc.com/comm/PC/Geomagnetic Disturbance Task Force GMDTF 2013/GMD
Planning Guide_approved.pdf
[R100] A. Pulkkinen, S. Lindahl, A. Viljanen, and R. Pirjola, “Geomagnetic storm of 29–31 October
2003: Geomagnetically induced currents and their relation to problems in the Swedish high-
voltage power transmission system,” Space Weather, vol. 3, no. 8, Aug. 2005.
[R101] Geomagnetic Disturbance Planning Tools, NERC,
https://www.nerc.com/comm/PC/Pages/Geomagnetic-Disturbance-Planning-Tools.aspx
[R102] R. P. Jayasinghe, “Investigation of protection problems due to geomagnetically induced
currents,” Ph.D. dissertation, Dept. Elec. Eng., Univ. Manitoba, Winnipeg, MB, Canada,
1997.
[R103] “Reliability standard for geomagnetic disturbance operation,” FERC, Docket No. RM14-1-
000, Order No. 797, Jun. 2014.
[R104] NERC Std. TPL-007-3, “Transmission system planned performance for geomagnetic
disturbance events,” 2019.
[R105] R. Pirjola, “Properties of matrices included in the calculation of geo-magnetically induced
currents (GICs) in power systems and introduction of a test model for GIC computation
algorithms,” Earth Planet Sp., vol. 61, no. 2, pp. 263–272, Feb. 2009.
[R106] R. Horton, D. H. Boteler, T. J. Overbye, R. Pirjola, and R. C. Dugan, “A test case for the
calculation of geomagnetically induced currents,” IEEE Trans. Power Del., vol. 27, no. 4, pp.
2368–2373, Oct. 2012.
[R107] D. H. Boteler, “The use of linear superposition in modelling geomagnetically induced
currents,” IEEE Power Engineering Society General Meeting, 2013.
[R108] V. D. Albertson, J. G. Kappenman, N. Mohan, and G. A. Skarbakka, “Load-flow studies in
the presence of geomagnetically-induced currents,” IEEE Trans. Power App. & Sys., vol.
PAS-100, no. 2, pp. 594–607, Feb. 1981.
[R109] T. J. Overbye, T. R. Hutchins, K. Shetye, J. Weber, and S. Dahman, “Integration of
geomagnetic disturbance modelling into the power flow: A methodology for large-scale
system studies,” North American Power Symposium (NAPS), pp. 1–7, Sep. 2012.
[R110] Texas A&M University Electric Grid Datasets, http://electricgrids.engr.tamu.edu
[R111] Y. Zhang, K. S. Shetye, R. H. Lee, T. J. Overbye, “Impact of geomagnetic disturbances on
power system transient stability,” Proc. 2018 North American Power Symposium, Fargo, ND,
Sep. 2018.
[R112] T. R. Hutchins and T. J. Overbye, “Power system dynamic performance during the late-time
(E3) high-altitude electromagnetic pulse,” Proc. 19th Power Systems Computation
Conference (PSCC), Genoa, Italy, Jun. 2016.

158
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

[R113] T. R. Hutchins, “Modelling, Simulation, and Mitigation of the Impacts of the Late Time (E3)
High-Altitude Electromagnetic Pulse on Power Systems,” PhD Thesis, University of Illinois,
Champaign, IL, USA., 2016.
[R114] A. Haddadi, A. Rezaei-Zare, L. Gérin-Lajoie, R. Hassani, and J. Mahseredjian, “A modified
IEEE 118-bus test case for geomagnetic disturbance studies—Part I: Model data,” IEEE
Trans. Electromagn. Compat. (Early Access), doi: 10.1109/TEMC.2019.2920271, May 2019.
[R115] A. Haddadi, L. Gérin-Lajoie, A. Rezaei-Zare, R. Hassani, and J. Mahseredjian, “A modified
IEEE 118-bus test case for geomagnetic disturbance studies—Part II: Simulation results,”
IEEE Trans. Electromagn. Compat. (Early Access), doi: 10.1109/TEMC.2019.2920259, May
2019.
[R116] Technical Brochure 736 (2018) Power system test cases for EMT-type simulation studies,
WG C4.503, https://e-cigre.org/
[R117] X. Dong, Y. Liu, and J. G. Kappenman, “Comparative analysis of exciting current harmonics
and reactive power consumption from GIC saturated transformers,” in Power Engineering
Society Winter Meeting, 2001.
[R118] L. Bolduc, A. Gaudreau, and A. Dutil, “Saturation time of transformers under dc excitation,”
Electr. Power Syst. Res., vol. 56, pp. 95–102, 2000.
[R119] A. Haddadi, A. Rezaei-Zare, J. Mahseredjian, and R. F. Arritt, “Modelling of GMD related
reactive power losses: Three-phase transformer k-factor analysis,” EPRI, Palo Alto, CA,
3002014847, 2019.
[R120] J. Mahseredjian, S. Dennetière, L. Dubé, B. Khodabakhchian, and L. Gérin-Lajoie, “On a
new approach for the simulation of transients in power systems,” Elect. Power Syst. Res.,
vol. 77, no. 11, pp. 1514–1520, Sep. 2007.
[R121] I. Babaeiyazdi, M. Rezaei-Zare, and A. Rezaei-Zare, “Wind Farm Operating Conditions
Under Geomagnetic Disturbance”, IEEE Trans. Power Del., vol. 35, no. 3, pp. 1357 - 1364,
June 2020.
[R122] R. Gagnon, M. Fecteau, P. Prud’Homme, E. Lemieux, G. Turmel, D. Paré, F. Duong, "Hydro-
Québec Strategy to Evaluate Electrical Transients Following Wind Power Plant Integration
in the Gaspésie Transmission System," in IEEE Transactions on Sustainable Energy, vol. 3,
no. 4, pp. 880-889, Oct. 2012, doi: 10.1109/TSTE.2012.2200304.
[R123] V.Q. Do, J.C. Soumagne, G. Sybille, G. Turmel, P. Giroux, G. Cloutier, S. Poulin. "Hypersim,
an integrated real-time simulator for power network and control systems," Third International
Conference on Digital Power System Simulators (ICDS), Vasteras, Sweden, May 25-28,
1999.
[R124] R. Gagnon, J. Brochu, G. Turmel; "Variable Slip Wind Generator Modeling for Real-Time
Simulation"; CIGRE Canada - Conference on Power Systems, October 2006.
[R125] R. Gagnon, G. Turmel, C. Larose, J. Brochu, G. Sybille, M. Fecteau, “Large-Scale Real-
Time Simulation of Wind Power Plants into Hydro-Québec Power System,” 9th International
Workshop on Large-Scale Integration of Wind Power into Power Systems as well as on
Transmission Networks for Offshore Wind Power Plants. Québec Canada, October, 2010,
pp. 73-80.
[R126] N. Miller, W. Price, J. Sanchez-Gasca, Dynamic Modeling of GE 1.5 and 3.6 Wind Turbine-
Generators, October 27, 2003, GE-Power Systems Energy Consulting
[R127] B. Khodabakhchian, S. Breault, E. Portales, H. Huynh, Y. Hotte: “TRV and the non-zero
crossing phenomenon in Hydro-Quebec’s projected 735 kV series compensated system”,
CIGRE, Aug 30 – Sept 5, 1992, paper 13-303.
[R128] J. Mahseredjian, P. J. Lagacé et al.: “Superposition technique for MOV protected series
capacitors in short-circuit calculations”. IEEE Trans. On Power Delivery, July 1995, Vol. 10,
Issue 3, pages 1394-1400.
[R129] J. Mahseredjian, Test cases for the simulation of electromagnetic transients,
https://github.com/JeanMahseredjian/EMT_Benchmarks

159
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

[R130] L. Bolduc, “GIC observations and studies in the Hydro-Québec power system,” J. Atmos.
Sol.-Terr. Phys., 64 (11), pp. 1793–1802, Nov. 2002.
[R131] A. Rezaei-Zare, A. Haddadi, J. Mahseredjian, and R. F. Arritt, “Geomagnetic disturbance
transformer thermal analysis tool: EPRI transformer thermal model (ETTM), version 1.0–
Beta,” EPRI, Palo Alto, CA, 3002014059, 2018.
[R132] Y. Cheng, M. Sahni, J. Conto, S.-H. Huang, J. Schmall “Voltage-profile-based approach for
developing collection system aggregated models for wind generation resources for grid
voltage ride-through studies” IET Renewable Power Generation, 2011, Vol. 5, Iss. 5, pp.
332–346

160
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

APPENDIX A Dynamic Phasors

A.1 Dynamic Phasors
When the simple circuit of Figure A-1 is energized, it can be shown that its current response in time-
domain is given by Equation 12
R
Vm Vm − t
i(t) = sin(t +  − ) − sin( − )e L

(R ) (R )
1/2 1/2 Equation 12
2
+ 2L2 2
+ 2L2

Where Vm represents the amplitude of the sinusoidal voltage source vS = Vm sin(t + ) ,  is the
closing angle (source phase) and  is the source frequency. It is noticed that the left-hand side of this
equation corresponds to circuit’s steady-state and the right-hand side corresponds to its natural
response. The steady-state is the forced response and can be found directly using phasors. In this case
it becomes:
Vm  V 
I= = m Equation 13
R + jL Z

where Z is the impedance of the energized circuit and  is its phase. Upper case voltage and current
symbols are used. Here the current and voltage phasors are in peak quantities, but RMS values can be
used. Equation 12 is in time-domain and Equation 13 is in phasor-domain. In fact, Equation 13 can be
solved for changing Vm and  in time for finding changing (time-varying) the steady-states of the
circuit.
R
+ +
L +
+

Figure A-1 Basic RL circuit energization

The phasor solution can be found for any circuit using computerized analysis methods. Nodal analysis
or more recently, modified-augmented-nodal analysis, can be used to formulate network equations
based on Kirchoff’s laws. It is clear that in phasor-domain, the circuit elements are transformed into
complex impedances with j . It is also possible to use state-space representation for steady-state
solution, which is less common when solving large-scale systems.
As an illustrative example, the circuit of Figure A-2 is solved using the system of equations

 1 1 −1 −1 
 sL + sC1 + R R1
0
sL1
0
 1 1 
 −1 1 1 −1   V1   I 
 + 0 0    s 
 R1 R1 sL 2 sL 2   V2   0 
 −1 1  V  =  0  Equation 14
 0 + sC2 0 0  3   
 sL 2 sL 2   V4   0 
 −1  I   V 
1  v s   s 
1
 0 0
 sL 1 sL1 
 0 
 0 0 0 1

161
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

where the circuit node numbering is arbitrary, s = j , the unknowns are the node voltages and v s
I

(current entering voltage source), Vs and Is are phasors (casted into complex numbers) for the network
forced solution.

L1 R1 L2
4 + 1 + 2 + 3

Vs +

+
C1 C2

+
Is

Figure A-2 Simple Circuit

It is noticed here the circuit of Figure A-2 is a generic single wire circuit, but it can become three-phase
with coupling between phases or be used directly as the positive-sequence network of a grid. The
network remains linear.
Various perturbations can be applied to the circuit of Figure A-2 for finding consecutive steady-states.
The phasors of voltages and currents can vary in time. Switch (fault) conditions can be represented by
combining nodes (elimination of row and column in the matrix of Equation 14) or more elegantly through
an ideal switch equation added into Equation 14. If a switch is connected between any two nodes k and
m, then its closed state equation is given by:
Vk − Vm = 0 Equation 15

An extra row and column must be added (modified-augmented-nodal formulation [R93]) to Equation 14
to include Equation 15. When the switch is open, its equation becomes:
Ikm = 0 Equation 16

The driving source can be, for example, a synchronous generator with its differential equations solved
in time-domain and interfaced with the network in phasor-domain, using, for example a Norton
equivalent.

In the above procedure it is assumed that the network is in steady-state and solved using
s = j .
Harmonics are not considered. Slowly evolving phenomena is assumed. The phasor-domain solution is
predominantly used for simulating electromechanical transients.
To be able to solve the circuit of Figure A-2 in full time-domain it is necessary to solve not only its source
equations in time-domain, but also its circuit differential equations. For a generic branch, the ordinary
differential equation (ODE) is given by
dx
= f(x,t) Equation 17
dt

with x being the branch state-variable and f its function. Initial conditions can be found from the
steady-state solution discussed above:
x(0) = x0 . The basic approach is to find the solution at discrete
time-points as illustrated in Figure A-3.

162
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

with initial conditions

x

no initial conditions

0 1 t 2 t 3 t 4 t
t
Figure A-3 Time-domain solution time-points for an ODE solution
The time-step t is selected to account for the highest frequencies involved in the simulated transient,
typically it should be ten times smaller than the period of such transient. The time-step t is typically
fixed to achieve highest computational efficiency, but variable time-step methods also exist. The solution
method is called off-line when the computer time required to solve for t step is not necessarily equal
to t . In real-time simulation, the computing time must not exceed t .
The formulation of equations for Figure A-2 requires replacing the inductive and capacitive components
by the well-known companion circuit [R94] model. Different numerical integration methods can be used,
but the most predominant is trapezoidal integration. For Equation 17, trapezoidal integration allows to
create the discretization:
t t
xt = ft + x t −t + ft −t Equation 18
2 2
dikm
vkm = L
For an inductor connected between the nodes k and m, dt and its companion circuit is found
using trapezoidal integration:
t t t
ikmt = vkmt + vkmt −t + ikmt −t = vkmt + ihkmt Equation 19
2L 2L 2L
As shown in Figure A-4, Equation 19 results into a simple resistive circuit with a history (superscript h)
current. The history current is found from the previous time-point solution. Such a conversion can be
also applied to a capacitor. At a given time-point t, the solution of the circuit in Figure A-2 is formulated
by

 t 2C1 1 −1 t 
 2L + t + R R1
0 −
2L1
0
 1 1  i +i −i
 −1 1 t t   v1   s L1h C1h 
 + − 0 0    −iL2h
 R1 R1 2L 2 2L 2   v 2   
 t t 2C2   v 3  = iL h − iC h 
0    2
Equation 20
 0 − + 0 2

 2L 2 2L 2 t   v 4   −iL1h 

 t t  i   
 1  s  
v
− 0 0 vs 
 2L 1 2L1 
 
 0 0 0 1 0

where now lower-case symbols are used to denote time-domain quantities. The history terms for
inductors and capacitors are denoted using the subscript h. The right-hand side of this equation varies
in time. The voltage and current sources can be ideal or the system may include, for example, rotating
machines represented by Norton equivalents from accurate solution of machine differential equations.
Switch equations can be inserted as for Equation 14.

163
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

It is noted that Equation 20 can be written as a generic system of equations at a given time-point:
A t x t = bt Equation 21

where A t is the matrix of coefficients, x t is the vector of unknowns and b t is the vector of known
variables at a give time-point. Since the network may encounter topological changes in the genetic case,
the matrix A t is time-varying.

2L
t
+

k m
+
ikm
t ihkm
t

Figure A-4 Inductor companion circuit model for time-domains solution

Sparse matrix techniques are used to solve accurately and efficiently the system Equation 20. The
sparse matrices are essential when solving very large-scale power grids. Typical sparsity in a large
power grid is above 94%.
In a given grid, the transmission line or cable models can be modeled using delay propagation-based
models. Such models are derived from constant parameter or wideband line/cable models. In the
sample case of Figure A-5 the three subnetworks are separated by two transmission lines (TLM1 and
TLM2). The inherent propagation delays of these lines allow to formulate the matrix A t as shown in
Figure A-6. This block-diagonal form can be found automatically through matrix factorization or
topological analysis. In this way, the matrices can be solved in parallel without any loss of accuracy.
This basic principle is used for parallelizing off-line and real-time applications.

N1 N2 N3
TLM1 TLM2
+ +

Figure A-5 Three subnetworks separated by transmission lines

Figure A-6 Sparse matrix for circuit diagram of Figure A-5, the number of non-zeros is 60
The time-domain approach depicted above is very accurate and can effectively account for balanced or
unbalanced networks, it is solved using multiphase representation and nonlinear network models can
be included directly. There is no assumption on network conditions and both very fast and very slow
transients can be simulated. This is called circuit-based electromagnetic transient solution.

164
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

For comparison purposes, the graph of Figure A-7 presents a comparison between phasor-domain and
time-domain solutions. It is apparent here, for a simple example, that the phasor-domain approach is
unable to present the higher frequency behaviour of the circuit.

Figure A-7 Comparison of time-domain (EMT) in blue and phasor-domain (TS, voltage phasor magnitude)
in red, voltages (kV)
In between phasor-domain and time-domain, there is a method named dynamic-phasors (DP)
[R95][R96]. This method is not commonly used, but it is widely researched in the literature and has
potential to eventually develop the concept of accuracy navigator since it can, in theory, navigate
between the time-domain and phasor-domain methods. To quickly introduce the DP method, it is
recalled that
k = k =N
x (t) =  X k e jkst 
t
 X k e jkst
t
Equation 22
k =− k =−N

X
where kt is the complex Fourier coefficient of the kth harmonic, s is the fundamental frequency of
the grid and N is the approximated number of terms in the Fourier series. The differential of the kth term
is given by
dxk dXk
= + jks Xk Equation 23
dt dt
where the time variable is dropped for convenience. For the case of an inductor
k =N k =N k =N dIkmk
 Vkm e jk t = 
k
s jks LIkmk e jks t + L dt
e jks t Equation 24
k =−N k =−N k =−N

and
dIkmk
Vkmk = jks LIkmk + L Equation 25
dt
The discretization of this equation using trapezoidal integration, results into
1
Ikmk = Vkmk + Ihistk Equation 26
ZLk

with

165
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

2L
ZLk = + jksL Equation 27
t

ZL* k 1
Ihistk = Ikmt −t + Vkmt −t Equation 28
ZLk ZLk

Equation 26 resembles to the formulation achieved with Equation 19. It is apparent from Equation 27
that when t is small, the DP method approaches the time-domain formulation and when t is large,
the DP method approaches the phasor-domain solution. Although in theory, it is possible to solve DP
network equations for harmonics, it is typically solved only at fundamental frequency in linear grids for
maintaining computational efficiency. The main advantage is usage of larger t for accelerating
computations. The DP method is more accurate than the phasor-domain method, but it remains less
competitive when compared to time-domain methods capable of maintaining accuracy even with large
time-steps. The programming of the DP method is complicated, and the presence of complex numbers
increases computing times.

166
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

APPENDIX B EMT Studies for

Geomagnetic Disturbances
B.1 EMT Studies for Geomagnetic Disturbances
A Geomagnetic Disturbance (GMD) occurs whenever a large mass of solar energetic particles released
from the sun’s outer layer reaches the Earth’s proximity [R97]-[R99]. The interaction of this stream of
charged particle with the magnetosphere causes a slow temporal variation in the magnetic field of the
Earth. These variations induce a slowly varying potential, known as Geoelectric Field (GEF), along
conducting loops formed by the Earth’s surface and the transmission lines. This electric field causes an
induced voltage in the transmission lines which in turn results in low-frequency (0.1 Hz or lower)
Geomagnetically Induced Currents (GICs) to flow through the transmission lines and grounded
transformers to ground. The flow of these quasi-dc currents in the transformer winding biases the
operating point of the magnetization characteristics to one side, thus causing prolonged uni-directional
saturation of the transformer. This uni-directional saturation has the following main detrimental impacts
on the electric grid: (i) increased transformer excitation current and reactive power losses which can
lead to voltage regulation problems or even voltage instability [R99],[R100]; (ii) transformer hotspot
formation on winding and/or structural parts due to the extension of the magnetic flux beyond the
transformer core [R101]; and (iii) injection of current harmonics into the grid by the saturated transformer
which can in turn lead to mis-operation of protection and control devices and loss of reactive power
equipment [R102].
To assess the risks of a major GMD and evaluate potential mitigation strategies, North American Electric
Reliability Corporation (NERC) has developed planning and operating standard TPL-007 [R103][R104],
to ensure the safe operation of grids under GMDs. These standards and subsequent guidelines
recommend system studies which require the application of complex analysis techniques and tools. The
main challenge is that under a GMD, dc quantities (i.e., GICs and GEFs) and odd and even harmonics
are present in substantial levels and interact with fundamental-frequency voltage and current
components.
Utility planners use various analysis techniques and simulation tools for the assessment of GMD system
impacts. These techniques can be broadly categorized into load-flow-based (LF) [R105]-[R109], PDT
[R110]-[R113], and EMT [R114]-[R116]. These techniques are based on different modeling assumptions
and solution methods which need verification.
Specifically, the EMT method considers GMD-related nonlinear effects and solves the GICs, var losses,
system voltages, and harmonics at the same time. By contrast, the LF and PDT methods solve GICs
using a dc model assuming linear superposition. This assumption needs evaluation since in an actual
power system, GICs depend on both dc quantities and transformer ac voltage due to the nonlinear
magnetization of transformers. The LF and PDT methods do not consider the impact of ac voltage and
only represent the dc effects, which may lead to inaccuracies. Modeling the additional var losses in each
transformer due to half-cycle saturation is a key assumption that directly influences the system GMD
vulnerability assessment results. LF and PDT methods model the var losses using the “k-factor” method
[R117] which relates the total var loss of a transformer to the effective GIC flowing through it; however,
this method does not consider the impact of changing transformer terminal voltage due to harmonics
coupling, propagating, and entering other transformers which also impacts the var loss. This may
introduce inconsistencies in the calculated var loss and subsequent GMD vulnerability assessment
studies.
Finally, GMD voltage stability assessment is commonly conducted assuming the power system to be in
a complete steady state. This assumption ignores power system dynamic response to a GMD, such as
transformer time delays caused by nonlinear magnetization [R115],[R118] which may affect the
accuracy of GMD analysis results. PDT packages such as [R110]-[R113] artificially incorporate the
transformer time delay without representing the nonlinear magnetization. Given these effects, the
commonly used methods of power system analysis which are based exclusively on the fundamental-
frequency component may not be sufficient for GMD analysis, underscoring the need for the evaluation
of modeling assumptions and solution methods.
The objective of this section is to present an evaluation on existing methods for analysis of GMD system
impacts, identify the limitations of each method and inconsistencies, and provide assumptions on the
use of each method. Specifically, this section evaluates the assumption of linear superposition in the

167
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

context of GMD calculations, the k-factor method, and the impact of GMD-related system dynamics on
voltage stability results. To achieve the objectives, this section cross-examines the simulation results
of the LF, PDT, and EMT methods on the same basis.
The EMT method has been used as reference since it is not inherently limited by the GMD modeling
assumptions of the PDT and LF methods. This allows for evaluation of these assumptions and
identification of potential inconsistencies in GMD calculations due to modeling differences.

B.2 GMD Analysis Methods

Figure B-1 presents a graphical overview of the GMD vulnerability assessment process [R104]. The
starting point is the identification of the peak GEF, based on a provided benchmark GMD event. The
next steps include calculating GIC flows, var losses, thermal response of transformers, and power
system response including bus voltages. This response is then examined against a set of performance
criteria, and mitigation measures are taken if the response does not meet the criteria. The focus of this
report is on the three highlighted steps of namely Step 1 for GIC calculation, Step 2 for var calculation,
and Step 3 for voltage stability analysis.

Figure B-1 GMD vulnerability assessment process of TPL-007 [R104].

B.2.1 Load flow

Step 1 of the LF method consists of creating a dc model of the network and calculating GIC flows. This
modeling approach uses linear superposition which assumes the dc quantities (GICs and GEFs) to be
fully decoupled from ac quantities (var losses and bus voltages). This is incorrect in theory due to the
nonlinear behavior of transformer core which causes GICs to become dependent on both dc quantities
and ac voltage. In Step 2, the calculated GICs are fed to an electrical model of transformers which
calculates var losses. This commonly used k-factor approach calculates var losses based on
Urated
Qexe =kIeff Equation 29
Ubase

where Qexe stands for the total three-phase reactive power losses in Mvar, k is a constant for each core
type in Mvar/A, Ieff represents the effective GIC flowing through the transformer in A per-phase, Urated
is the transformer high-side voltage rating in kV, and Ubase (in kV) denotes the reference voltage used
in determining k factor. This analysis considers single-phase two-limb transformers for which the linear
k-factor method of Equation 29 is the common approach used by the LF and PDT methods. However,
the linear k-factor is not required by the LF/PDT methods. For other core types the GIC-var relationship
may not be linear [R99],[R119] , particularly the three-leg type, and a piecewise linear approximation is
used by some commercial LF/PDT codes. In Step 3, the calculated var losses are used to create an ac
representation of the network for load-flow studies including voltage stability.
B.2.2 PDT
The PDT method shares Step 1 and Step 2 of the LF method; however, it uses PDT simulations instead
of LF ones in Step 3 to calculate bus voltages. The TS simulation requires additional power system
modeling details including electrical and mechanical model of synchronous machines (SMs) and their
controls, which are provided by PDT tools.

168
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

B.2.3 EMT
In practice, GICs, var losses, harmonics and bus voltages are not decoupled from one another and can
dynamically interact. The EMT method [R120] accounts for this consideration by representing the full
network including dc quantities and nonlinear effects as a set of differential equations, thereby solving
the GICs, var losses, harmonics, bus voltages, and nonlinearities at the same time. In terms of solution
approach, the main difference with respect to the LF and PDT methods is that the EMT method carries
out Step 1-Step 3 at the same time. In terms of modeling assumptions, the main difference is the
incorporation of network nonlinearities including transformer iron core saturation characteristics which
is essential for the accurate representation of var losses. A piecewise linear function represents the
nonlinear saturation characteristic of each transformer. Such a representation obviates the need for the
k-factor method of var loss calculation, and the var loss is instead calculated as
Qexe =3 ph (t )Iph (t ) Equation 30

where  is the angular frequency in rad/s, ph (t ) signifies the amplitude of the fundamental-frequency
per-phase magnetic flux of the magnetization branch at time t in Wb, and Iph (t ) represents the
fundamental-frequency per-phase current of the magnetization branch at time t in A. Further than the
solution of nonlinear effects, the simultaneous solution of dc and ac quantities obviates the need for the
application of linear superposition. This feature of the EMT method enables evaluating the validity of the
modeling assumptions of the LF and PDT methods as studied in the test case presented in the following
section.

B.3 Test system

The test system used in this section is the IEEE 118-bus benchmark [R114]-[R116]. Hereinafter in this
report, the test system is referred to as 118-GMD. The model includes 173 power transformers with two-
and three-winding configuration modeled as single-phase two-limb units including a model of nonlinear
iron core. Load transformers include on-load-tap-changer (OLTC). The synchronous machine models
include mechanical and electrical dynamics, turbine-governor, control system with automatic voltage
regulator (AVR), power system stabilizer (PSS), and over-excitation limiter (OEL).
Although the 118-GMD test case does not contain inverter-based resources, the analysis presented
here is applicable to the presence such resources. In fact, the presence of power electronics devices,
further favors the usage of EMT simulations. Some results with a wind park case are also presented
below.
The GEFs have been modeled as lumped controllable dc voltage sources in series with the lines; the
dc voltage is calculated based on the GPS coordinates of line terminals and GEF amplitude and
orientation. Substation grounding has also been incorporated. References [R114]-[R116] present full
modeling data of 118-GMD.
To cross-examine LF, TS, and EMT GMD results, this report develops three implementations of 118-
GMD.
It should be mentioned that the transmission lines in 118-GMD have been represented by their PI model
in the LF, TS, and EMT methods to reduce modeling inconsistencies between the three methods. The
EMT method allows more accurate line models including frequency-dependent line models [R114].
B.3.1 Load-Flow Model Implementation
The LF model has been developed within the PDT and has two modules: an equivalent dc network and
an ac load-flow model. The dc network is modeled as an equivalent single-phase circuit whose topology
is consistent with that of the ac LF model. It contains equivalent dc models for each component in the
system, including dc resistance data, transformer configuration, substation GPS coordinates, and the
GEF model.
Transmission lines have been represented by the dc resistance of phase conductors. Transformers with
a grounded-wye winding have been modeled by the dc resistance of the grounded-wye winding.
Substation grounding has been represented by the resistance of the grounding grid in parallel to remote
earth resistance. Shunt capacitors are modeled as an open circuit while shunt reactors have been
represented by their resistive dc equivalent.
The ac LF model contains standard models as well as a model of GMD-related var losses. To obtain
these var losses, first the dc network is solved and the effective GIC flowing through the transformers is
calculated. This effective GIC is then translated into var loss using the k-factor approach assuming a

169
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

base value of k = 1.18 Mvar/A for single-phase two-limb transformers scaled to the high-side voltage of
the transformer. The calculated var loss is represented in the ac LF model as a shunt inductor in parallel
to the high-voltage-side of the transformer.
B.3.2 PDT Model Implementation
The PDT model has been developed has two modules, namely an equivalent dc network which is the
same as that of the LF model and a PDT ac model. The main difference with respect to the ac LF model
is the additional machine dynamic models and controls which allow simulating the ac system response
in time-domain. GICs and var losses are the same as those of the LF method.
B.3.3 EMT Model Implementation
The EMT model is presented in [R114]-[R116]. This model incorporates the nonlinear magnetization of
transformers and uses an iterative solver for solution of nonlinearities. Since this model is not limited by
the modeling assumptions of the LF and PDT implementations, it has been used in this report as
reference for cross-examination of GMD results.
B.3.4 Cross-Examination of Implementations (Without GMD)
To ensure that the three implementations are consistent, first their simulation results have been
compared without GMD. The comparisons of the load-flow solutions of the EMT and LF/PDT
implementations suggest a close match with the largest error in voltage amplitude and phase angle
being 0.02% and 0.01°, respectively.
The time-domain simulations and comparisons for slower electromechanical transients are also verified
to be consistent between EMT and PDT.

B.4 Comparison of GMD simulation results

The same GMD event has been simulated using the LF, PDT, and EMT models, and three sets of results
have been cross-examined, namely GICs, transformer var losses and voltage stability results. To
facilitate the cross-examination, the case studies assume a uniform GEF with a constant amplitude and
orientation over time. In practice, GEFs do not appear as a step function, but exhibit variations with
spectral characteristics in the same range as that of the saturation-delaying time constants of flux offset
[R97],[R99],[R104],[R115].
The test system has been subjected to a GMD with an electric field orientation of GEF =35 at 4 different
amplitudes of GEF = {1, 2, 3, 4} V/km. The orientation angle GEF follows the convention of [R97] (see
Figure B-2). In this reference frame, the x and y axes represent the geographical South–North and
West–East directions, Ex and Ey represent the electric field in the x and y directions, GEF (t ) and L
represent the angle of GEF vector and transmission line with respect to x axis, and the i and j symbols
signify the two terminal substations of the line. Reference [R114] has presented further details. The
effective GIC flowing through each transformer has been calculated based on the approach of [R109]
under each GEF level using the LF, PDT, and EMT methods. The results of the LF and PDT methods
are identical, and the results of the EMT method have been taken as reference.

Figure B-2 Definition of x and y coordinates, the direction of GEF vector, and the orientation of
transmission line.

B.5 Geomagnetically-Induced Currents

Figure B-3 shows the GIC solution of the EMT model including GIC flow in transmission lines and
transformer neutral under GEF=1 V/km. As shown, GICs exhibit an initial transient following the on-set
of GMD and reach a steady-state value after a time delay. The LF/PDT models do not capture this

170
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

transient state and only provide a steady-state solution. For comparison, the EMT simulation has been
run for a sufficiently long time until GICs reach a steady-state condition, and the steady-state values are
compared to the GICs calculated by the LF/PDT methods.
Cross-examination between LF/PDT and EMT solutions for the calculated GICs in transmission lines,
load substation transformer neutrals and three-winding grid transformer neutrals, indicate that in most
cases the percentage error is less than 10%. In a few cases, the percentage error is more than 10%.
This error is due to the omission of the impact of ac voltage on GICs by the LF and PDT methods. In an
actual power system, GICs depend on both dc quantities and transformer ac voltage due to the nonlinear
magnetization of transformers. The LF and PDTs do not consider the impact of ac voltage and only
represent the dc effects, thus leading to an error.
To further illustrate the cause of the error, a simulation case study has been carried out in which the
load transformer on bus 59 of 118-GMD has been subjected to a changing ac voltage under a fixed dc
voltage. The test system consists of the transformer connected to a variable ac voltage source
(emulating the impact of changing grid ac voltage) and series fixed dc voltage sources in each phase
(emulating a constant GEF). The dc source produces a dc current in the transformer windings (emulating
the GIC) while the ac voltage source provides ac excitation. Figure B-4 shows the response of the
transformer under a change in the ac voltage from Vac=1 pu to 0.92 pu suggesting that ac voltage level
impacts the level of dc current flowing into the transformer winding. This is due to the nonlinear
behaviour of the transformer magnetization; a change in the amplitude of ac voltage results in a change
in the peak-to-peak amplitude of flux in transformer core, which in turn leads to a change in the peak
value and the dc component of the magnetization current, and hence, the GIC.
Additional case studies have further cross-examined GIC solutions under GEF levels of {2, 3, 4} V/km.
The error tends to decrease with increasing GEF level. The reason is that the above-mentioned impact
of ac voltage on GIC tends to be more significant at lower GIC levels where the transformer is less
saturated, and a change of the transformer terminal voltage changes the generated GIC more
significantly.

B.6 Transformer VAR Loss

Transformer var loss has been calculated based on Equation 29 for the LF and PDTs and Equation 30
for the EMT method. Figure B-5 presents the transformer var loss under GEF=1 V/km obtained in EMT.
As shown, var losses exhibit an initial transient following the on-set of GMD and reach a steady-state
value after a time delay. This time delay for each individual transformer depends on factors such as GIC
level, delta winding resistance, and magnetization characteristics [R115],[R118]. The k-factor approach
of the LF/PDT models Equation 29 does not capture this transient state and only provides a steady-
state var solution.
The cross-examination of the calculated var loss for transformers, under GEF=1 V/km, generally exhibits
a more significant error compared to that of the GIC solution. The key reason for this error is the omission
of the impact of ac voltage on var loss under the k-factor approach of the LF/PDT methods. In practice,
the increased var loss of transformers reduces the ac voltage level which in turn results in the reduction
of var loss. The calculated var loss of the k-factor approach does not consider this interaction and hence
calculates larger var loss values than those of the EMT method. The var loss error found with the LF/PDT
method increases with increasing GEF level.

B.7 Voltage Stability Results

The LF, TS, and EMT methods use different methods and modeling assumptions to analyze voltage
stability due to GMD. The objective of this section is to compare their results. To that end, the same
GMD event is simulated using these methods, the amplitude of the applied GEF is changed until a
voltage collapse occurs, and this amplitude is compared between the three methods. Loads
corresponding to the normal operation of the system (without GMD) have been modeled by their
constant impedance representation in the LF, PDT, and EMT methods for consistency.
The LF method uses static analysis to determine voltage stability assuming a steady-state condition
during a GMD. The method calculates the additional var loss of each transformer using the k-factor
method assuming nominal voltage and models it as a load connected to the high-voltage side of the
transformer. These loads are modeled as constant-current to represent the excitation current during
transformer saturation and changes in the var loss as a function of ac voltage. The simulation test is
repeated at each GEF level by calculating a set of transformers var losses at the considered GEF level,
updating the constant-current loads, and running a load-flow simulation to determine voltage stability.
The criterion for declaring voltage instability in the LF method is the divergence of the load-flow solution.

171
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

The PDT method uses dynamic analysis based on time-domain simulations to assess voltage stability.
The system dynamics considered in the TS simulations include the electrical and mechanical dynamics
of the synchronous machines, machine controls including OEL, and the time delay of transformer var
loss increase. This delay has an impact on voltage stability analysis [R115] and hence should be
adequately represented in the TS simulation. As [R115], [R118] have shown, the delay of each individual
transformer depends on factors such as GIC level, delta winding resistance, and magnetization
characteristics. Reference [R118] has provided an approximate analytical formula which may be valid
under certain conditions. Reference [R113] has presented the numerical implementation of the delay in
a PDT; however, the delay is regarded as a user-defined parameter, and the more challenging problem
of calculating the delay is not addressed. Given these complexities, the TS simulation assumes a unique
delay for all transformers. This assumption creates a modeling difference with respect to the EMT model.
An objective of this section is to study the impact on voltage stability results. The delay has been
implemented by stepwise increasing of the constant-current load of each transformer. The PDT declares
voltage instability when the amplitude of the time-domain voltage of at least one bus drops to zero.
The EMT model inherently accounts for the time delay of var losses, and no additional modeling effort
is required.
Table B-1 compares the minimum GEF amplitude causing voltage collapse under the three methods,
Figure B-6 illustrates the simulated bus voltages in EMT under two GEF amplitudes of GEF=4V/km
(stable) and 5.5V/km (unstable).
The results of Table B-1 suggest an inconsistency in the calculated voltage stability limit. The error of
the LF method is mainly due to the assumption of steady-state which omits GMD-related system
dynamics influencing voltage stability limits including the time delay of var increase and machine
controls including OEL [R115]. Another cause of the inconsistency is the assumption of nominal voltage
in the calculation of the increased var loss.
The error of the PDT is mainly due to the modeling of the time delay of var loss increase. In , the delay
is assumed to be 0s for the PDT as the transformers var loss is increased instantaneously in a single
step. Table B-2 presents the results of other saturation time delays of {20, 40, 80, 140, 300}s realized
by stepwise increase of the individual var loss of each transformer in {1, 2, 4, 7, 15} steps with a step
duration of 20s. As shown, voltage collapse occurs at GEF={9, 12, 12, 14, 15}V/km, respectively. The
results show an inconsistency between TS and EMT solutions due to the different modeling of saturation
delays.

Figure B-3 GIC flow in line and transformer neutral under GEF=1 V/km, simulated in EMT

172
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure B-4 The dependence of GIC on ac voltage level: (a) instantaneous value of magnetic flux in
transformer core; (b) the dc component of magnetic flux; (c) instantaneous value of transformer
magnetization current; (d) the dc component of transformer magnetization.

Figure B-5 Transformer var loss under GEF=1 V/km simulated in EMT.
Table B-1 Comparison minimum GEF amplitudes causing voltage collapse calculated by the LF, TS, and
EMT methods
Method LF PDT EMT
Min. GEF causing voltage collapse (V/km) 7 9* 5.5

* var loss increased in one step.

Figure B-6 Simulation of bus voltages in EMT under: (a) GEF=4V/km (stable) and (b) GEF=5.5V/km
(unstable).
Table B-2 Minimum GEF amplitude causing voltage collapse calculated by the PDT under different
saturation time delay values.

173
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Saturation time delay (s) 0 20 60 120 280

GEF amplitude causing voltage collapse (V/km) 9 12 12 14 15

B.8 GMD Impacts on Wind Farms

With ever increasing penetration of IBRs, the characteristics and GMD vulnerability of the future power
grids will be different from those observed in the conventional power systems. Based on variety of
control strategies and flexibilities, the power converters of the IBRs are sensitive and quickly respond to
the abnormal GMD conditions and distorted voltage and current waveforms. [R121] investigates the
operating conditions of a typical wind farm during GMD, and thus the simulation results of this part is
represented from [R121]. The points of concerns include the impacts of GIC on the wind farm equipment
such as the main output transformer (MOT) and the doubly fed induction generator (DFIG), under
various GMD intensities. Also, it is crucial to known how the power electronic converters and the control
systems of the wind farm respond to the distorted voltage and current waveforms and reactive power
demand resulting from the GMD. Due to nonlinear nature of power system under the GMD conditions,
the EMT approach must be employed for accurate analysis of system behavior.
Figure B-7 shows a typical wind farm under study consisting of eighteen 2.5 MVA DFIG wind turbines
connected to a 230 kV power grid through the associated 50 MVA, 230 kV/34.5 kV main output
transformer (MOT) and a double-circuit transmission line. The details of the employed models and
parameters can be found in [R121]. In this study, the operating conditions of the wind farm is investigated
under i) various GIC levels ranging from 0 to 200 A at the neutral point of the MOT and ii) the benchmark
GIC signature of [R99].
Figure B-8 shows the voltage and current of phase A of the MOT when the magnitude of the GIC reaches
200 A at the neutral of the MOT HV winding. Under such conditions, the harmonic components
contribute to eddy current losses in the core and windings of the DFIGs and this should be considered
when the vulnerability of the DFIGs are assessed, during a GMD. Figure B-9 shows the total harmonic
distortion (THD) of the MOT and DFIGs for the full range of the simulated neutral GIC from 0 to 200 A.
While the acceptable voltage and current THD is 5% on the LV side (collector side), Figure B-9 depicts
the violations from this limit at less than 50 A GIC for DFIG voltage and current. It should be noted that
50 A neutral GIC is relatively small and during a severe GMD, higher magnitudes of GIC is likely.
For GMD vulnerability assessment, NERC has determined a benchmark GIC signature [R99] to be used
in the GMD vulnerability studies. Figure B-10-(a) shows this benchmark waveform when the peak of the
waveform is adjusted at 100 A per phase. Such a GIC is applied on the studied system and the MOT
winding temperature rise is estimated as an indication of the MOT withstand capability to determine
whether or not the wind farm can stay in service under the benchmark GIC conditions. Figure B-10-(b)
shows the MOT winding temperature rises for the peak benchmark GIC of 70A, 75A, and 100A. The
figure clearly shows that the MOT winding temperature exceeds the maximum permissible standard
limit of 180 degrees and the MOT is prone to failure.

Double Circuit
Bus 1 Bus 2 Transmission Bus 3
MOT Line

18-Turbine Grid
wind farm

Filter

Figure B-7 Wind farm under study with 18 DFIG turbines

Figure B-8 MOT operating conditions under GIC of 200 A at neutral a) voltage, b) current

174
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure B-9 Wind farm THDs versus GIC at full load

Figure B-10 MOT hotspot temperature during the benchmark GIC event, a) benchmark GIC signature, b)
MOT winding hotspot temperature under the GIC benchmark with various magnitudes.

B.9 Conclusion
This report has studied the LF, PDT, and EMT methods of GMD system impact analysis, evaluated their
modelling assumptions, and identified potential inconsistencies in GMD calculations due to modelling
differences. Since the EMT method is not inherently limited by the modelling assumptions of the LF and
PDTs, it has been used in this report as the reference for cross-examination.
GIC calculations showed an inconsistency which was negligible in most cases and more significant at
lower GIC levels where the transformer is less saturated. The cause was the different modelling of
nonlinear magnetization and the resulting impact of ac voltage on GIC. The LF and PDT do not consider
this impact and solve GICs using a dc model based on the assumption of linear superposition, potentially
leading to inaccuracies. At higher GIC levels, the impact of ac voltage and hence the inaccuracy became
negligible. The results suggest that the assumption of linear superposition in a power system under
GMD should be further studied.
Var loss calculations of the LF and PDT also showed an inconsistency with respect to the EMT method.
The cause was the omission of the impact of ac voltage on var loss under the k-factor approach used
by the LF and PDT. In a real-world power system, the increased var loss of transformers reduces the
ac voltage level which in turn results in the reduction of var loss, and therefore it is necessary to consider
the interaction of ac voltage and var loss. Given the direct influence of var loss modelling on the system
GMD vulnerability assessment results, such an error may cause further inaccuracies.
Voltage stability analysis results of the three methods also showed an inconsistency. In case of the LF
method, the cause was the assumption of steady state which neglects the impact of power system
dynamics on voltage stability. These omitted dynamics include the transformer time delay. Another
contributing factor was the inaccuracy of calculated var loss due to the omission of the impact of ac
voltage. In the case of TS, the cause was the different modelling of transformers saturation time delays.
The report has shown the sensitivity of the calculated voltage stability limit to this time delay, thereby
highlighting the importance of incorporating the time delay in voltage stability analysis.

B.10 Simulation Results

Table B-3 Cross-examination of load-flow solution of the EMT and LF/PDT implementations (without
GMD)
EMT PDT Error
Bus Voltage Angle Voltage Angle Voltage Error Angle Voltage Angle
Number (pu) (°) (pu) (°) (pu) Error (°) Error % Error %
1 0.97 5.62 0.97 5.62 0 0 0.01 0.07
2 0.97 6.28 0.97 6.27 0 0 0.01 0.06

175
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

3 0.97 6.81 0.97 6.8 0 0 0.01 0.05

4 0.99 10.98 0.99 10.97 0 0 0.01 0.02
5 0.98 11.87 0.98 11.87 0 0 0.01 0.02
6 0.97 8.89 0.97 8.89 0 0 0.01 0.04
7 0.97 8.43 0.97 8.43 0 0 0.01 0.04
8 0.98 16.32 0.98 16.33 0 0 0.01 -0.02
9 0.99 29.06 0.99 29.06 0 -0.01 0.01 -0.02
10 1.04 44.05 1.04 44.06 0 -0.01 0.01 -0.02
11 0.97 8.2 0.97 8.2 0 0 0.01 0.04
12 0.98 7.7 0.98 7.69 0 0 0.01 0.05
13 0.98 6.26 0.98 6.26 0 0 0.01 0.06
14 0.97 6.81 0.97 6.8 0 0 0.01 0.06
15 0.97 6.51 0.97 6.51 0 0 0.01 0.07
16 0.99 8.86 0.99 8.86 0 0 0.01 0.04
17 0.99 9.74 0.99 9.74 0 0 0.01 0.03
18 0.97 7.48 0.97 7.48 0 0 0.02 0.04
19 0.96 6.75 0.96 6.75 0 0 0.01 0.06
20 0.97 7.11 0.97 7.11 0 0 0.01 0.06
21 0.97 8.36 0.97 8.36 0 0 0.01 0.05
22 0.98 10.69 0.98 10.69 0 0 0.01 0.03
23 1 16.19 1 16.19 0 0 0.01 0.02
24 1 15.91 1 15.91 0 0 0.01 0.02
25 1.04 26.06 1.04 26.06 0 0 0.01 0
26 1.04 26.56 1.04 26.56 0 0 0.01 -0.02
27 0.97 5.18 0.97 5.17 0 0 0.01 0.09
28 0.97 4.08 0.97 4.07 0 0 0 0.12
29 0.97 3.87 0.97 3.87 0 0 0.01 0.12
30 0.98 13.37 0.98 13.38 0 0 0.01 -0.02
31 0.96 4.34 0.96 4.34 0 0 0.01 0.11
32 0.96 4.51 0.96 4.51 0 0 0.01 0.1
33 0.98 7.38 0.98 7.37 0 0 0.01 0.06
34 0.99 9.49 0.99 9.48 0 0 0.01 0.05
35 0.99 9.13 0.99 9.13 0 0 0.01 0.05
36 0.99 9.09 0.99 9.08 0 0 0.01 0.05
37 1 10.33 1 10.32 0 0 0.01 0.04
38 0.97 13.21 0.97 13.21 0 0 0 -0.01
39 0.99 8.68 0.99 8.68 0 0 0.01 0.05
40 0.98 9.02 0.98 9.02 0 0 0.01 0.04
41 0.96 8.55 0.96 8.54 0 0 0 0.04
42 0.97 11.47 0.97 11.47 0 0 0.01 0.03
43 0.98 9.56 0.98 9.55 0 0 0.01 0.04
44 0.96 12.6 0.96 12.6 0 0 0.01 0.02
45 0.96 14.67 0.96 14.67 0 0 0 0.02
46 0.97 17.79 0.97 17.78 0 0 0.01 0.01
47 0.99 20.11 0.99 20.11 0 0 0 0.01
48 0.98 19.73 0.98 19.73 0 0 0.01 0.01
49 0.99 20.84 0.99 20.84 0 0 0.01 0

176
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

50 0.98 18.49 0.98 18.49 0 0 0 0.01

51 0.97 16.62 0.97 16.61 0 0 0.01 0.01
52 0.97 15.81 0.97 15.81 0 0 0.01 0.02
53 0.97 15.46 0.97 15.46 0 0 0.01 0.02
54 0.98 17.04 0.98 17.04 0 0 0 0.02
55 0.97 16.24 0.97 16.24 0 0 0 0.02
56 0.97 16.11 0.97 16.11 0 0 0 0.02
57 0.98 16.37 0.98 16.36 0 0 0 0.02
58 0.97 16.1 0.97 16.1 0 0 0.01 0.02
59 1 20.77 1 20.76 0 0 0.01 0.02
60 1.01 23.88 1.01 23.87 0 0 0.01 0.01
61 1.02 26.05 1.02 26.05 0 0 0.01 0.01
62 1.01 24.22 1.01 24.22 0 0 0.01 0.01
63 0.99 24.24 0.99 24.24 0 0 0 0
64 1.01 26.18 1.01 26.18 0 0 0 0
65 1.03 29.19 1.03 29.19 0 0 0.01 0.01
66 1.01 28.67 1.01 28.67 0 0 0 0
67 1 25.7 1 25.7 0 0 0 0
68 1 29.64 1 29.64 0 0 0.01 0.01
69 1.05 27.48 1.05 27.48 0 0 0 0
70 0.98 15.55 0.98 15.55 0 0 0.01 0.02
71 0.99 15.57 0.99 15.57 0 0 0.01 0.02
72 1 15.68 1 15.67 0 0 0.01 0.02
73 0.99 15.57 0.99 15.57 0 0 0.01 0.02
74 0.97 14.94 0.97 14.94 0 0 0.01 0.02
75 0.98 16.5 0.98 16.5 0 0 0.01 0.02
76 0.99 16.43 0.99 16.43 0 0 0 0.02
77 1 24.17 1 24.17 0 0 0 0.02
78 0.98 24.65 0.98 24.65 0 0.01 0.01 0.02
79 0.98 26.05 0.98 26.04 0 0.01 0.01 0.02
80 1.01 30.89 1.01 30.88 0 0.01 0.01 0.02
81 0.98 30.25 0.98 30.25 0 0 0 0
82 0.97 32.15 0.97 32.15 0 0 0 0.01
83 0.97 36.39 0.97 36.39 0 0 0 0.01
84 0.97 38.7 0.97 38.7 0 0 0 0.01
85 0.98 41.59 0.98 41.58 0 0 0 0.01
86 0.98 41.29 0.98 41.28 0 0 0 0.01
87 0.98 42.54 0.98 42.53 0 0 0 0.01
88 1.01 48.13 1.01 48.12 0 0 0.01 0.01
89 1.03 49.88 1.03 49.88 0 0 0.01 0.01
90 1.01 48.43 1.01 48.43 0 0 0.01 0.01
91 1.01 47.36 1.01 47.35 0 0 0.01 0.01
92 1.01 45.57 1.01 45.57 0 0 0.01 0.01
93 0.99 41.44 0.99 41.43 0 0 0 0.01
94 0.99 38.36 0.99 38.36 0 0 0 0.01
95 0.98 35.81 0.98 35.8 0 0 0 0.01
96 0.98 33.97 0.98 33.96 0 0 0 0.01

177
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

97 0.99 32.04 0.99 32.04 0 0.01 0.01 0.02

98 1 32.46 1 32.46 0 0.01 0.01 0.02
99 1 36.37 1 36.36 0 0 0.01 0.01
100 0.99 38.88 0.99 38.87 0 0 0.01 0.01
101 0.98 39.37 0.98 39.37 0 0 0.01 0.01
102 0.99 41.83 0.99 41.82 0 0 0.01 0.01
103 0.98 38.29 0.98 38.29 0 0 0.01 0.01
104 0.97 36.59 0.97 36.59 0 0 0.01 0.01
105 0.97 36.7 0.97 36.69 0 0 0.01 0.01
106 0.98 37.54 0.98 37.53 0 0 0 0.01
107 0.98 36.62 0.98 36.61 0 0 0 0.01
108 0.97 36.46 0.97 36.46 0 0 0 0.01
109 0.97 36.39 0.97 36.38 0 0 0 0.01
110 0.97 36.68 0.97 36.67 0 0 0.01 0.01
111 0.99 40.09 0.99 40.08 0 0 0.01 0.01
112 0.96 34.96 0.96 34.96 0 0.01 0.01 0.01
113 0.99 7.64 0.99 7.64 0 0 0.01 0.05
114 0.96 4.15 0.96 4.14 0 0 0 0.12
115 0.96 4.05 0.96 4.04 0 0 0 0.12
116 1 30.59 1 30.59 0 0 0.01 0.01
117 0.98 7.62 0.98 7.62 0 0 0.01 0.05
118 0.98 15.82 0.98 15.82 0 0 0.01 0.02

Figure B-11 Response of the SM on bus Breed to a three-phase 100-ms bolted fault on bus 4 of 118-GMD
within EMT and PDT: (a) machine speed; (b) electric power output; (c) turbine mechanical power; (d) field
voltage; (e) PSS control signal output; and (f) OEL control

178
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure B-12 GIC flow in line and transformer neutral under GEF=1 V/km and θGEF=35° simulated in EMT

Table B-4 Cross-examination of GIC in the neutral of load substation transformers under GEF=1 V/km and
θGEF=35° calculated by the LF/PDT and EMT methods
Line GIC Erro Line GIC Erro Line Line GIC Erro
(A) r (A) r (A) r
Fro To LF/PD EMT (%) Fro To LF/PD EMT (%) Fro To LF/PD EMT (%)
m T m T m T
bus bu bus bu bus bu
s s s
1 2 -1.21 - 0.0 39 40 9.48 9.48 0 75 11 1.77 1.77 0.0
1.21 3 8 4
1 3 -6.42 - 0.0 40 41 -3.12 - 0.0 76 77 -1.29 -1.3 0.2
6.42 3 3.12 2 8
2 12 -7.91 - 0.0 40 42 -5.23 - 0.0 76 11 -3.73 - 0.0
7.91 2 5.23 2 8 3.73 3
3 5 -7.35 - 0.0 41 42 -2.06 - 0.1 77 78 4.03 4.03 0.0
7.36 5 2.06 2
3 12 -2.74 - 0.0 42 49 -6.44 - 0.0 77 80 -0.98 - 0.2
2.74 2 6.44 4 0.99
4 5 -3.21 - 0.1 42 49 -6.44 - 0.0 77 80 -0.98 - 0.2
3.21 6.44 4 0.99
4 11 3.52 3.52 0.0 43 44 3.05 3.05 0.0 77 82 -13.17 - 0.0
9 2 13.1 1
7
5 6 -0.1 -0.1 4.6 44 45 -1.04 - 0.1 78 79 -2.44 - 0.1
9 1.04 6 2.44
5 11 4.14 4.14 0.0 45 46 -3.86 - 0.0 79 80 -10.22 - 0.0
9 3.87 5 10.2 6
3
6 7 2.87 2.87 0.0 45 49 4.64 4.64 0.0 80 96 -13.4 - 0
6 6 13.4
7 12 4.02 4.02 0.0 46 47 3.48 3.48 0.0 80 97 -12.47 - 0.0
2 5 12.4 1
7
8 9 -18.94 - 0.0 46 48 6.22 6.22 0.0 80 98 -10.34 - 0
18.9 3 2 10.3
4 4
8 30 -0.34 - 0.1 47 49 15.16 15.1 0.0 80 99 -12.59 - 0.0
0.34 6 6 1 12.5 1
9
9 10 -18.94 - 0.0 47 69 -11.46 - 0.0 82 83 -16.94 - 0.0
18.9 2 11.4 2 16.9 1
4 7 4

179
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

11 12 0.71 0.71 0.1 48 49 4.29 4.29 0.0 82 96 -3.94 - 0.0

2 5 3.95 3
11 13 3.43 3.44 0.0 49 50 12.88 12.8 0.0 83 84 -11.29 - 0.0
4 9 2 11.2 1
9
12 14 0.57 0.57 0.1 49 51 15.58 15.5 0.0 83 85 -10.5 - 0.0
2 8 2 10.5 1
12 16 -4.35 - 0.0 49 54 12.26 12.2 0.0 84 85 -13.66 - 0.0
4.36 1 6 1 13.6 1
7
12 11 -0.74 - 0.1 49 54 12.26 12.2 0.0 85 86 -10.79 - 0.0
7 0.74 4 6 1 10.7 1
8
13 15 -1.81 - 0.0 49 66 -7.44 - 0.0 85 88 -7.07 - 0.0
1.81 1 7.44 8 7.08 4
14 15 -0.96 - 0 49 66 -7.44 - 0.0 85 89 -5.4 - 0.0
0.96 7.44 8 5.41 5
15 17 -14.24 - 0.0 49 69 -12.84 - 0.0 86 87 -7.79 - 0.0
14.2 3 12.8 1 7.79 1
4 4
15 19 -0.4 -0.4 0.0 50 57 13.93 13.9 0.0 88 89 0.11 0.11 3.7
1 4 1 6
15 33 5.02 5.02 0.0 51 52 2.25 2.25 0.0 89 90 -1.74 - 0.1
2 5 1.74 3
16 17 -2.69 - 0.0 51 58 12.94 12.9 0 89 90 -1.74 - 0.1
2.69 8 4 1.74 3
17 18 9.6 9.61 0.0 52 53 7.99 7.99 0 89 92 5.31 5.31 0.0
5 7
17 31 -14.14 - 0.0 53 54 5.84 5.84 0.0 89 92 3.57 3.57 0.0
14.1 2 2 7
4
17 11 -9.09 - 0.0 54 55 -3.23 - 0.0 90 91 2.49 2.49 0.0
3 9.09 2 3.23 3 4
18 19 5.64 5.64 0.0 54 56 -6.29 - 0.0 91 92 4.69 4.69 0.0
1 6.29 1 2
19 20 -10.65 - 0 54 59 -8.25 - 0.0 92 93 10.32 10.3 0.0
10.6 8.25 1 3 2
5
19 34 6.39 6.39 0.0 55 56 -9.45 - 0 92 94 10.84 10.8 0.0
4 9.45 4 2
20 21 -10.22 - 0.0 55 59 -12.12 - 0.0 92 10 5.58 5.58 0.0
10.2 1 12.1 2 0 1
3 3
21 22 -9.1 -9.1 0.0 56 57 -10.83 - 0.0 92 10 -2.37 - 0.0
2 10.8 1 2 2.37 5
3
22 23 -10.34 - 0.0 56 58 -9.37 - 0 93 94 11.31 11.3 0.0
10.3 2 9.37 1 1
4
23 24 1.09 1.09 0.0 56 59 -4.36 - 0.0 94 95 13.15 13.1 0.0
2 4.36 3 6 2
23 25 -9.12 - 0.0 56 59 -4.57 - 0.0 94 96 15.67 15.6 0.0
9.12 6 4.57 3 8 1
23 32 3.9 3.91 0.0 59 60 -9.71 - 0.0 94 10 -5.98 - 0.0
8 9.71 3 0 5.99 5
24 70 0.93 0.93 0.0 59 61 -8.77 - 0.0 95 96 14.44 14.4 0
2 8.77 4 4
24 72 2.33 2.33 0 60 61 2.59 2.58 0.1 96 97 14.29 14.2 0
8 9
25 27 4.53 4.54 0.1 60 62 -8.29 - 0.0 98 10 -15.46 - 0.0
1 8.29 1 0 15.4 1
6
26 30 10.25 10.2 0.0 61 62 -6.32 - 0.0 99 10 -14.47 - 0
5 3 6.32 2 0 14.4
7

180
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

27 28 4.95 4.95 0.0 62 66 -4.83 - 0.0 100 10 -17.99 - 0

3 4.84 2 1 17.9
9
27 32 4.95 4.95 0.0 62 67 -2.26 - 0.0 100 10 -7.87 - 0.0
1 2.26 1 3 7.87 1
27 11 3.58 3.58 0.0 63 64 -7.13 - 0.0 100 10 0.39 0.39 0.2
5 1 7.13 4 4 2
28 29 8.13 8.13 0 64 65 -9.43 - 0.0 100 10 -1.84 - 0.2
9.43 2 6 1.84 2
29 31 8.61 8.61 0.0 65 68 -22.11 - 0.0 101 10 -9.05 - 0.0
1 22.1 1 2 9.05 1
1
30 38 6.65 6.65 0 66 67 7.73 7.73 0.0 103 10 4.88 4.88 0.0
2 4 1
31 32 -4.53 - 0.0 68 81 -12.33 - 0.0 103 10 2.97 2.97 0.0
4.53 1 12.3 1 5 3
3
32 11 7 7 0.0 68 11 -3.92 - 0.0 103 11 -6.08 - 0.0
3 2 6 3.92 7 0 6.08 1
32 11 -4.91 - 0.0 69 70 -9.26 - 0.0 104 10 -3.62 - 0
4 4.91 3 9.26 4 5 3.62
33 37 2.88 2.88 0.0 69 75 -12.55 - 0.0 105 10 3.06 3.06 0.0
8 12.5 3 6 8
5
34 36 -1.28 - 0.1 69 77 -15 -15 0.0 105 10 2.06 2.06 0.0
1.28 2 1 7 4
34 37 10.12 10.1 0.0 70 71 3.23 3.23 0.0 105 10 -5.31 - 0
1 8 1 8 5.31
34 43 1.32 1.32 0.0 70 74 -1.97 - 0.0 106 10 0.09 0.09 1.2
4 1.97 4 7 8
35 36 -1.76 - 0.0 70 75 -0.92 - 0.0 108 10 -8.44 - 0
1.76 2 0.92 2 9 8.44
35 37 2.89 2.89 0.0 71 72 -1.28 - 0.0 109 11 -9.85 - 0
9 1.28 1 0 9.85
37 39 11.3 11.3 0.0 71 73 4.52 4.52 0 110 11 -4.51 - 0.0
1 2 1 4.51 5
37 40 12.25 12.2 0.0 74 75 3.02 3.02 0.1 110 11 -0.42 - 0.2
5 1 1 2 0.42
38 65 1.73 1.73 0.3 75 77 0.81 0.81 0.3 114 11 -4.25 - 0.0
3 7 5 4.25 2

Table B-5 Cross-examination of GIC in the neutral of load substation transformers under GEF=1 V/km and
θGEF=35° calculated by the LF/PDT and EMT methods
Xfo GIC (A) Error Xfo GIC (A) Error Xfo GIC (A) Error
bus LF/PD EMT (%) bus LF/PD EMT (%) bus LF/PD EMT (%)
T T T
1 11.92 11.92 0 43 -5.19 -5.19 0.03 80 11.54 11.54 0.05
2 20.12 20.12 0.03 44 12.28 12.28 0 82 23.15 23.15 0
3 11.01 11.01 0.01 45 -5.46 -5.46 0.04 83 14.55 14.55 0.03
4 -0.21 -0.18 15.73 46 -32.06 -32.06 0.02 84 7.12 7.12 0
6 -4.64 -4.63 0.11 47 -0.66 -0.65 0.21 85 -0.61 -0.6 0.95
7 -3.45 -3.45 0.02 48 5.78 5.78 0.03 86 -8.97 -8.97 0.01
11 10.53 10.53 0.01 49 -19.13 -19.13 0.02 88 -21.56 -21.57 0.05
12 -2.8 -2.78 0.81 50 -3.16 -3.16 0.01 90 -9.34 -9.33 0.01
13 15.72 15.72 0 51 1.16 1.16 0.22 92 -16.88 -16.88 0.02
14 4.59 4.59 0 52 -17.22 -17.22 0.03 93 -2.95 -2.95 0.03
15 10.72 10.72 0 53 6.46 6.46 0.02 94 -2.09 -2.09 0.02
16 -4.99 -4.99 0.03 54 49.62 49.62 0 95 -3.87 -3.87 0.08
17 -0.28 -0.16 80.67 54 49.62 49.62 0 96 -4.55 -4.55 0.09
18 6.2 6.19 0.07 55 28.66 28.66 0 97 5.47 5.47 0.01

181
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

19 14.83 14.83 0.01 56 36.55 36.55 0 98 15.34 15.34 0.01

20 -1.27 -1.27 0.06 57 9.3 9.3 0.01 100 -2.27 -2.25 1.15
21 -3.38 -3.38 0.01 58 10.72 10.71 0.01 101 -26.81 -26.83 0.05
22 3.73 3.73 0.03 59 -3.57 -3.44 3.77 102 -34.26 -34.28 0.05
23 -18.65 -18.65 0.03 59 -3.57 -3.44 3.77 103 -9.75 -9.75 0.03
27 -4.88 -4.89 0.24 59 -3.57 -3.44 3.77 104 13.88 13.88 0
28 -9.53 -9.53 0.05 60 -11.99 -11.99 0 105 -0.31 -0.31 0.12
29 -1.45 -1.45 0.07 62 -11.75 -11.75 0.01 106 3.39 3.39 0.06
31 -2.94 -2.94 0.04 66 -8.33 -8.32 0.06 107 5.27 5.27 0.03
32 3.48 3.47 0.33 67 16.39 16.39 0.01 108 9.39 9.39 0.03
33 6.41 6.41 0.02 70 -13.54 -13.54 0 109 4.22 4.22 0.03
34 -5.88 -5.88 0.08 74 -7.8 -7.81 0.02 110 -17.18 -17.18 0
35 -3.4 -3.4 0 75 -39.1 -39.09 0.03 112 -1.13 -1.13 0.24
36 -2.05 -2.05 0.02 76 7.85 7.85 0.01 114 -1.98 -1.98 0.02
39 5.48 5.48 0.02 77 -6.83 -6.83 0.05 115 -2.01 -2.01 0.02
40 20.28 20.28 0.01 78 19.39 19.4 0.01 117 -2.21 -2.21 0.02
41 -3.18 -3.18 0.02 79 23.35 23.35 0 118 -5.88 -5.88 0.03
42 14.11 14.11 0 80 11.54 11.54 0.05

Table B-6 Cross-examination of GIC in the neutral of generator substation transformers under GEF=1
V/km and θGEF=35° calculated by the LF/PDT and EMT methods
Xfo GIC (A) Error Xfo GIC (A) Error Xfo GIC (A) Error
bus LF/PD EMT (%) bus LF/PD EMT (%) bus LF/PD EMT (%)
T T T
1 10.97 10.96 0.01 42 2.65 2.64 0.02 77 -6.28 -6.28 0.01
4 -0.73 -0.75 3.78 46 -4.31 -4.31 0.01 80 17.31 17.31 0.04
6 -4.27 -4.27 0.15 46 -4.31 -4.31 0.01 80 17.31 17.31 0.04
8 13.96 13.95 0 49 -22.96 -22.96 0.02 85 -2.1 -2.09 0.55
10 -28.41 -28.41 0.01 54 15.04 15.03 0.01 87 -23.38 -23.39 0.04
10 -28.41 -28.41 0.01 54 15.04 15.03 0.01 89 -10.69 -10.68 0.07
12 -0.35 -0.36 1.74 54 15.04 15.03 0.01 89 -10.69 -10.68 0.07
12 -0.35 -0.36 1.74 55 26.36 26.36 0 89 -10.69 -10.68 0.07
12 -0.35 -0.36 1.74 56 3.65 3.65 0.01 90 -8.59 -8.59 0.01
12 -0.35 -0.36 1.74 59 -3.28 -3.27 0.41 91 -6.58 -6.59 0.05
15 9.86 9.86 0.01 59 -2.38 -2.41 1.4 92 -15.53 -15.53 0.02
18 5.7 5.7 0.04 61 0.13 0.29 53.16 99 5.63 5.63 0.04
19 13.64 13.64 0 62 -10.81 -10.81 0 100 -1.82 -1.82 0.38
24 -6.51 -6.51 0.01 65 19.48 19.47 0.02 100 -4.96 -4.95 0.15
25 -7.28 -7.28 0.02 65 19.48 19.47 0.02 103 -19.18 -19.19 0.06
26 -11.13 -11.13 0.03 66 -18.17 -18.17 0.02 104 12.77 12.77 0
26 -11.13 -11.13 0.03 66 -18.17 -18.17 0.02 105 -1.06 -1.04 1.78
27 -21.95 -21.94 0.05 69 5.06 4.93 2.64 107 1.19 1.19 0.01
31 -0.07 -0.07 1.97 70 -12.46 -12.46 0.01 110 -15.81 -15.81 0
32 3.2 3.21 0.17 72 3.12 3.12 0.05 111 -13.54 -13.54 0.02
34 -5.41 -5.41 0.05 73 13.56 13.55 0.01 112 -0.13 -0.13 0.23
36 -7.08 -7.08 0.02 74 -7.18 -7.18 0.03 113 -6.28 -6.28 0.07
40 69.95 69.96 0.01 76 7.23 7.22 0.01 116 -11.76 -11.77 0.05

182
TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Table B-7 Cross-examination of effective GIC in the neutral of three-winding grid transformers under
GEF=1 V/km and θGEF=35° calculated by the LF/PDT and EMT methods.
Transformer GIC (A) Error (%)
From bus To bus LF/PDT EMT
5 8 0.29 0.29 0.7
17 30 0.16 0.03 500.76
25 26 -42.16 -42.16 0.01
37 38 -8.18 -8.18 0.08
59 63 5.26 4.9 7.4
61 64 7.19 7.04 2.2
65 66 -33.34 -33.36 0.06
68 69 14.87 15.01 0.9
80 81 15.15 15.11 0.23

Figure B-13 The dependence of GIC on ac voltage level: (a) instantaneous value of magnetic flux in
transformer core; (b) the dc component of magnetic flux; (c) instantaneous value of transformer
magnetisation current; (d) the dc component of transformer magnetisation
Table B-8 GIC calculation error of the LF/PDT methods as a function of GEF amplitude
GEF
amplitude
(V/km)
1 2 3 4
Mean percentage error (%) 4.1 0.5 0.3 0.3
Standard deviation of percentage error (%) 7.5 2.5 1.4 1.8
Percentage of GIC solutions with error ≥ 1.7 1.2 0.6 0.6
10%

183
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure B-14 Transformer var loss under GEF=1 V/km and θGEF=35° simulated in EMT

Table B-9 Cross-examination of var loss of load substation transformers under GEF=1 V/km and θ GEF=35°
calculated by the LF/PDT and EMT methods
Xfo Q Error Xfo Q Error Xfo Q Error
(Mvar) (Mvar) (Mvar)
bus LF/PD EMT (%) bus LF/PD EMT (%) bus LF/PD EMT (%)
T T T
1 1.29 1.19 8.48 43 0.56 0.52 7.43 80 1.25 1.2 4.04
2 2.18 2.05 6.6 44 1.33 1.24 7.61 82 2.51 2.37 6.17
3 1.2 1.09 9.24 45 0.59 0.53 12.88 83 1.58 1.5 5.11
4 0.02 0.01 62.09 46 3.48 3.28 6.02 84 0.77 0.72 7.31
6 0.5 0.45 12.53 47 0.07 0.06 23.49 85 0.07 0.05 21.24
7 0.37 0.34 8.88 48 0.63 0.59 7.05 86 0.97 0.91 6.87
11 1.14 1.04 9.91 49 2.08 1.97 5.65 88 2.34 2.29 2.02
12 0.3 0.27 14.22 50 0.34 0.32 8.27 90 1.01 0.97 5
13 1.71 1.6 6.51 51 0.13 0.11 16.16 92 1.83 1.78 3.16
14 0.5 0.46 7.74 52 1.87 1.76 6.39 93 0.32 0.3 6.86
15 1.16 1.05 10.58 53 0.7 0.65 7.49 94 0.23 0.2 11.57
16 0.54 0.5 7.48 54 5.39 5.09 5.84 95 0.42 0.37 12.83
17 0.03 0.01 139.0 54 5.39 5.09 5.84 96 0.49 0.45 10.85
2
18 0.67 0.6 12.92 55 3.11 2.88 7.95 97 0.59 0.56 6.01
19 1.61 1.47 9.73 56 3.97 3.73 6.28 98 1.67 1.59 4.75
20 0.14 0.12 14 57 1.01 0.95 6.63 100 0.25 0.22 12.59
21 0.37 0.34 8.11 58 1.16 1.09 6.54 101 2.91 2.75 5.89
22 0.41 0.37 8.07 59 0.39 0.34 13.38 102 3.72 3.59 3.66
23 2.02 1.98 2.09 59 0.39 0.34 13.38 103 1.06 0.99 6.64
27 0.53 0.48 10.57 59 0.39 0.34 13.38 104 1.51 1.38 9.2
28 1.03 0.96 7.33 60 1.3 1.25 3.92 105 0.03 0.03 27.65
29 0.16 0.14 11.37 62 1.28 1.23 3.85 106 0.37 0.33 10.31
31 0.32 0.27 16.65 66 0.9 0.86 4.64 107 0.57 0.54 6.52
32 0.38 0.33 15.17 67 1.78 1.74 2.23 108 1.02 0.97 5.52
33 0.7 0.65 7.74 70 1.47 1.38 6.89 109 0.46 0.42 8.5

184
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

34 0.64 0.59 9.07 74 0.85 0.77 9.5 110 1.87 1.72 8.51
35 0.37 0.34 9.9 75 4.24 4.06 4.53 112 0.12 0.11 16.26
36 0.22 0.2 11.12 76 0.85 0.8 7.06 114 0.21 0.19 12.45
39 0.59 0.54 9.38 77 0.74 0.7 5.7 115 0.22 0.19 13.42
40 2.2 2.08 6.02 78 2.11 1.98 6.13 117 0.24 0.22 9.86
41 0.35 0.3 15.97 79 2.53 2.37 6.94 118 0.64 0.59 9.15
42 1.53 1.4 9.48 80 1.25 1.2 4.04

Table B-10 Cross-examination of var loss of generator substation transformers GEF=1 V/km and
θGEF=35° calculated by the LF/PDT and EMT methods.
Xfo Q Error Xfo Q Error Xfo Q Error
(Mvar) (Mvar) (Mvar)
bus LF/PD EMT (%) bus LF/PD EMT (%) bus LF/PD EMT (%)
T T T
1 1.19 1.12 6.5 42 0.29 0.27 6.42 77 0.68 0.61 12.07
4 0.08 0.04 96.16 46 0.47 0.46 1.89 80 1.88 1.77 5.93
6 0.46 0.43 8.93 46 0.47 0.46 1.89 80 1.88 1.77 5.93
8 3.79 3.73 1.64 49 2.49 2.43 2.58 85 0.23 0.2 12.67
10 7.71 7.62 1.22 54 1.63 1.58 3.16 87 2.54 2.42 4.75
10 7.71 7.62 1.22 54 1.63 1.58 3.16 89 1.16 0.91 26.96
12 0.04 0.03 10.13 54 1.63 1.58 3.17 89 1.16 0.91 26.96
12 0.04 0.03 10.13 55 2.86 2.71 5.79 89 1.16 0.91 26.96
12 0.04 0.03 10.13 56 0.4 0.38 3.9 90 0.93 0.88 6.04
12 0.04 0.03 10.13 59 0.36 0.32 10.5 91 0.71 0.7 1.8
15 1.07 1 7.35 59 0.26 0.23 13.58 92 1.69 1.64 2.62
18 0.62 0.56 9.77 61 0.01 -0.02 196.2 99 0.61 0.58 5.5
2
19 1.48 1.37 7.72 62 1.17 1.11 5.25 100 0.2 0.18 8.14
24 0.71 0.64 11.25 65 5.29 5.21 1.53 100 0.54 0.5 7.84
25 0.79 0.53 49.71 65 5.29 5.21 1.53 103 2.08 2.03 2.68
26 3.02 2.89 4.34 66 1.97 1.86 5.94 104 1.39 1.29 7.2
26 3.02 2.89 4.34 66 1.97 1.86 5.94 105 0.12 0.09 23.16
27 2.38 2.19 8.61 69 0.55 0.32 70.67 107 0.13 0.12 6.79
31 0.01 0.01 20.9 70 1.35 1.28 5.6 110 1.72 1.61 6.66
32 0.35 0.31 12.12 72 0.34 0.32 6.16 111 1.47 1.44 1.87
34 0.59 0.56 5.79 73 1.47 1.41 4.09 112 0.01 0.01 17.62
36 0.77 0.73 5.59 74 0.78 0.72 8.87 113 0.68 0.67 1.88
40 7.59 7.36 3.17 76 0.78 0.74 5.66 116 3.19 3.11 2.52

Table B-11 Cross-examination of var loss of three-winding transformers under GEF=1 V/km and θGEF=35°
calculated by the LF/PDT and EMT methods.
Transformer Q (Mvar) Error (%)
From bus To bus LF/PDT EMT
5 8 7.18 6.73 6.69
17 30 1.61 1.39 15.58
25 26 5.96 5.64 5.71
37 38 1.51 1.19 26.91
59 63 4.1 3.69 11.13
61 64 1.91 1.77 7.8

185
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

65 66 2.92 2.82 3.69

68 69 1.25 0.4 211.56
80 81 4.38 3.71 18.17

Table B-12 Total var loss of transformers calculated by the LF, TS, and EMT methods.
GEF
amplitude
(V/km)
1 2 3 4
Qexe LF/PDT 242.7 485.2 727.8 970.4
(Mvar)
EMT 227.4 459.7 686.6 908.7

Table B-13 Error of calculated transformer var loss by the LF/PDT methods as a function of GEF
amplitude.
GEF
amplitude
(V/km)
1 2 3 4
Mean percentage error (%) 13 8.3 8.6 8.9
Standard deviation of percentage error 25.3 8.3 12.9 12.7
(%)
Percentage of GIC solutions with error 31.2 20.2 24.3 26
≥ 10%

Table B-14 Comparison of the minimum GEF amplitude causing voltage collapse calculated by the LF, TS,
and EMT methods.
Method LF TS EMT
Min. GEF causing voltage collapse 7 9* 5.5
(V/km)
* Var loss increased in one step.

Figure B-15 Voltage of bus 9 simulated in PDT under: (a) GEF=8V/km (stable) and (b) GEF=9V/km (unstable).

186
 TB 881 - Electromagnetic transient simulation models for large-scale system impact studies in power
systems having a high penetration of inverter-connected generation

Figure B-16 Simulation of bus voltages in EMT under: (a) GEF=4V/km (stable) and (b) GEF=5.5V/km
(unstable).

Table B-15 Minimum GEF amplitude causing voltage collapse calculated by the PDT under different
saturation time delay values.
Saturation time delay (s) 0 20 60 120 280
GEF amplitude causing voltage collapse 9 12 12 14 15
(V/km)

187
 ISBN : 978-2-85873-586-0

TECHNICAL BROCHURES
©2022 - CIGRE
Reference 881 - September 2022