727

MODELLING OF INVERTER-BASED
GENERATION FOR POWER SYSTEM
DYNAMIC STUDIES

JOINT WORKING GROUP

C 4/C6.35/CIRED

MAY 2018
MODELLING OF INVERTER-
BASED GENERATION FOR
POWER SYSTEM DYNAMIC
STUDIES
JWG C4/C6.35/CIRED
Members
K. YAMASHITA, Convenor CIGRE JP H. RENNER, Convenor CIRED AT
S. MARTINEZ VILLANUEVA, Secretary ES P. ARISTIDOU UK
T. VAN CUTSEM BE I. GREEN US
G. IRWIN CA G. LAMMERT DE
J. CARVALHO MARTINS PT L.D. PABON OSPINA DE
Z. SONG UK K. VENNEMANN DE
L. ZHU CN

Corresponding Authors
R. ADAMS AU I. ARONOVICH IS J. MA AU
B. BADRZADEH AU A.H. BAKAR MY T.E. MCDERMOTT US
M. BARBIERI IT E.T. DE BERARDINIS IT P. POURBEIK US
J.C. BOEMER US M. BRAUN DE R. TURRI IT
H. BRONZEADO BR N. CAMMALLERI IT E. VITTAL US
A. CERRETTI IT K. CHAN CH C. ZHAN DE
F.E. CIAUSIU RO V. DEBUSSCHERE FR P. MATTAVELLI IT
L. GE CN D. GEIBEL DE J. MILANOVIC UK
A. HALLEY AU S. JANKOVIC DE M. STEURER US
K. KAROUI BE L.M. KORUNOVIC RS S. UTTS RU
X. WU US

Copyright © 2018
“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 : 978-2-85873-429-0

“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any responsibility, as to the
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

EXECUTIVE SUMMARY
Over the past decades, inverter-based generators (IBGs) such as modern wind turbine generators
(WTGs) and photovoltaics (PVs), have spread around the world in response to the commitment by
numerous governments to increase renewable energy production to deal with the global warming and
other environmental concerns. In the past, the dynamics and resulting security of power systems were
largely determined by the characteristics of (large) synchronous generators connected at the
transmission system level, whereas nowadays, the impact of IBGs and their specific characteristics can
no longer be neglected and they are beginning to dominate the dynamic performance of the power
system.
In the past when the percentage penetration of IBGs was low, their impact on power system security
and performance was minimal or even negligible. In contrast, Transmission System Operators (TSOs)
are today facing operational situations where the penetration of IBGs is reaching over 50%. A number
of power systems are now operating at times with over 60% of the instantaneous load demand being
supplied from IBGs1. The increasing penetration of IBGs affects the resilience of networks to withstand
a wide range of contingency events if they are not integrated appropriately. This is in part due to the
displacement of conventional large synchronous generators with their stabilising controls (such as AVR
and PSS). The dynamic response of synchronous generators is defined by their physics (flux linkage
etc.) and controllers, whereas the dynamic response of IBGs is defined by their controllers or control
algorithms only i.e. without the physics of synchronous generators, which in turn, is specified to meet
the requirements of the relevant grid code. Such difference is likely to be a trigger to evolve some grid
codes to require new IBGs to contribute to grid stability and operation by mandating certain ancillary
service capabilities such as voltage and frequency controls. In other words, grid codes have driven the
development of IBGs.
Dynamic simulations have played an important role for many years in assessing the stability and security
of power systems. Such studies are usually performed by power system planners and operators by
means of mathematical simulation models within commercially available software tools. With this
purpose, tailored dynamic models representing all critical elements in the power system are developed,
with model complexity adjusted to account for the physical phenomena being investigated. Models for
synchronous generators and their associated controls have been developed over many years, especially
by IEEE, and are well understood and standardised. In comparison, the development and availability of
public and generic models for representing the various types of IBGs has only recently been achieved
for large utility scale IBG power plants [1], [2], and is still in its infancy for mini and micro installations
that represent a growing percentage of embedded (distributed) generation connections. In fact, for
representing IBGs for the distributed generation, industry research indicates that around one third of
utilities and system operators still model IBGs through negative loads in bulk power system dynamic
studies, effectively neglecting their dynamic behaviour [3]. According to the results of the questionnaire
survey performed as part of this Joint Working Group (JWG), the rationale behind this approach is
described as follows:
 Lack of defined modelling requirements for IBGs specific to particular power system phenomena.
 Limited access to well-validated, detailed IBG models.
 Lack of widely accepted generic IBG models for distributed generation and associated parameters.
 Varying grid code requirements.
 A lack of information about the power system at the lower voltage levels associated with distribution
and sub-transmission networks.
 A lack of an accepted (agreed) methodology for the aggregation of distributed IBGs.
 Insufficient knowledge and experience about the practical operation of IBGs in the power system.
Significant efforts have been made in the past by modelling experts to establish generic Root Mean
Square (RMS) type models through organisations like the International Electrotechnical Commission
(IEC), CIGRE and the Western Electricity Coordinating Council (WECC) in the United States [2],[4],[5].
The recent activities of the IEC working group have focused on the development of generic wind turbine
generator and wind power plant models, while the WECC have focused on large-scale IBGs connected
to the transmission system level. However, these generic models are not yet widely applied to power

1
Which may in some power systems include HVDC import from neighbouring regions or countries.

3
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

system dynamic studies which are regularly performed by TSOs and DSOs, especially in Europe. In
regards to IBGs connected to distribution networks (e.g. residential PVs), there are still no generally
accepted aggregated dynamic models available for use that adequately capture the dynamic
performance of such equipment and the impact it may have on the bulk power system.
The objective of this CIGRE and CIRED Joint Working Group is to review and report on the latest
developments relating to IBG modelling for power system dynamic studies. The scope has included both
large-scale IBGs connected to a single point of common coupling at the transmission level as well as
distributed IBGs on the medium and low voltage level. Given previous work on the modelling of WTGs,
special focus has been given to PV and battery system modelling. The Technical Brochure (TB) provides
guidance on the selection of appropriate IBG models and the required characteristics/functions that
should be represented. Model structure and functionality are described in terms of the type of dynamic
study to be undertaken and the general characteristics of the associated power system.
This TB identifies and categorises the main differences of characteristics between IBGs with minimum
functionalities and with no advanced capability and the conventional large capacity synchronous
generators which will be displaced by IBGs where there is a high level of penetration of renewables.
The major differences are summarised as follows:
 The short-term dynamic response of synchronous generators is mainly defined by physics (flux
linkage etc.)
 The short-term dynamic response of IBGs is defined by their control algorithms, which in turn
follows the requirements established in technical specifications.
It may be concluded that system inertia, short-circuit current/strength, synchronising capabilities
(existence of a synchronous torque component) and a constant internal voltage source (grid forming
capability) are inherent to synchronous generators but cannot always be easily emulated (if at all) by
IBGs from either technical or commercial perspectives. Based on these differences, the main features
that the Joint Working Group concluded that needs to be integrated into future IBG designs are identified
and detailed in the TB. Moreover, a comprehensive list of functions already offered by IBGs has been
reported, as well as the corresponding model components required to emulate the resulting
performance characteristics. The components have been categorised into three: a) Inverter control, b)
Inverter protection, c) Grid support capability.
This TB investigates two types of models: Electromagnetic Transient (EMT) and Root Mean Square
(RMS), the latter also being referred to as positive sequence modelling of the fundamental frequency
dynamic response. The benefits and limitations of each model type are presented, along with the
functionalities that need to be implemented depending on the dynamic study being undertaken. EMT
models are capable of incorporating significant levels of detail. They are typically more complex than
RMS models and generally require advanced knowledge of the equipment componentry and control
system design. They are generally unsuitable for large scale studies (incorporating hundreds or even
thousands of IBGs) due largely to the computational burden that comes with running complex models
at time steps typically in the order of ‘tens of microseconds’, as well as the difficulty of the post-
processing more complex output data from detailed 3-phase EMT models. On the contrary, RMS models
are computationally efficient and allow large scale simulations to be performed in minutes rather than
hours. Furthermore, the data inputs and post-processing of output data for RMS models is far less
burdensome. Nevertheless, RMS models have their limitations and have been identified in this TB as
being inadequate to accurately model IBGs in the following circumstances:
 Weak system conditions (typically characterised as having very low short-circuit ratio (SCR)).
 For undertaking detailed inverter and collector system design.
 For performing certain system interaction studies such as those involving sub-synchronous
resonance (SSR) and sub-synchronous control interactions (SSCI).
 For analysing the response of IBGs to unbalanced faults and resulting voltage phase angle shifts.
It remains the responsibility of the power system engineer setting out to perform dynamic simulations,
to understand the benefits and limitations of each modelling method and tool. Judicious selection of
model type is required if the power system dynamics of interest are to be properly identified and
analysed. It is emphasized that the above remarks related to the selection of the model type are more
critical especially when the penetration of IBGs becomes high.
This TB has catalogued the components and functions that need to be included in the IBG model,
depending on the power system phenomena being studied. Twenty-five functions are classified into the

4
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

three categories as outlined above. While the classifications are not without any ambiguity, they do
provide a reasonable indication of the relevance of each function for different types of power system
dynamic studies. The necessity of each function is examined for the following five power system
phenomena that are of common interest to system operators:
 Frequency deviations.
 Large signal voltage deviations (large voltage deviations associated with transient network faults
and temporary over voltages).
 Small signal voltage deviations (smaller magnitude but longer duration changes in network
voltage).
 Small signal analysis (oscillatory stability & damping studies).
 Examining network performance during unintentional islanding events.
The TB discusses how certain functions may be critical for performing one type of study, but can be
reasonably neglected when performing another. A selection of representative power system dynamic
simulation studies is also illustrated to demonstrate how certain power system phenomena interact. For
example, large voltage deviations are relevant when considering short-term voltage stability, transient
stability and LVRT/HVRT studies and there may be overlap between these issues depending on the
characteristics of the power system being considered. The necessity of each model component is
discussed, with focus on the impact that omitting certain functionalities may have when performing
specific types of analysis. Secondary modelling components, i.e. unnecessary model components to be
modelled are also identified in the TB. It is noted that so long as the IBG dynamic behaviour is sufficiently
accurate for the type of phenomenon being studied, applying appropriate simplifications which exclude
secondary components can help to reduce the computational burden and resulting time to perform
simulations.
In this TB, the model components used to represent key functionalities are further classified into two
sub-categories: a) Local, b) Plant level. This categorisation recognises the fact that single IBG
installations (such as rooftop PVs) will typically rely only on local controls within a single inverter,
whereas utility scale IBGs that potentially combine tens of hundreds of individual inverters to form an
aggregated generating system, will apply over-arching plant level controls to enable a coordinated
response to be delivered at the point of connection. When considering the structure of each model
component, there is also a need to differentiate between the RMS and EMT model types. While the
high-level controls are usually the same both for RMS and EMT models, the representation of low-level
control equipment could be significantly different depending on what the EMT model is to be used for.
Various EMT models and the corresponding positive sequence (RMS) representations are presented in
each chapter. This TB provides block diagrams for both existing functionalities already known to be in-
service, as well as future (planned) functionalities that are likely to become more common going
forward. Two complete examples of generic RMS models with representative model parameters are
provided as an appendix.
Aggregation methodologies for IBGs, specifically distributed PV installations, are not adequately defined
at the present time. This TB reviews one of the most advanced and recent aggregation methodologies
proposed by the WECC. The methodology is categorised into the following two sub-groups:
 Aggregation principles suitable for steady-state power flow and simplified short-circuit studies.
 Aggregation principles for dynamic simulations.
Prior to aggregating multiple individual IBGs, consideration needs to be given as to what functionalities
are provided by the units and whether there is an adequate level of commonality. For instance, as there
may be different grid code requirements for MV and LV connected IBGs, it may not be appropriate to
aggregate all units across multiple voltage levels to create a single model. Depending on the analysis
being undertaken, this TB asserts that it may be more appropriate to capture the individual
characteristics of MV and LV IBGs as two separate aggregated models. The same principle applies when
the rated capacity of a single IBG exceeds a certain threshold, irrespective of its connection voltage. As
a dominant source in a particular area of a network, consideration needs to be given to representing
such plant separately, using specific models.
This TB summarises the main IBG model validation methodologies that are currently used by industry.
Given that relevant work is still ongoing within the IEC [6] and has already completed in Germany [7]
to define the process of validating IBG models, this TB focuses more on available mechanisms that can
be used for validation purposes. These include the use of dedicated testing facilities, as well as the use

5
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

of real time monitoring systems to capture the performance of in-service equipment during actual power
system disturbances to validate the IBG model output provided by simulation tools. A “general model
validation iterative procedure” is provided in this TB.
This TB reviews state-of-the-art and current industry practices relating to the modelling of IBGs. It adds
to the existing narrative by providing recommendations for the ongoing development and use of IBG
models in power system dynamic studies. It has been identified that the functionality that needs to be
incorporated into IBG models is different depending on the type of dynamic study being undertaken, as
well as the characteristics of the connection point and/or power system being analysed, e.g. ‘system
strength’ as one consideration. The control block diagrams introduced in this TB are provided as
examples and are not the only way of modelling various IBG functionalities. As such, the TB does not
recommend the application of any specific dynamic model for a given power system dynamic study, but
rather identifies models which can be applied and provides some fundamental information and guidance
on their use. Based on the key findings and observations coming from the Joint Working Group activities,
this TB emphasises the necessity and importance of the proper use of the various IBG models that are
available. The objective is to encourage utilities, system operators, research institutes and academia to
focus on selecting what functionalities need to be properly represented in IBG models as well as the
type of model that is most appropriate. The need to appropriately capture the response characteristics
of embedded IBG units is also highlighted noting that, in aggregate, they may represent a substantial
contribution to the overall generation of a grid.

References

[1] P. Pourbeik, J. Sanchez-Gasca, J. Senthil, J. Weber, P. Zadehkhost, Y. Kazachkov, S. Tacke and

J. Wen, “Generic Dynamic Models for Modeling Wind Power Plants and other Renewable
Technologies in Large Scale Power System Studies”, IEEE Trans. on Energy Conversion,
September 2017, Vol. 32, No. 3, Page(s): 1108 – 1116.
[2] Ö. Göksu, P. Sørensen, J. Fortmann, A. Morales, S. Weigel and P. Pourbeik, “Compatibility of
IEC 61400-27-1 Ed 1 and WECC 2nd Generation Wind Turbine Models,” Conference: 15th
International Workshop on Large-Scale Integration of Wind Power into Power Systems as well
as on Transmission Networks for Offshore Wind Power Plants, November 2016.
[3] G. Lammert, K. Yamashita, L. D. Pabón. Ospina, et. al., “International Industry Practice on
Modelling and Dynamic Performance of Inverter Based Generation in Power System Studies,”
CIGRE Science & Engineering Vol. 8, June 2017.
[4] CIGRE C4.601 WG, "Wind turbines - Electrical simulation models for wind power generation,"
TB328, Aug. 2007.
[5] International Electrotechnical Commission (IEC) TC88 WG27, "Wind turbines - Electrical
simulation models for wind power generation," [Online]. Available:
http://www.iec.ch/dyn/www/f?p=103:14:0::::FSP_ORG_ID,FSP_LANG_ID:5613,25
[6] IEC 61400-27-2 Ed. 1: “Wind energy generation systems – Part 27-2: Electrical simulation
models – Model validation,” CD as of Feb. 2017.
[7] FGW, “Technical guidelines for power generating Units and Systems Part 4, -Demands on
Modelling and Validating Simulation Models of the Electrical Characteristics of Power Generating
Units and Systems,” FGW -TR4, Rev-08, Jan. 2016.

6
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

CONTENTS
EXECUTIVE SUMMARY ............................................................................................................................... 3

CONTENTS ................................................................................................................................................... 7

1. INTRODUCTION ............................................................................................................................. 17
1.1 BACKGROUND ................................................................................................................................................................. 17
1.2 SCOPE ................................................................................................................................................................................. 18
1.3 STRUCTURE ........................................................................................................................................................................ 19

2. CHARACTERISTICS OF INVERTER BASED GENERATION ........................................................ 21

2.1 MAIN DIFFERENCES OF CHARACTERISTICS BETWEEN AN INVERTER BASED GENERATOR AND A
SYNCHRONOUS GENERATOR ................................................................................................................................................... 21
2.2 INVERTER CHARACTERISTICS ......................................................................................................................................... 30
2.3 PRIMARY ENERGY SOURCES OF INVERTER BASED GENERATORS ...................................................................... 34
2.4 CONCLUSION ................................................................................................................................................................... 36

3. NECESSARY FUNCTIONALITIES OF INVERTER BASED GENERATORS FOR KEY

PHENOMENA ............................................................................................................................................. 39
3.1 INTRODUCTION ................................................................................................................................................................ 39
3.2 BEHAVIOUR IN RESPONSE TO FREQUENCY DEVIATIONS ..................................................................................... 41
3.3 BEHAVIOUR IN RESPONSE TO LARGE VOLTAGE DEVIATIONS............................................................................ 53
3.4 BEHAVIOUR IN RESPONSE TO SMALL AND LONG-TERM VOLTAGE DEVIATIONS ......................................... 73
3.5 SMALL DISTURBANCE ANALYSIS .................................................................................................................................. 79
3.6 UNINTENTIONAL ISLANDING OPERATION ................................................................................................................ 82
3.7 OTHER PHENOMENA AND STUDIES ............................................................................................................................ 86
3.8 SELECTION OF TYPE OF MODELS [4].......................................................................................................................... 89
3.9 CONCLUSION ................................................................................................................................................................... 91

4. EMT MODELS FOR INVERTER BASED GENERATION............................................................... 93

4.1 INTRODUCTION ................................................................................................................................................................ 93
4.2 POWER SYSTEM PHENOMENA REQUIRING EMT SIMULATION ........................................................................... 98
4.3 SELECTION OF ELECTRICAL INTERFACING MODELS AND CONTROL MODELS................................................ 98
4.4 DETAILED DESCRIPTION OF INVERTER MODELS FOR EMT STUDIES ................................................................. 101
4.5 INVERTER CONTROLS AND INTERNAL PROTECTION ........................................................................................... 105
4.6 SYSTEM MODELS FOR EMT STUDIES ........................................................................................................................ 109
4.7 CONCLUSIONS .............................................................................................................................................................. 110

5. RMS MODELS FOR INVERTER BASED GENERATION ........................................................... 111

5.1 INTRODUCTION ............................................................................................................................................................. 111
5.2 POWER SYSTEM PHENOMENA REQUIRING RMS SIMULATION........................................................................ 112
5.3 MODELLING OVERVIEW ............................................................................................................................................. 113

7
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

5.4 INVERTER MODEL, CONTROL AND PROTECTION ................................................................................................. 114

5.5 FREQUENCY AND VOLTAGE CONTROL AT COMPONENT/LOCAL LEVEL ..................................................... 127
5.6 FREQUENCY AND VOLTAGE CONTROL AT PLANT LEVEL .................................................................................. 131
5.7 FUTURE TECHNICAL CHALLENGES OF IBG MODELLING ..................................................................................... 137
5.8 REMAINING TASKS ....................................................................................................................................................... 137
5.9 CONCLUSIONS .............................................................................................................................................................. 138

6. MODELLING OF AGGREGATED DISTRIBUTED INVERTER-BASED GENERATION ........... 139

6.1 BACKGROUND AND MOTIVATION ......................................................................................................................... 139
6.2 POWER FLOW REPRESENTATION ............................................................................................................................ 141
6.3 SIMPLIFIED DYNAMIC MODEL .................................................................................................................................... 152
6.4 CONCLUSION ................................................................................................................................................................ 155

7. VALIDATION OF INVERTER-BASED GENERATOR MODEL .................................................. 157

7.1 INTRODUCTION ............................................................................................................................................................. 157
7.2 SOURCE DATA FOR MODEL VALIDATION .............................................................................................................. 157
7.3 RECOMMENDED SPECIFICATION OF MEASUREMENT DEVICES ........................................................................ 164
7.4 VALIDATION PROCEDURES......................................................................................................................................... 164
7.5 STUDIES AND TESTS TO BE CARRIED OUT FOR MODEL VALIDATION ............................................................. 165
7.6 MODEL VALIDATION EXAMPLE .................................................................................................................................. 168
7.7 CONCLUSION ................................................................................................................................................................ 175

8. CONCLUSIONS AND FUTURE WORK .................................................................................... 177

APPENDIX A. DEFINITIONS, ABBREVIATIONS AND SYMBOLS .................................................... 181

A.1. GENERAL TERMS ............................................................................................................................................................ 181
A.2. SPECIFIC TERMS ............................................................................................................................................................. 185

APPENDIX B. LINKS AND REFERENCES ............................................................................................. 187

APPENDIX C. EXAMPLE STUDIES........................................................................................................ 195

APPENDIX 1-A INTERNATIONAL INDUSTRY PRACTICE ON MODELLING AND DYNAMIC PERFORMANCE OF
INVERTER BASED GENERATION IN POWER SYSTEM STUDIES......................................................................................... 195
APPENDIX 1-B MODELLING AND DYNAMIC PERFORMANCE OF INVERTER-BASED GENERATION IN POWER
SYSTEM STUDIES: AN INTERNATIONAL QUESTIONNAIRE SURVEY ................................................................................ 207
APPENDIX 2-A DYNAMIC VOLTAGE BEHAVIOUR BEFORE AND AFTER ISLANDING IN RESPONSE TO REACTIVE
POWER VARIATION ................................................................................................................................................................... 213
APPENDIX 3-A EXAMPLE PROTECTION FOR INVERTER-BASED GENERATOR CONNECTING IN MV AND LV
NETWORK 215
APPENDIX 3-B CALIFORNIA ISO EXPERIENCE: OVER-GENERATION ASSESSMENT ............................................... 222
APPENDIX 3-C LVRT STUDY FOR FREQUENCY STABILITY IN TASMANIAN GRID .................................................... 234
APPENDIX 3-D IMPACT OF CCGT TRIPPING WITH PV ON FREQUENCY STABILITY IN ISRAELI GRID .............. 235
APPENDIX 3-E CALIFORNIA ISO EXPERIENCE: IMPACT OF PV ON SHORT-TERM VOLTAGE STABILITY ........... 239
APPENDIX 3-F EXAMPLE MODEL COMPONENTS FOR FREQUENCY STABILITY STUDY ........................................ 249

8
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 3-G HIGHLIGHT OF GRID-CODE REQUIREMENTS IN STEADY STATE .................................................... 250

APPENDIX 3-K PROBABILISTIC APPAROCH FOR DYNAMIC STABILITY STUDY ........................................................ 252
APPENDIX 5-A REPRESENTATION OF UNBALANCED FAULT USING RMS MODEL [44], [45] ............................... 266
APPENDIX 5-B PV ARRAY MODEL ...................................................................................................................................... 270
APPENDIX 5-C EXAMPLES OF RMS-TYPE INTEGRATED PV MODELS ......................................................................... 272
APPENDIX 5-D BATTERY ENERGY STORAGE MODELS .................................................................................................. 284
APPENDIX 6-A EXAMPLE OF TYPICAL DATA FOR OVERHEAD AND UNDERGROUND CABLES IN PORTUGAL 288
APPENDIX 7-A AN OPEN-SOURCE DISTRIBUTED CONTROL PLATFORM FOR HIL-BASED TESTING AND
DEMONSTRATION OF ADVANCED POWER SYSTEMS ...................................................................................................... 289
APPENDIX 7-B TYPE OF STUDIES AND TESTS FOR MODEL VALIDATION ................................................................. 292
LINKS AND REFERENCES OF APPENDIX C ............................................................................................................................. 294

9
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

FIGURES AND ILLUSTRATIONS

Figure 2.1 Image of generator output and frequency reponses in case of generator tripping. ........... 22
Figure 2.2 Example transition of synchronizing torque coefficient in case of a generator tripping. ..... 24
Figure 2.3 General control circuit structure of PV. .......................................................................... 31
Figure 2.4 Main components of generic inverter-based generator. .................................................. 35
Figure 2.5 PV systems, battery systems or fuel cells (a) micro-turbine or full converter WTG (b) and
DFIG (c). .................................................................................................................................... 35
Figure 3.1 Transient stability scenario due to loss of generation. .................................................... 42
Figure 3.2 WAMS recording (50 Hz sampling)................................................................................ 43
Figure 3.3 System response after the normative contingency in interconnected operation [42]. ....... 44
Figure 3.4 Example of measured PV output (12 kW three phase inverter). ...................................... 45
Figure 3.5 July 2002 peak week wind and load in NYISO [44]. ....................................................... 46
Figure 3.6 Solar energy production during the solar eclipse and on the adjacent days in CAISO [45]. 47
Figure 3.7 Solar energy production during the solar eclipse and the system frequency [45]. ............. 47
Figure 3.8 Frequency on a 500 kV bus in central California with an outage of two nuclear units in the
cases with high and low headroom and inertia. ............................................................................. 48
Figure 3.9 Simulation of 50.2 Hz effect, tripping and automatic reconnection of PV in Europe [46]. .. 49
Figure 3.10 Voltage phrase jump mechanism by means of active power [47]. ................................. 51
Figure 3.11 One example mechanism how frequency is forced to be oscillated via Q variation type
active anti-islanding control signal. ............................................................................................... 52
Figure 3.12 Voltage phase jump mechanism by means of reactive power [47]. ............................... 52
Figure 3.13 Image of dynamic behaviour of voltage and current in case of OOS. ............................. 54
Figure 3.14 Example short-circuit current (schematic diagram) [50]. .............................................. 55
Figure 3.15 Type of studies and phenomena with large voltage deviation. ...................................... 56
Figure 3.16 Measured response obtained in power system simulator [53]. ...................................... 59
Figure 3.17 WECC test system with large-scale PV plants [54]. ...................................................... 60
Figure 3.18 Voltage at PCC (230 kV) [54]. .................................................................................... 62
Figure 3.19 Voltage at load bus [54]. ............................................................................................ 62
Figure 3.20 Example of short-circuit current provision measured in PV plant. .................................. 64
Figure 3.21 Two equivalent machine system and example power swing oscillation. ......................... 65
Figure 3.22 Change in power flow over transmission lines before and after disconnection of PVs at
different locations [16]. ............................................................................................................... 65
Figure 3.23 Rotor angle of G1 with relative to centre of inertia with various level of LVRT requirements
[16]. ........................................................................................................................................... 66
Figure 3.24 Western interconnection frequency during fault [58].................................................... 66
Figure 3.25 SCE solar resonance output SCADA graph [58]. ........................................................... 67
Figure 3.26 System voltage with PLL (a) tracks (b) loses voltage [57]. ........................................... 70
Figure 3.27 Expanded Nordic system [65]. .................................................................................... 75
Figure 3.28 Detailed distribution network model [63]. .................................................................... 75
Figure 3.29 LVRT capability curve of RES [65]. .............................................................................. 76
Figure 3.30 Voltage on TN bus 4044 [65]. .................................................................................... 76
Figure 3.31 Terminal voltages of two RES units in two different DNs (The one in DN1 violates the
LVFRT curve while the other not) [63]. ......................................................................................... 77
Figure 3.32 Total active power generation by RES in all the DNs [65]. ............................................ 77
Figure 3.33 Schematic representation of unintentional islanding. .................................................... 82
Figure 3.34 Classification of Sub-synchronous interaction [31]. ...................................................... 87
Figure 3.35 Cumulative distribution function of transient stability index .......................................... 88
Figure 3.36 Frequency nadir with response to the instant penetration of RES (40% loading) [86]. ... 89
Figure 3.37 Grid frequency on bus 60 [86]. ................................................................................... 89
Figure 4.1 EMT formulation of inductor differential equation as a difference equation via Trapezoidal
integration. ................................................................................................................................. 94
Figure 4.2. EMT formulation of a capacitor differential equation as a difference equation via
Trapezoidal integration. ............................................................................................................... 94
Figure 4.3. Linear equivalent circuit - conversion of differential equations to finite difference
equations. ................................................................................................................................... 95
Figure 4.4. Structure of discrete switch model. ............................................................................. 101

13
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 4.5. Stages of forward collapsing and back substitution for derivation of an averaged switch
model. ....................................................................................................................................... 102
Figure 4.6. Comparison of currents (a,b,c) from averaged switch model and exact discrete switch
model for a transient event [103]. ............................................................................................... 104
Figure 4.7. Simple current source model interfaced to the EMT electrical circuit. ............................ 104
Figure 4.8. Typical cascaded PI control structure for VSC inverters. ............................................... 105
Figure 4.9. Typical phase locked loop (PLL) .................................................................................. 106
Figure 4.10. Example of PWM firing for 2 level VSC converters. ..................................................... 107
Figure 4.11. Example of multi-level converter output (400 MW with 200 power modules per converter
arm). ......................................................................................................................................... 108
Figure 5.1 Overview of IBG model. .............................................................................................. 114
Figure 5.2 Inverter generator model. ........................................................................................... 114
Figure 5.3 Reference frame for phasors and active/reactive currents. ............................................ 115
Figure 5.4 Example of converter with LVPL block diagram with peripheral blocks [114]. ................. 116
Figure 5.5 Example of current limiter block diagram of PVs [113]. ................................................. 116
Figure 5.6 Example of HVRT block diagram of PVs [113], [114]. ................................................... 117
Figure 5.7 Example of frequency ride-through block diagram. ....................................................... 118
Figure 5.8 Example of LVRT characteristics [17]. .......................................................................... 118
Figure 5.9 Example pseudo code of LVRT characteristics [116]. .................................................... 119
Figure 5.10. Generic PLL structure and operation [117]. ............................................................... 120
Figure 5.11 Example of simplified PLL behaviour [114]. ................................................................ 120
Figure 5.12 Example of DC source model [118]. ........................................................................... 121
Figure 5.13 Example of restarting sequences for PVs in LV network [120]...................................... 122
Figure 5.14 Example of grid protection system control block diagram [113]. .................................. 123
Figure 5.15 Example of over- and under- voltage relay models for single-phase PV inverter. ........... 123
Figure 5.16 Example of over- and under-frequency relay models for single-phase PV inverter. ........ 124
Figure 5.17 Example of instantaneous overvoltage relay models for single phase PV inverter. ......... 125
Figure 5.18 Example of voltage phase jump type passive anti-islanding protection block diagram
[124]. ........................................................................................................................................ 125
Figure 5.19 Example of rate of change of frequency type passive anti-islanding protection block
diagram [124]. ........................................................................................................................... 126
Figure 5.20 Example of reactive power variation type active anti-islanding protection block diagram
[124]. ........................................................................................................................................ 126
Figure 5.21 Example of frequency-shift type active anti-islanding protection block diagram [124].... 127
Figure 5.22 Local frequency-and voltage controller [111]. ............................................................. 127
Figure 5.23 Example of reactive power control block diagram [114]. ............................................. 128
Figure 5.24 Example of reactive power control block diagram [114]. ............................................. 129
Figure 5.25 Example of active power control block diagram [113]. ................................................ 130
Figure 5.26 Representative flowchart of voltage rise mitigation functions [125]. ............................. 131
Figure 5.27 Representative PV power plant topology [111]. .......................................................... 132
Figure 5.28 Plant level Q control block diagram for PV plants and wind farms [111]. ...................... 133
Figure 5.29 Plant level Q control block diagram for PV plants [113]. .............................................. 133
Figure 5.30 Dynamic voltage control characteristics [126]. ............................................................ 133
Figure 5.31 Example of frequency control categories [127]. .......................................................... 134
Figure 5.32 Plant level primary frequency control block diagram for PV plants and wind farms [109].
................................................................................................................................................. 134
Figure 5.33 Plant level active power control block diagram for PV plants [113]. .............................. 135
Figure 5.34 Inertial response profile [48]. .................................................................................... 136
Figure 5.35 Functional representation of a closed-loop inertia-based FFR control [9]. ..................... 136
Figure 5.36 Example of LFSM-O block diagram. ............................................................................ 137
Figure 6.1 Response to a three-phase fault with varying control parameters for PV in LV networks
[134]. ........................................................................................................................................ 140
Figure 6.2 Example of utility-scale PV power plant topology [7]. .................................................... 142
Figure 6.3 Single-machine equivalent representation of an example solar PV plant [7]. .................. 142
Figure 6.4 Sample utility-scale PV plant topology [137]. ................................................................ 144
Figure 6.5 Composite load model with distributed behind the meter generation [111]. ................... 147
Figure 6.6 Recommended power flow representation for study of medium-penetration PV scenarios
with single equivalent distribution impedance [7]. ........................................................................ 148

14
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 6.7 Recommended power flow representation for study of medium-penetration PV scenarios
with multiple equivalent distribution impedance and changing interconnection requirements [130],
[131]. ........................................................................................................................................ 150
Figure 6.8 Block diagram of WECC small PV plant model [111]. ..................................................... 153
Figure 7.1 Model validation test systems for PV inverter in CEPRI. ................................................. 159
Figure 7.2 Model validation test systems for PV inverter in CRIEPI. ............................................... 160
Figure 7.3 Example interactions between synchronous generator and 24 single-phase PV inverters. 161
Figure 7.4 The 5 MW PHIL facility at Florida state university-CAPS. ............................................... 163
Figure 7.5 CAPS open-source distributed control system layout. .................................................... 163
Figure 7.6 Flowchart of general model validation procedure. ......................................................... 165
Figure 7.7 Model validation of large voltage disturbance. .............................................................. 169
Figure 7.8 Model validation of active power command disturbance. ............................................... 169
Figure 7.9 Schematic diagram of Ningxia network. ....................................................................... 170
Figure 7.10 Structure of PV power plant A. .................................................................................. 171
Figure 7.11 Representative structure of PV plant A [157]. ............................................................. 172
Figure 7.12 Example of measured and simulated responses following faults at 750 kV [157]. .......... 173
Figure 7.13 Chinese standard local controller module model [113]. ............................................... 174

TABLES
Table 2.1 Major existing and/or potential differences between IBGs and synchronous generators .... 28
Table 2.2 Requirement for IBG in grid code and standard .............................................................. 32
Table 2.3 Classification of inverter-based generator technologies ................................................... 35
Table 2.4 List of functionalities of IBGs ......................................................................................... 37
Table 3.1 Type of phenomena and type of studies ........................................................................ 39
Table 3.2 Size of grid and key frequency control ........................................................................... 45
Table 3.3 Necessary functionalities for frequency deviations .......................................................... 50
Table 3.4 Difference between normal short-circuit and out-of-step phenomenon ............................. 54
Table 3.5 Condition of voltage and reactive power control [54] ...................................................... 61
Table 3.6 Summary of short-term dynamic voltage response [54] .................................................. 61
Table 3.7 Necessary functionalities for large voltage deviations ...................................................... 69
Table 3.8 Necessary functionalities for small and long-term voltage deviations ................................ 78
Table 3.9 Necessary IBG’s functionalities for small signal stability analysis ...................................... 82
Table 3.10 Necessary functionalities for unintentional islanding operation ....................................... 86
Table 3.11 Necessary functionalities for controller interaction studies (simplified table).................... 88
Table 3.12 Selection of type of models for typical power system dynamic studies ............................ 90
Table 4.1 Type of phenomena and studies typically performed using EMT-type models ................... 98
Table 4.2 Control model for type of study .................................................................................... 101
Table 5.1 Type of phenomena and studies typically performed using RMS-type models .................. 112
Table 5.2 Overview of components and selection in RMS-type models ........................................... 113
Table 5.3 Recommended performance guidelines for inertial response profile in Hydro-Quebec [48]136
Table 6.1 Computation of collector system’s equivalent parameters for sample system in Figure 6.4.
................................................................................................................................................. 145
Table 6.2 Sample equivalent collector system parameters [139] .................................................... 145
Table 6.3 Suggested data for single distribution network equivalent [139] ..................................... 149
Table 6.4 Example of modelling equivalent solar PV on distribution network .................................. 149
Table 6.5 Suggested data for multiple distribution system equivalents for typical German active
distribution systems (Data provided here is simply an example, more detail discussion required for
specific systems) [6], [131], [137] .............................................................................................. 150
Table 6.6 Input parameters and sample settings of PVD1 [111] .................................................... 154
Table 6.7 Internal variables of PVD1 [111] ................................................................................... 154
Table 7.1 Pros and cons of play-back method and full grid simulation method [155] ...................... 165
Table 7.2 Type of studies and tests for model validation ............................................................... 166
Table 7.3 Major specification of PV inverter .................................................................................. 171
Table 7.4 Derived model parameters ........................................................................................... 174

15
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

EQUATIONS
Equation 2.1 ............................................................................................................................... 23
Equation 3.1 ............................................................................................................................... 51
Equation 3.2 ............................................................................................................................... 52
Equation 3.3 ............................................................................................................................... 72
Equation 3.4 ............................................................................................................................... 79
Equation 3.5 ............................................................................................................................... 79
Equation 4.1 .............................................................................................................................. 109
Equation 5.1 .............................................................................................................................. 115
Equation 5.2 .............................................................................................................................. 117
Equation 5.3 .............................................................................................................................. 119
Equation 5.4 .............................................................................................................................. 119
Equation 5.5 .............................................................................................................................. 122
Equation 5.6 .............................................................................................................................. 129
Equation 5.7 .............................................................................................................................. 136
Equation 6.1 .............................................................................................................................. 143
Equation 6.2 .............................................................................................................................. 144
Equation 6.3 .............................................................................................................................. 145
Equation 6.4 .............................................................................................................................. 146
Equation 6.5 .............................................................................................................................. 146
Equation 6.6 .............................................................................................................................. 146

16
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

1. INTRODUCTION
1.1 BACKGROUND
With the sudden uptake of renewable energy sources (RES) in many power systems around the world,
the demand for high quality, validated dynamic models to capture their performance characteristics has
dramatically increased. While significant efforts have been made recently to document and validate wind
turbine models [1], [2], [3], other types of RES are just starting to gain attention. These include
photovoltaics (PVs), micro-turbines of various configurations, as well as battery storage systems, either
forming part of a renewable energy generating system or standing alone. A common characteristic of
RES is that they are typically interfaced to the grid through power electronic inverters. As a result, they
have a different dynamic response characteristic when compared to that of classical synchronous
generators. None-the-less, there are a few recent examples of both wind and PV model validation using
recently developed generic models for large scale winds and PVs at the transmission system level [1],
[2].
A notable observation is the variety of RES, both in terms of scale as well as diversity of network
connection points. Transmission connected wind and solar plants are now relatively common place, with
single network connection points facilitating many hundreds of megawatts (MW) or more of installed
capacity. However, rooftop PV has also found favour in many countries but is comprised of a multitude
of small installations ranging from a few kilowatts (kW) to tens or hundreds of kW at a single location.
This type of RES is normally connected to the local distribution network. While individual unit sizes may
be small, the aggregate capacity can be very significant and can represent a major generation source
for the broader power systems. For example, the total installed capacity of rooftop PV in the Australian
National Electricity Market (NEM) has reached approximately 5 gigawatts (GW) in 2017 and continues
to grow. This compares with a typical NEM wide system load demand of approximately 25-30 GW.
The existing and forecast prevalence of RES has provoked serious concerns in the industry as to how
these new technologies can and should be represented in dynamic simulations. In practice, there is a
lack of validated dynamic models available for many individual RES technologies such as photovoltaics,
fuel cells, micro turbines and other inverter-based sources. In addition, there is no agreed methodology
as to how to aggregate and represent the enormous number of distributed RES so that their response
characteristics can be accounted for as part of system-wide dynamic simulations.
As the penetration of such RES technologies continues to increase, the stability and dynamic
performance characteristics of the power system will change as will the impacts on network protection
systems and various aspects of power quality. Higher penetrations of RES will also make real time
system operation more challenging than in the past, for both transmission system operators (TSOs) and
distribution system operators (DSOs).
TSOs routinely perform dynamic time-domain simulations to assess the stability of their power systems.
The requirements to do so are often embedded within grid codes, systems standards or rules that govern
e.g. the connection to their networks and the operation of the electricity system. While technical
information (including modelling data) is generally available for the transmission system and the large
centralised generating systems that connect to it, the models that are currently used to represent
distribution networks (and any RES embedded within) are typically based on a very limited amount of
information. In many cases, TSOs simplify the sub-transmission and/or distribution network down to a
‘net load’ representing the power exchange at the HV-MV boundary. The ‘net load’ is assigned a model
which attempts to capture all of the downstream dynamic response characteristics including embedded
RES and all load devices.
DSOs may or may not have a detailed representation of their network for use in simulation software
packages and often rely on load flow analysis rather than dynamic studies. They typically have steady
state data available for electricity consumers and (to some extent) embedded generators which may
include load profiles, equipment ratings, installed capacities etc, however these data are generally not
suitable for developing dynamic models. Larger DSOs may have access to significantly more technical
information depending on the complexity of the networks they are responsible for.
In order to better understand what information is available to various parties and explore what IBG
models are currently being used for dynamic time-domain simulations by TSOs and DSOs, a
comprehensive questionnaire was developed by the JWG and distributed in 2015 [4],[5]. The

17
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

questionnaire was sent to 63 utilities and system operators covering some 21 countries on all continents.
The results of the survey, based on 45 responses (71% response rate) are summarised in this TB.
The survey revealed that the ‘negative load model2’ is still the most widely used IBG representation
within dynamic simulations intended to investigate frequency stability and rotor angle stability. It also
showed that Root Mean Square (RMS) IBG models are more likely to be used for frequency stability
studies and rotor angle stability studies, while the Electro-Magnetic Transient (EMT) type IBG models
are more likely to be used for short-term voltage stability studies, fault ride through (FRT) studies, and
various EMT studies. A full analysis of the survey results is given in Appendices 1-A and 1-B.
It should be emphasised that it is necessary to use simplified models for most bulk power system
dynamic studies so that the simulation run times can be maintained in a reasonable range (See also
Sub-Chapter 6.1). This TB discusses the importance of understanding the impact that various functions
and characteristics embedded in IBGs can have on different types of power system studies and ensuring
that they are suitably represented in any model that is then applied when performing those studies. An
acceptable model should capture what is important and apply simplifications where appropriate in the
interests of being efficient.
It can be seen that there are a number of important issues facing the power industry:
 Increasing penetration of IBG technologies.
 Lack of validated dynamic models available for many IBGs including PV, the installed capacity
of which is already significant and growing in a number of countries3.
 Lack of understanding as to what functionalities are important for different types of power system
studies, i.e. what should be modelled explicitly and what can be reasonably ignored or
represented in a simplified way. This issue inherently includes a consideration of when RMS or
EMT models are most applicable for use.
 No agreed or well documented methodology to perform aggregation of embedded IBGs.
 Given that detailed design information may simply not be available to develop explicit models
of some types of IBGs, there is a need for more generic modelling information with appropriate
guidance provided on its use.
On the other hand, no particular guidance for the model selection is also seen as an important issue
facing the academia. Inappropriate model selection with inappropriate model parameter(s) can be
observed even in journal papers, which can cause further inappropriate model selection in other
research studies.
It is in this context that this TB has been developed.

1.2 SCOPE
The CIGRE and CIRED JWG, “Modelling and dynamic performance of inverter based generation in
power system transmission and distribution studies” was established in 2014. The aims of this JWG, as
described in its terms of reference, were to address the following issues:
 Provide a critical overview of existing RES dynamic simulation models and modelling
methodologies, focusing primarily on photovoltaics and some other inverter-based sources. The
review should include relevant model parameters for both distribution and transmission system
studies.
 Generation technologies for which adequate dynamic simulation models do not presently exist or
are not appropriate for the expected purposes will be identified. The activities of existing working
groups within the IEC and WECC will be considered. Suggestions will be offered on potential
improvements to existing models and modelling methods as appropriate.
 Develop a set of recommendations and step-by-step procedures for developing dynamic models
for RES and consider how such models may be validated.
 Provide recommendations for developing equivalent aggregated models for simulating clusters of
similar types of RES technologies.

2
Load voltage characteristics and load frequency characteristics are not specified in the survey.
3
There are now validated models of relevant manufacturers of WTGs and PVs which are available. For example, the list of
validated models currently includes over 1000 models of different manufacturers in Germany.

18
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 Provide an overview of new system performance issues that may arise as a result of very large
penetration of IBG (and load) technologies.
It should be noted that models for distributed generation will be aggregated models as seen at the MV-
LV and/or HV-MV interface and should therefore try to account for:
 Extension, configuration and composition (bare conductors, cables, etc.) of the LV and MV
networks.
 Characteristics of any embedded generation (types of IBG, installed capacity, built-in protective
functions, control capabilities and associated settings, etc.).
 Automated operation and/or protection systems such as load-shedding functions, self-healing
characteristics, etc.
 Issues associated with islanded operation of MV networks.
The appropriate characterisation of loads is also important in these activities. Much of this work has
already been completed by CIGRE WG C4.605 “Modelling and aggregation of loads in flexible power
networks” [6]. The intent of this JWG has not been to significantly expand upon the work of C4.605 but
rather focus on the modelling of embedded generation that may form part of the overall aggregate
response at the point of common coupling (PCC).

1.3 STRUCTURE
In addition to the introductory chapter, this TB contains further seven chapters and a number of
appendices. The appendices include further detailed analyses, case studies and descriptions of different
IBG models (with sample parameters).

1.3.1 Characteristics of IBG (Chapter 2)

Originally, IBGs were designed with a minimum set of functions, driven by the limited technical
requirements necessary for their connection to the network at the time. As a result, key capabilities
which contribute to system reliability and security were not implemented. Because synchronous
generators (which inherently offer many of these capabilities) are now being displaced with IBGs, the
increasing levels of RES integration is beginning to have negative impacts on power system security
and dynamic performance.
In recent years, grid code requirements have evolved such that new connected IBGs need to provide
more functionalities and capabilities, similar to that offered by synchronous generators. Nevertheless, a
significant percentage of existing IBGs in many countries still remain connected without necessarily
complying with the technical requirements unless a retrofitting campaign is enforced.
Chapter 2 examines the important technical characteristics of IBGs that need to be accounted for when
developing dynamic models for use in power system simulation studies.

1.3.2 Necessary functionalities of IBG for key phenomena (Chapter 3)

To decrease the computational burden involved in large-scale stability studies, Chapter 3 lists which
functions should be represented in a dynamic model for each type of power system phenomenon
typically studied. It also notes which functions can be reasonably neglected. Trying to include all
functionalities in every type of study could be inefficient and is unnecessary. Once the engineer selects
the type of study to be performed, the TB helps to define the necessary IBG functions that should be
included in the dynamic model.

1.3.3 EMT models for IBG (Chapter 4)

When the phenomena to be studied is significantly outside the bandwidth of RMS models, i.e.
fundamental frequency models (for example analysis of switching transients or sub-synchronous
torsional interactions, etc.), then EMT simulations should be conducted using detailed, equipment
specific models. EMT analysis tools solve the differential-algebraic equations of a three-phase electrical
network (as compared to transient stability analysis which generally uses RMS positive sequence
phasor equations to represent the fundamental frequency response of the electrical network).
This distinction means that EMT analyses using appropriately detailed models are capable of
representing the non-linear response of electrical devices (e.g. transformer saturation or surge

19
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

arresters) and are suitable for investigating issues such as harmonic instability phenomena, sub-
synchronous resonances, AC transient overvoltages, lightning surges, and the control interactions of
power electronic devices. Chapter four investigates some of these issues in the context of IBG impacts
on AC power systems.

1.3.4 RMS models for IBG (Chapter 5)

RMS models are mainly used to study the stability of large interconnected power systems, including
phenomena such as electromechanical oscillations (small-signal stability), rotor-angle stability of
synchronous generators and voltage and frequency stability. Phasor simulation methods, using RMS
models, are used when the fundamental frequency behaviour is of interest.
The network is simulated with fixed complex impedances for modelling its fundamental frequency
behaviour. Converters are included in RMS programs using their positive sequence equivalent models.
They capture the fundamental frequency behaviour of the converter while ignoring fast switching
transients and simplifying control and protection functions that would otherwise require a more detailed
representation to capture the behaviour and their operation fully. Chapter 5 explores the benefits and
limitations of RMS modelling techniques in the context of representing IBGs in power system simulation
studies.

1.3.5 Modelling of aggregated distributed IBG (Chapter 6)

It is well known that wind and solar plants (parks) may contain many individual wind turbine generators
(WTG) and individual PV inverter units, respectively. As different IBG could have different dynamic
behaviours following faults, the individual modelling of each IBG type is an ideal solution for accurately
representing such dynamic behaviours. However, as the number of RES increases, it is becoming more
challenging to model the huge number of individual generators as part of large-scale dynamic stability
studies (mainly due to the high computational burden and the limited assigned time for completing
analysis activities).
Therefore, aggregation techniques need to be applied to achieve a reasonable balance between the
accuracy and the computational burden of the time-domain simulation. The key observations for the
modelling of aggregated IBGs are summarised in Chapter 6 including that provided by the WECC [7].

1.3.6 Validation of IBG models (Chapter 7)

Model validation is an important aspect of any model development process. It is important for the model
vendor to ensure the validity of its products as well as the end-users who rely on the models for a host
of reasons, including maintaining power system security and reliability. It should be noted that there are
often many differences between the models used by manufacturers and the ones made available to
utilities. The former can be built on the individual cell-inverter level with very detailed and complicated
control and protection logical circuits for equipment design. The model used by utilities and system
operators is often simplified with many devices being represented by a lumped element in the model.
The guidance provided in this chapter on model validation approaches applies only for lumped models
representing an aggregated PV power plant connected to the grid. It may be partially applied to wind
farms, too.

20
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

2. CHARACTERISTICS OF INVERTER BASED

GENERATION
2.1 MAIN DIFFERENCES OF CHARACTERISTICS BETWEEN AN INVERTER BASED
GENERATOR AND A SYNCHRONOUS GENERATOR
As interconnection requirements (grid codes) evolve, inverter-based generator (IBG) functionalities and
their behaviours will approach that of large capacity synchronous generators but at the present time, i.e.
as of 2017, their behaviours are quite different as detailed in the following paragraphs. The term, “IBG”
which is used in Sub-Chapter 2.1 only denotes IBGs with minimum functionalities and with no advanced
capability4. The term, “synchronous generator” which is used in Sub-Chapter 2.1 only denotes large
capacity synchronous generators and which it is assumed will be replaced with the IBGs where there is
a high level of penetration of renewables. The main difference of characteristics and behaviours between
the IBG and the conventional synchronous generator are summarized as follows:
(1) Rotating mass/inertia: Inverters do not have a rotating mass component; i.e. there is no inherent
inertia. The prime mover behind the inverter might have the inertia, but its “usage” has to be
achieved via the inverter controls and the inverter size because all IBGs are limited in terms of
maximum current through the power electronics devices, as well as maximum voltage. To use the
real available inertia, if any, of the “prime mover”, a significant oversize of the inverter generator 5
or room to increase output may be necessary because the available energy reserve in IBGs is very
limited and thus the inertial response of IBGs is eventually withdrawn. Moreover, synthetic inertia
cannot be considered completely equivalent to the inertia provided by conventional synchronous
generators which are directly connected to the grid as measuring devices and controls introduce
delays to the synthetic inertia reaction to events in the grid. Although the fast frequency response
has been commercially available, the synthetic inertia is still carefully evaluated and not in practical
use.
One of the promising schemes for representing the synthetic inertia captures the Rate of Change
of Frequency (ROCOF) and increases or decreases the IBG output so that the frequency change
is mitigated. This concept enables the reduction of the mismatch between the mechanical output
and the electrical output when ROCOF is not zero. When a generator tripping is considered as an
example, the power output of remaining synchronous generators shows the stepwise increase right
when the generator tripping occurs (See Figure 2-1).
It should be noted that the ROCOF is zero at the moment the generator trip occurs because the
system frequency is the pre-disturbance/initial frequency at the moment. Even if the primary
frequency response can be ideally emulated in IBGs, the immediate increase in the IBG output
cannot be observed without the inertia effect (See pink dotted line of PG2 in Figure 2.1). Such
stepwise increase in the synchronous generator output will definitely alleviate the frequency drop.
Synthetic inertia cannot achieve this behaviour mainly due to delay in ROCOF measurement,
filtering and control. Therefore, the synthetic inertia concept of modifying the controls dependent
on the measured ROCOF cannot be currently considered completely equivalent to the inertia
provided by conventional synchronous generators. It should be noted that other concepts for
inverter controls are also under discussion [8], [9], which may offer other means of control in the
future, e.g. concepts such as using a battery system for providing the fast frequency response
and/or the synthetic inertia. (See bullet point 4).

4
In other words, IBGs which were in their infancy and the size of which was unlikely to be categorized as large utility scale.
5
Air-cooled IGBT converters have substantial short-term overload capability (for around up to 1 s).

21
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Generator Tripping

Output
Active
Power PG1

0
With inertia of PG2
Output
Active
Power

PG1 G
PG2
W/o inertia of PG2 PG2 G
0 Load
W/o inertia of PG2 PG3 G
Output
Active
Power

PG3
With inertia of PG2
+ 1
Dw
0 Pm -Pe
Ms
50 Hz -
Primary
DPm Frequency
Frequency
System

System Frequency Controller

Pm: Mechanical Power
Pe: Electrical Power
w: Rotating Speed (=2pf)
Time M: Inertia (=2H)

Figure 2.1 Image of generator output and frequency reponses in case of generator tripping.
(2) Fault current contribution: Inverters predominately lack inductive characteristics that are
associated with rotating machines because it is controlled by power electronic equipment and not
by electrical machines. The classical short circuit current contribution expected from synchronous
machines does not apply (as caused by the law of constant flux in rotating machines). Instead, a
short circuit contribution is possible by means of inverter control. However, this contribution is
typically limited to slightly above 1 p.u. current (limited overload capability of a semiconductor power
electronics device), provided that all the active power supplied to the network is reduced to zero
and all the current which is able to flow through the power electronics devices without damaging
them is turned into reactive power. Of course, a certain oversize of IBGs would help to reduce also
this gap with respect to traditional synchronous generators. If the voltage at the PCC during a fault
is very low, the phase angle of the current injected by the inverter may be ill defined, which means,
the expected fault current is unlikely to be provided no matter how oversized IBGs are applied.
Therefore, many grid codes exempt IBGs from providing reactive current and allow to cease the
current injection when the residual voltage is below a threshold value, such as 20% of rated voltage.
It is noted that the limited infeed of the fault current is revealed when the PCC is located near the
fault point only. In other words, if the voltage at the PCC during a fault is not too low, the substantial
infeed of the fault current may be expected regardless of the electrical distance between the PCC
and the fault point.
(3) Control response capability: The control response of inverters can be extremely fast (certainly
faster than a rated frequency cycle). This offers the opportunity to design the inverter response to
be quite flexible. Thereby, both the needs of distribution and transmission system can be taken into
account, even implementing different behaviours/responses in the inverter generators according to
external signals/commands, voltage/frequency measurements, presence of local fault or
perturbation on transmission system, etc. Conversely, inadequate design of controls may result in
abnormal behaviours affecting the power system, both in normal operation and unintentional
islanding (described in Chapter 3), e.g. because of a too fast response by the control loops to even
small voltage and frequency variations.
(4) Constant voltage source6: The voltage induced in the windings of a synchronous generator (also
known as internal induced voltage) is typically larger than the grid voltage. Moreover, this internal

6
The voltage denotes the synchronous internal voltage. It does not denote the terminal voltage of synchronous generators.

22
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

induced voltage is independently regulated from the grid voltage. It will cause increased current
injection as the grid voltage sags and hence typically contributes positively to network stability.
IBGs do not have such an inherent internal voltage source. The current that can be provided to the
grid during a voltage sag is dominated by the IBG control behaviour and typically limited to 1 p.u. It
should be noted that the operation mode of IBGs typically cannot change without stopping the
inverter, although the IBGs also have the ability to create voltage through U-F mode (also called
isolated operation mode [10]7) instead of P-Q mode.
(5) Transmission-level voltage support: Large capacity synchronous generators generally operate
in AVR mode. That means such generators have an ability to regulate the terminal voltage and the
system voltage in HV network (e.g. typically equal to or higher than 200 kV) near the generator bus.
Small capacity synchronous generators generally operate with Automatic Q Regulator (AQR) or
Automatic Power Factor Regulator (APFR) mode (this means they do not regulate the terminal
voltage but the reactive power or the power factor coming from the terminals). That is because the
reactive power injection of these units is limited, thus they do not have a capability for changing the
terminal voltage and regulating the system voltage (See Appendix 2-A). The IBGs are generally
assumed to operate with a unity power factor. That means most of the IBGs operate with AQR
mode, the power factor of which is one8. Thus, actual voltage control for transmission-level voltage
support cannot be expected or achieved. However large-scale IBGs can support the transmission-
level voltage with the aid of other external voltage controls such as reactive power compensator
and/or Static Var Compensator (SVC). The IBGs themselves can change the reactive power output
through the oversized inverter and/or through the reduction of active power output. Modern grid
codes such as VDE4120 in Germany have required the voltage control9 at the IBG’s PCC.
(6) Synchronization (torque) capability: The synchronous generators have the synchronizing torque
capability which is a very important factor for angle stability [11]. The synchronizing torque index,
Kij is proportional to the internal voltage of the synchronous generator and the equivalent
synchronous generators and/or the angle difference between the synchronous generators and the
equivalent synchronous generator (See Equation 2.1). Such generators can automatically change
their active power output so as to mitigate the change in the angle difference. It is noted that this
capability does not denote the ability which tells how the IBGs in general capture the voltage angle
through a Phase Locked Loop (PLL) algorithm in order to output the active power and reactive
power in a correct phasor form. This capability reacts not to the voltage angle itself but the angle
difference between two different points in order to contain such angle difference within 180 degrees.
This ability is one of the important contributions especially for rotor angle stability.
It should be emphasized that this capability is not literally required for IBGs because they have no
rotor-angle stability issue. On the other hand, the IBGs might be required to have the synchronizing
torque capability in the future although IBGs do not need to be synchronized. In such a case, this
is not easy to be achieved because the communication infrastructure for measuring the
aforementioned angle difference is basically required. Even such angle difference is assumed to
be measured nearly in real time, tremendous number of measurements are required in wide area,
because the equivalent synchronous generator to be measured for calculating the angle difference
is not always the same and significantly changes especially when a synchronous generator in a
network of equivalent synchronous generators is disconnected (See Figure 2.2).
dP d  Vi  V j  Vi  V j
K ij    sin  ij   cos  ij
d d  X  X
Equation 2.1
Where, Kij denotes synchronizing torque coefficient induced between generator i and j,
P denotes active power output of a generator
Vi denotes internal induced voltage of generator i,

7
Recent developments in power electronics allow for the inverters to operate in U-F mode (also called voltage-frequency mode
and grid-forming), giving them the ability to start and maintain the system frequency and voltage using the appropriate controls.
It should be noted that this mode of operation is not widely used and it requires inverters with increase capabilities. Moreover,
in general, the operation mode of IBGs typically cannot promptly change without stopping the inverter.
8
In some countries, the power factor has been set as a value less than 1 even for IBGs which are connected to LV networks.
9
For example, this voltage control is known as fast voltage control in German grid code, VDE4120.

23
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Vj denotes internal induced voltage of generator j,

X denotes reactance between internal induced voltages of generators i and j,
ij denotes angle difference between internal induced voltages of generators i and j.

(a) Before generator tripping of G3 (b) After generator tripping of G3

Figure 2.2 Example transition of synchronizing torque coefficient in case of a generator tripping.
(7) Loss of synchronism: Synchronous generators cannot avoid loss of synchronism when angle
stability cannot be maintained, while the IBGs do not have a rotor angle and keep synchronism
inherently. As mentioned earlier, IBGs in general capture the voltage angle through a PLL algorithm
in order to output the active power and reactive power in a correct phasor form. These
characteristics can be also treated as a sort of synchronization capability. IBGs are required to be
synchronised with the AC grid by PLL. The characteristics of these PLL algorithms, in particular
during system disturbances, might impact the inverter response. It is noted that the IBGs might also
lose synchronism i.e. might be disconnected due to the significant voltage dip, but do not have
transient stability problem.
(8) Damping torque capability (power oscillation damping: POD): Oscillations can be damped
when extra power is injected into the system in phase with the rotor speed deviation, which is
instantaneously decelerated, and/or when extra power is consumed in the system, which is
instantaneously accelerated. In real power systems, the damping power is obtained by the
modulation of load or generation for a period of time, typically in the range of 5 to 10 seconds.
This damping torque can be achieved in two ways 10. Inherently, synchronous generators have
short-circuited damper, or amortisseur windings, to help damp mechanical oscillations of the rotor
if the rotor speed deviates from synchronous speed, the flux will not be stationary with respect to
the rotor and currents will be induced in the damper windings. According to Lenz’s law, these
currents will oppose the flux change that has produced them and so help restore synchronous
speed and damp the rotor oscillations. Supplementary controls, called power system stabilizers
(PSS), can be used to further enhance the damping of local and inter-area modes of rotor oscillation
among generators [11].
In the case of IBGs, the modulation of active power dampens the oscillation directly, whereas
modulation of reactive power dampens indirectly by modulation of the system voltage and therefore
by modulation of the voltage dependent loads. The way this modulation achieves mitigation of
oscillations is by means of active and reactive power injection, which can be implemented in a POD
controller. The control scheme of the active power injection is the same as the PSS which is often
applied to large capacity synchronous generators. For example, an additional control loop could be
used to modulate the voltage at the PCC and to achieve the damping effect via the connected load
with the load voltage characteristics.
(9) Frequency control capability (primary, secondary and tertiary): Turbines directly linked to
synchronous generators can have primary, secondary and tertiary frequency control capabilities.
This capability strongly depends on the prime mover characteristics, not the generator. In order to
emulate those capabilities, the IBGs need to increase or decrease their active power output.
However, the energy sources connected to the grid via inverters are in the majority of cases not
controlled. IBGs can rather easily decrease their active power output but it is not easy for them to
increase their active power output. An option is to reduce their active power reference intentionally

10
Another proven means of damping such oscillations, is through the use of power oscillation dampers (POD) installed on
active power electronic devices such as SVCs and HVDC.

24
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

at steady state in order to ensure an adequate upper margin or headroom. This has obvious
economic implications, particularly for RESs where curtailing the IBG means essentially lost
opportunity costs that cannot be recovered since the energy source is variable. Another option is
to marry an energy storage option with the IBGs. It should be noted that this could cause additional
burden to the IBG owners. Also, it has to be considered that even for traditional power plants with
synchronous generators keeping additional generation margins available for regulation represents
an additional cost.
(10) Limited frequency sensitive mode: In the case of significant frequency rise, power plants need
to decrease their outputs. This emergency corrective action is called "Limited Frequency Sensitive
Mode – over frequency" It is important to note that any kind of generator can operate in the limited
frequency sensitivity mode, but their prime movers may not be able to provide this operating mode.
(For example, in the case of gas turbine power plants, the sudden decrease of fuel input will
increase the air-fuel rate and could cause the undesired/unintentional flame-out of the combustion
system.11). This capability strongly depends on the prime mover characteristics, not the generator.
For many IBGs, this is not a limitation. For example, PV can easily reduce its output if the inverter
is controlled according to this mode.
(11) Maintenance: The periodical maintenance for synchronous generators is more onerous as longer
down time and more expensive intervention is required than compared to IBGs
(12) FRT capability: Synchronous generators are required to withstand without failure a short circuit of
any kind at its terminals by IEC Standards (IEC 60034-3 Clause 4.16). On the other hand, the prime
movers do not always have the fault ride-through capability. While most of the representative prime
movers of large-scale hydro power plants, coal-fired power plants and nuclear power plants have
such capability, some prime movers of medium-scale thermal power plants have a shear pin
embedded in the rotating shaft and designed to break during severe voltage dip (when the shear
pin breaks during severe voltage dip, the power plant is tripped). To date, distributed IBG typically
do not ride through severe three phase faults because the voltage phase angle could not be
detected when the line voltage is very low, e.g. less than 30% because the magnetic contactor
which is placed between the inverter and the grid will open due to the loss of its excitation of the
magnetic coil. It is noted that this is the issue of the contactor and not of the inverter. New
techniques such as higher resolution frequency calculation, and the use of the off-delay release
type magnetic contactor or the UPS can now achieve the fault ride-through capability12.
(13) Reactive power support:
(a) V-Q control during steady-state: The rated power factor of the synchronous generators is
generally in the range of 0.80 - 0.95, with the higher value being typical of modern units. The rated
power factor for distributed IBGs is not often provided, which means the rated power factor is
assumed to be unity13. Distribution-connected PVs are still operated at unity power factor over their
entire active power output range in many countries. Most of these inverters are not sized to provide
any reactive current at full output. In order to provide reactive power support at full output, larger
inverters will be required. On the other hand, system operators usually require that IBGs include
reactive power control at the PCC. In addition, this reactive power control to be independently
activated by such multiple alternatives as voltage regulation, reactive power regulation or power
factor regulation14.
(b) Reactive current control during network incidents: A synchronous generator can
dynamically support the reactive power output from the moment when the system fault occurs thus
providing an immediate/instant increase of reactive power output. In contrast, it cannot be
guaranteed that the IBG can increase the reactive power output from the moment when the fault
occurs to mitigate the voltage drop mainly because the detection time of the voltage magnitude of

11
As the countermeasure, there is typically a rate limit on how quickly a gas turbine can reduce its output.
12
Due to the advanced technologies, most large scale (many tens of MWs) utility scale wind power plants connected to the
transmission system do have FRT capability and some WTGs are able to ride through solid faults up to 3 s, which is most
unlikely to be achieved by synchronous generators from mechanical point of view and from transient stability point of view.
13
Nowadays, the power factor in some countries has been set as a value less than 1 even for IBGs which are connected to LV
networks.
14
Large scale IBGs do provide these same capabilities, i.e. voltage regulation, however, this requires more work and research
to identify the proper and suitable means of providing such capabilities at the distribution level. One such effort is the current
revision of the IEEE Standard 1547 in North America.

25
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

the IBG cannot physically be instantaneous. The IBG enables an increase in the reactive power
output with some delay during the fault by decreasing the active power output within the rated
current. Therefore, no matter how quickly the IBG is able to control voltage and current, the IBG
cannot show the same immediate/instant increase of the reactive power support provided by the
synchronous generator. In addition, a synchronous generator allows a negative sequence current
to flow, whereas the IBG is often designed to block negative sequence currents. In the future a
TSO may require that the IBG provides a negative sequence current in case of unsymmetrical faults
mainly to ensure sufficient voltage recovery for all three phases 15.
(14) Harmonic emission: Inverters may produce non-sinusoidal currents that can be described and
quantified as harmonic emission in frequency domain. The harmonic emissions need to be
assessed and controlled before the connection is permitted. The IEC has standards of harmonic
emissions and some countries impose their own limits for connection of nonlinear appliances to the
grid. Harmonic currents emitted by synchronous generators (airgap flux harmonics, slot harmonics,
etc.) are usually negligible. The harmonic current emission of IBGs depends on the following; type
of technology used, control strategy of the DC/AC-inverter, existence of high- or low-frequency
coupling transformer and the harmonic voltages prevailing in the AC-power system.
(15) Harmonic voltage reduction: Since the effective impedance of synchronous generators for low
order harmonics is based on the small sub-transient reactance, synchronous generators provide a
rather low impedance path for harmonic currents and thus tend to reduce harmonic voltages. All
voltage source converters absorb harmonics because inverters can act as an impedance using the
voltage source converter technologies.
(16) Black start: is the ability of the power system to restart itself after a full or partial system black out.
Most conventional generators are designed to require an electrical supply from the power system
to start up. Normally this is provided from the transmission or distribution system, however under
black start this supply is not available. Therefore, to restart the system it is required to some power
stations have their own auxiliary supplies to they can restart themselves. These power stations can
then be used to restart other power stations and thus the whole system can be restarted.
Traditionally black start capability relies on large transmission-connected synchronous generators.
Over the coming years the trend of reduction in the number of these plants is expected to continue
leading to fewer traditional black start providers being available. For an IBG the ability of the
technology to achieve black start is more limited under extreme network conditions due to factors
such less inertia, less overload capacity to provide inrush current for energization and the use of
PLL technology [12].
Most of the power systems around the world are undergoing fundamental changes. This includes strong
moves away from heavy reliance on fossil fuels as the primary energy source mainly provided by large
synchronous generators connected to the transmission systems, towards a decarbonised future supply
relying increasingly on variable renewable energy sources (RES) using non-synchronous generation
predominantly connected to the network via power electronics and extensively connected deeply
embedded in the distribution networks. Some countries in Europe have already experienced times in
which in some periods the national demand for electricity has been exceeded by the RES production
alone [13] That means the functionalities which the conventional generators have and which the IBGs
do not have, will be lost and the system stability could be affected. In order to cope with this, such
functionalities have been required by the IBGs through updating of grid codes. It should be noted that
the aforementioned advanced functionalities and capabilities could usually be require to newly installed
IBGs.
In general, if equipped with proper control logics, the IBGs could offer to the grid many flexible features,
like:

 Frequency regulation
 Reactive power/voltage regulation
 Insensitivity/immunity to large electrical torque variation (i.e. due to automatic reclosing near the
plant)
 Low Voltage Ride-Through (LVRT)

15
This is now regulated in Germany (See VDE-AR-4120).

26
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 2-1 summarizes the main existing and/or potential differences between IBGs and conventional
synchronous generators.
In terms of simulation models for IBGs the following challenges and requirements can be highlighted:
 Recently a lot of new capabilities (with reference to IBGs, because many of these were common
for traditional synchronous generators) have been required for IBGs in grid codes, some of them
are still at the definition stage, to comply with both DSOs and TSOs needs. Those capabilities have
to be represented in each model (at least according to the specific simulation to be performed).
 Specific capabilities are already available on the market (e.g. simulation of inertia), even if obtained
by additional devices (storage). However, they are not described in detail in any standard approach
in terms of algorithms, performance, implementation, compliance assessments, etc., making it
difficult to develop appropriate and generic models.
 From a “model definition” perspective, it is very important to be aware that IBGs have no “natural”
features (because the feature can depend on how the IBG controller is designed), while the
synchronous generators have these.
 The scope of application (area of validity) of any given model has to be defined. For example, a
specific capability (frequency support, voltage support etc.) might be represented in a Root Mean
Square (RMS) model for stability analysis, but the model is only valid under certain conditions (e.g.
no smaller than the minimum short circuit ratio).

27
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 2.1 Major existing and/or potential differences between IBGs and synchronous generators
Relevant Conventional IBG with minimum Advanced capability /
phenomena synchronous functionalities Advanced feasibility of
generator with (before IBG
standard AVR and interconnection (after interconnection
turbine governor requirements are requirements are
evolved) evolved)
Rotating Frequency Yes No Yes (prime mover
mass/inertia Stability dependent), but enough
headroom (or unloaded
synchronized capacity) is
required (The use of
battery energy storage
could be needed
depending on the type of
devices). The capability
might also depend on the
direction of frequency
deviation (over- or
under-frequency). In
addition, exact emulation
cannot be performed16.
See (1) above.
Frequency Frequency Yes* No Yes (prime mover
response Stability dependent), but enough
capability headroom and/or upper
(primary, margin need to be
secondary and ensured. See (9) above.
tertiary)
Limited Frequency Yes** No Yes (prime mover
frequency Stability (over- dependent). See (10)
sensitive mode frequency) above.
Constant Voltage Yes, internal No, if connected to Yes, but isolated system
voltage source17 stability induced voltage the grid (inverter is is required. Stiff voltage
synchronized to and stiff frequency (U-F
external grid mode) are required for
frequency/phase) inverter). Oversized IBG
may be required. See (4)
above.
Transmission- Voltage Yes, but large No Yes (large-scale IBGs
level voltage Stability capacity machines only) often with large
support (steady with AVR only capacity reactive power
state) compensators such as
shunt capacitor/reactor,
SVC. See (5) above.
Reactive power Voltage Yes, according to No Yes, but larger IBG is
support (V-Q Stability PQ-capability required or active power
control during /support needs to be reduced
steady state) according to PQ
capability characteristics.
See (13) above.

16
None-the-less, studies have been shown that so-called “synthetic inertia” for wind turbine generators can be an effective
means of helping to reduce system ROCOF during frequency events. See https://www.nerl.gov/grid/wwsis.html
17
The voltage denotes the synchronous internal voltage. It does not denote the terminal voltage of synchronous generators.

28
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Relevant Conventional IBG with minimum Advanced capability /

phenomena synchronous functionalities Advanced feasibility of
generator with (before IBG
standard AVR and interconnection (after interconnection
turbine governor requirements are requirements are
evolved) evolved)
Reactive power Rotor angle Yes No Yes, usually during faults
support stability IBGs may be able to
(reactive Voltage provide a reactive
current control Stability current injection with
during network some delay which helps
incidents) keeping them grid
connected, as per
TSO/DSO requirements.
See (13) above.
Synchronization Rotor angle Yes No Yes (but almost
(torque) stability infeasible), angle
capability18 difference needs to be
observed without time
delay. See (6) above.
Damping torque Rotor angle Yes, damper No Yes, POD functionality.
capability stability windings and See (8) above.
(power addition of PSS
oscillation
damping
capability)
Loss of Rotor angle Yes Not applicable to See (7) above.
synchronism (transient) IBGs
stability
Protection
FRT capability Rotor angle Yes*** No. Yes (prime mover
(transient) dependent). See (12)
stability above.
Frequency
stability
Harmonic Power quality Low Yes, including the See (14) above.
emission frequency band
dedicated to utilities
for PLC based
communications
Harmonic Power quality Yes, for low order Yes Yes. See (15) above.
voltage harmonics
reduction

Short-circuit Protection, Yes No Yes, but contribution is

contribution in limit voltage limited to around 1 p.u.,
case of decline unless inverter has short-
symmetrical term rating to exceed 1
and p.u.). See (2) above.
unsymmetrical
faults

18
Note that this capability is NOT literally required for IBGs because they have no rotor-angle stability issue.

29
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Relevant Conventional IBG with minimum Advanced capability /

phenomena synchronous functionalities Advanced feasibility of
generator with (before IBG
standard AVR and interconnection (after interconnection
turbine governor requirements are requirements are
evolved) evolved)
Control Voltage and Fast, depending Inverter itself very See (3) above.
response frequency on the time fast (faster than
capability stability constants involved generator controllers
or other external of synchronous
dependencies**** generators),
limitations possible
due to measurement
delay, prime mover
or other external
dependencies****
Overload misc. Yes Limited (nearly Yes, but significant
capability (up to negligible), oversized IBGs are
few seconds) depending on required. See (2) above.
semiconductor
devices*****
Maintenance misc. regularly Low for inverter See (11) above.
itself, prime mover
depending on the
type
* Generator with enough headroom and/or upper margin.
** Depending on rotating prime mover characteristics.
*** Generators with damper windings are for local modes of oscillations, while generators with the PSS are
for inter-area modes of oscillations.
**** E.g. industrial process supplying primary energy to be converted in electricity
***** Air-cooled IGBT converters have pretty good short-term overload capability (E.g. up to 1s), while water
cooled MV-connected converters have almost no short-term overload capability.

2.2 INVERTER CHARACTERISTICS

2.2.1 Topology and controller structure
There is a variety of inverter topologies available on the market, developed by different manufacturers
and optimised for specific applications. Optimisation targets generally consider reliability, maintenance,
losses, overall costs etc. For example, 3-phase connected wind-turbines may be equipped with 2 level
voltage sourced back-to-back inverters or multilevel voltage sourced back-to-back inverters, depending
on the wind-turbine rated power.
Using different primary energy sources, the inverter topology remains substantially the same even using
different energy sources. The following is an example of the topology for a PV plant. As regards PV
inverters, among different possible topologies [14], two typical inverter families are:
 H-bridge topology
 Neutral point clamped topology
Most of the inverter topologies for PV available today have been derived from these basic design
structures. It is not the aim of this TB to describe all inverter topologies and their specifics for the different
energy resource. The principal scheme of a photovoltaic power plant with H-bridge as a representative
example is described in Figure 2-3 (a)
The PV plant consists of the following elements [14]:
 Photovoltaic array: The array of photovoltaic modules forms the solar generator. Several modules
can be connected in series, forming a so-called string. Strings in turn can be connected in parallel.
It is possible to connect either a common inverter to the entire array, with an inverter per string or
to connect a single inverter for each module.
 Chopper: Boost DC voltage, among others, to ensure Maximum Power Point Tracking (MPPT),
measuring the PV voltage (VPV) and PV current (IPV)

30
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 DC link: Filter/Energy storage

 Inverter: Conversion of DC power to AC power with the output filter (denoted as LCL filter in Figure
2-3 (a)).

PWM Vdc PWM

IPV Ig

VPV Vg

(a) Simplified PV block scheme with different control layers [14]

Chopper Inverter
iin
Filter
iac
Smoothing capacitor

vdc
vin vg Grid
PV Array

Valve device control Control

vin iin vdc iac vg
DC-ACR Circuit

IDC-cmd Comparator
IAC-cmd
VDC-ref Amplify Reference
MPPT DC-AVR voltage PLL
IDC-ref waveform

(b) Example of control circuit of PV [16]

Figure 2.3 General control circuit structure of PV.
The inverter control enables the AC (active) current waveform to match with the voltage at the PCC in
terms of the frequency and the phase at unity power factor. The inner current control monitors the
voltage at the PCC and creates the current waveform with the same frequency and the same phase of
the voltage at the PCC with the aid of the PLL. At the DC side, the Maximum Power Point Tracking
(MPPT) control maximizes the active power output. As shown in Figure. 2-3 (b), DC-ACR monitors input
active power and changes the input DC current periodically in order to control the input DC current via

31
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

controlling the switching action of the chopper. The DC-AVR monitors the internal DC voltage and
amplifies the reference voltage waveform (created via PLL) which results in the AC (Active) current
control signal. The power electronic device switching action is controlled by comparing the
aforementioned AC current control signal with the measured AC current.

For most of the phenomena described in this technical brochure, the modelling of the switching action
is generally not necessary. An average model, i.e. a model where all the variables are averaged over
the switching period, is able to precisely reproduce most of the phenomena reported in this chapter.
Only a few of them, such as the prediction of harmonic distortion and electromagnetic interference (EMI)
compliance, require a more precise switching model.

2.2.2 Ancillary functions

Because of the flexibility of the inverter control design, IBGs may be required from either the technical
standards and/or from grid codes, to provide some additional capabilities for grid support, among them:
 Negative sequence current injection; (ref 4.7.4.1.2) [15]
 Reactive current control calculated by mean of power factor input;
 Maximum reactive current injection;
 Reactive current level depending on voltage depth (ref 4.7.4.1.1) [15]
Table 2-2 shows the most relevant required IBGs capabilities from the EU Regulation 2016/631
“establishing a network code on requirements for grid connection of generators” [17] and the IEEE 1547
“Standard for Interconnecting Distributed Resources with Electric Power Systems” [18]. As IEEE
1547 and UL1741, “Standard for Inverters, Converters, Controllers and Interconnection System
Equipment for use with Distributed Energy Resources” [19] have been evolved, the latest requirements
coming from IEEE 1547-2018 are also introduced in this table.
Table 2.2 Requirement for IBG in grid code and standard
Requirement EU 2016/631 IEEE 1547-2014 IEEE 1547-2018
P(f) (over/under) x x
Voltage control by means of reactive power (Q(V)) x (x) x
Voltage control by means of active power (P(V)) x
Synthetic Inertia (x)
ROCOF immunity x x
FRT (LV/HV) x (LVRT only) (x) x
Inverter internal protections
Anti-islanding active detection methods x (ROCOF) x x
Dynamic voltage support during faults and voltage
(x) (x)
steps
Power Oscillation Damping (x)
Black start Capability (x)
Capability of Islanding Operation x
Automatic Disconnection with Abnormal Voltage x x
Automatic Connection with active power recovery
x x
speed
Constant Power at Low Voltage x
Constant Power at Low Frequency x
Note: The sign “x” means one or more classes/categories of the IBGs are required to meet with the designated requirement.
( ) denotes a non-mandatory requirement.

2.2.3 Inverter protection

An Inverter’s protection may be separated into two main classes, internal and external. This
classification has nothing to do with the physical location of the protections. Internal protections are
primarily to assure the safety of the inverter itself and may be not in accordance with relevant standards
of protection relays (IEC TC 95 “Measuring relays and protection equipment” [20]) and are applied by
the manufacturers. External protection is required to serve a different purpose and considers the network.
Physically, in some cases, the external protection may be the same as the internal inverter protection,
but, despite this, they are not “monitoring” the inverter (internal), but the network (external). Although
protection functions based on pure electric measurements (passive protection methods) shall comply

32
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

with requirements of IEC TC 95, other protections (not mentioned and/or based on active methods) are
not defined at all by IEC TC 95 at the present moment.
Internal protections are generally inserted in inverter models and do not affect IBG capabilities and
requirements. External protections, despite that they are physically inside the inverter control, may be
modelled separately, to allow for changes in the protection models or different regulation combinations
without any change in inverter model.
2.2.3.1 Internal Protection
Each IBG type has its own type of internal protections focused on avoiding damage to the inverter itself.
These internal protections are also known as generator protections (i.e. nothing to do with Interface
Protection).
Some examples of inverter internal protections are:
 Reduction of maximum inverter current when the DC voltage exceeds a certain limit;
 Limitation of inverter current’s variation rate after a fault;
 Limitation of total reactive current;
 Manual PV field shutdown with emergency stop;
 PV field insulation detection;
 DC overcurrent protection;
 Over/under voltage protection;
 Over/under frequency protection.
It should be noted that “Limitation of inverter current’s variation rate after a fault” and “Limitation of total
reactive current” are generally categorized as control instead of protection. Because their control
functions can operate for protection purposes as well as for control purposes, they are treated as the
internal protection in this technical brochure.
2.2.3.2 External Protection
IBGs may have external protections to:
 Detect uncontrolled local islanding situations and disconnect generators to shut down this island.
This functionally is also known as “Loss of Mains Protection”;
 Reduce the power production from the generating plant to prevent an over-voltage or over-
frequency situation in the network it is connected to;
 Assist the power system to reach a controlled state in case of voltage or frequency deviations
beyond corresponding regulation values.
These protections (or combination of different elementary protection functions) are usually referred to
as interface protection or interface protection system.
The interface protection system is generally based on combinations of over/under voltage and
over/under frequency protections.
It is not the purpose of the interface protection system to:
 Disconnect the generating plant from the network in case of faults internal to the power generating
plant. Protection against internal plant faults or abnormal operation conditions (short-circuits, earth
faults, overloads, etc.) is in charge of other external protection relays coordinated with network
protection, according to electric system operator protection criteria. Protection against electric
shock and against fire hazards are out of the scope of this TB;
 Prevent damages to the generating unit due to incidents (e.g. short circuits, asynchronous reclosing
operations) on the network. To avoid these possible damages, the generating unit shall have an
appropriate immunity level.
The interface protection system has to be coordinated with power system protections and power system
needs. Generator disconnection shall happen as fast as possible and with high reliability in case of local
faults or outside of normal operation conditions, and shall not happen in case of global perturbations
unless voltage and frequency values are far from normal operation conditions for a relatively long time.
The type of protection and the sensitivity and operating times of the interface protection system depend
on the electric system protections and on the characteristics of the network.

33
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The interface protection system is meant as a dedicated external device. Protection functions and other
features (e.g. EMC, mechanical, climatic requirements, etc.) shall be according to IEC TC 95 and other
relevant standards.
A good overview on external protection can be found in CIGRE TB 613 “Protection of Distribution
Systems with Distributed Energy Resources”[21] and CIGRE TB 421 “The impact of renewable energy
sources and distributed generation on substation protection automation” [22].
2.3 PRIMARY ENERGY SOURCES OF INVERTER BASED GENERATORS
Renewable Energy Sources (RES) are mainly connected to the grid via inverters. A power inverter, or
inverter, is an electric energy converter that converts direct current (DC) to single-phase or polyphase
alternating current (AC) [23]. These IBGs represent 100% of the total for the Photovoltaic (PV) plants
and an appreciable and continuously increasing percentage of wind plants (full inverter generators and
Doubly Fed Induction Generators (DFIG) reach sizes much higher than 4 MW). In addition, inverters
are also used in Organic Rankine Cycle (ORC) plants [24] and in micro turbines equipped with high
frequency permanent magnet synchronous generators. It is noted that the doubly fed technology has
originally been developed for hydro plants, e.g. variable speed pumped storage units [25]. The inverter
provides the interface between the grid and primary energy source to be transformed into electricity.
Although the inverter technologies may be similar to all devices, an appreciable difference may exist
related to the prime mover features, therefore influencing at least the inverter control. The response
and achievable performance of the combined system depends both on the capability of the inverter and
the capability of the primary energy source.
Examples:
 PV plants (PV array) neither have physical inertia nor mechanical/thermal processes involved.
Therefore nearly “real time” regulation is possible, limited only by inverter capabilities and inverter
control reaction time (for the whole chain, including measurement time of relevant quantities, such
as voltage, frequency, etc.; the theoretical reaction times may be some milliseconds or shorter).
There is no inherent energy storage (due to missing inertia) and thus, no possibility to support the
system in case of under-frequency due to incentive systems (unless additional storage devices are
foreseen or unless the generation is curtailed by several percent of the available active power [26]).
 The prime mover of wind turbines (rotor blades) exhibits mechanical effects, including for instance
inertia. The inverter control has to take into account the dynamics of the prime mover. For example,
in case that such power cannot be delivered to the grid (e.g. during a fault), even though pitch
control may be very fast, some energy may be injected in the DC stage of the inverter causing an
increase of the DC voltage (possibly limited by the activation of a chopper or of electronic operated
short time duty resistors)
 In case of ORC generators (not widely used at present), the capability and the reaction time
depends on the thermal cycle of the “prime mover” from the ORC itself, which is usually an industrial
process, thus it is not easily controlled in a rapid manner without serious consequences to the main
process itself.
 As stated before, some of the RES capabilities might not be used due to existing economic
incentives (or the lack thereof): RES is mostly allowed to operate at maximum momentary available
power supplied from the primary energy source (there are some exceptions at present, for instance
Eirgrid, Hydro Quebec, ERCOT (for those IBGs participating in the primary frequency response
market), etc. have already introduced the requirement to operate RES at a power some % below
the available maximum power, to allow for primary frequency regulation). Moreover, the power
factor is often equal to one, the reactive power exchange with the network for voltage control on
distribution networks and power flows on transmission system, not being generally requested, (the
main reason for this is to avoid any oversize of the inverter or any momentary active power
reduction to allow the Q exchange)
The generation units that use different technologies can be characterized according to their primary
energy source and the existence, or not, of a prime mover and a rotating electrical machine as a
generator [27], [28], as shown in Table 2-3 and Figure 2-4. It is noted that the generator protection
somehow depends on the type of primary source (See Appendix 3-A.2.3).

34
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 2.3 Classification of inverter-based generator technologies

Prime mover Rotating machine as
Technology Primary source
(rotating) electrical generator
Wind energy conversion system Wind Yes Yes
Micro-turbine Diesel or gas Yes Yes
ORC (based or not on micro-turbines) Waste heat Yes Yes
Fuel-cells Hydrogen No No (Yes optional)
Photovoltaics Sun No No
Superconducting magnetic energy storage Storage No No
Battery energy systems Storage No No
Flywheel Storage Yes Yes
Variable speed hydro plants (VSC or cyclo
Water Yes Yes
converter)

Figure 2.4 Main components of generic inverter-based generator.

An example of a typical structure of a power source with a converter interface is shown in Figure 2-5.
The energy source may be: (a) a DC-power source by itself or (b) an AC source, which is rectified into
DC. In any case, the source itself may include other power electronics converters (AC/DC and DC/DC),
in order to create and/or regulate the DC voltage or current [27].

Fuel Cell
or
Battery
DC = Grid
or
PV
a.

Gen
DC = Grid

b.
=

Gen Grid

=
c.
=
Figure 2.5 PV systems, battery systems or fuel cells (a) micro-turbine or full converter WTG (b)
and DFIG (c).
Contrary to the modelling of the conventional generation sources, such as thermal or hydro generators,
where dynamics of generators play a very important role, for IBGs, the electrical control model
(generator and electronic interface) is of vital importance. The inverter serves as an interface between
the energy source and the electricity network. This electrical control model primarily determines the
dynamic performance of the IBG. Thus, from the system analysis point of view, the primary source and
its controls are often neglected.
However, the primary energy source and the prime mover influence the capabilities of the IBG. The type
of the generator imposes some constraints on the electrical controls available and the generator
capabilities. For instance, regarding frequency response, the inverter can respond very quickly but its
primary source might be slower to follow, requiring a ramped change and thus constraining the mid- and
long-term behaviour.

35
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

When studying the short-term, transient and dynamic performance of the inverters at the grid terminals
or when the generator is interfaced to the grid with the auxiliary help of a storage system to supply the
transient need of power, the details on the primary source behaviour tend to lose importance. We can
regard the system as an inverter connected to a stiff DC source which simplifies the design and analysis
of the power electronic interface. However, the controls, features and capabilities of the generator’s
inverter following disturbances strongly rely on the type of its primary source.
PV plants that are fully interfaced by power electronic converters and have the ability to provide fast
response during frequency rise/drop have no rotating masses that introduce increased time responses.
The loss of stabilization effect due to the lack of inertia could be compensated from a much faster
reaction time. The fast reaction time may also have some adverse effects, which must be properly taken
into consideration in the dynamic studies, because these effects may affect different parts or levels of
the electric system from the one under consideration (for instance may increase uncontrolled islanding
on distribution systems).
Simple micro-turbines consist of a compressor, a combustor, a turbine, and a generator. The
compressors and turbines are typically radial flow designs that resemble automotive engine
turbochargers. Most designs are single-shaft and use a high-speed, permanent magnet generator to
produce variable-voltage, variable-frequency AC power.
Battery energy storages may heavily affect dynamic system behaviour according to the way they are
used [29].
For further detail, it is necessary to consider primary energy sources using inverters/converters to
interface with the network, but it is noted that providing the primary source model is out of the scope of
this TB. IBGs shall be considered as two components: inverter and generator which both have
independent requirements for frequency control and different frequency response characteristics.
2.4 CONCLUSION
Chapter 2 addressed the characteristics of IBGs focusing on the differences between IBGs with
minimum functionalities and large capacity synchronous generators. The representative natural feature
of synchronous generators is the inertia; fault current provision, synchronization capability and the
constant/fixed internal voltage19 source. They cannot easily be emulated (if at all or if needed) by IBGs
from technical or commercial perspectives20. On the other hand, many of the characteristics such as
limited frequency control capability and the reactive power control capability can be provided by IBGs.
Because of the increasing functionalities of IBGs, the IBG models have been further developed.
Although the synthetic inertia concept has emerged over the last few years, it is not completely
equivalent to inertia provided by synchronous machines mainly because the immediate/instant change
in active power is feasible by only synchronous generators. However, the control response of IBGs is
much faster than that of synchronous generators and such fast frequency response might be able to
compensate such lack of immediate/instant change in active power.
Chapter 2 also addressed the difference from a protection point of view, of the characteristics of IBGs
compared to synchronous generators. In general, IBGs are more likely to be disconnected due to the
high sensitivity of inverter protections. Therefore, modern grid codes are asking for IBGs to have FRT
capabilities. Because the operation of the inverter protection could result in the disconnection of the IBG,
the inverter protection models play an important role for most of the dynamic stability analyses. On the
other hand, the primary source and its controls may often be neglected for power system dynamic
studies.
Chapter 2 introduces in a basic fashion the type of models which is used for dynamic power system
analysis. The selection of the model type (EMT or RMS) is not discussed in Chapter 2 because it is very
dependent on the specific phenomena to be investigated. The selection of the model type with the
necessary model element for each type of phenomenon is discussed in Sub-Chapter 3.8. The addressed
characteristics in Chapter 2 are then used to extract the necessary functionalities of the IBG model
components which can be classified into three categories: 1) Control, 2) Protection and 3) Capability
(See Table 2.4). The definition of those categories will be provided in Chapter 3 (See also Appendix
A.2).

19
The voltage denotes the synchronous internal voltage. It does not denote the terminal voltage of synchronous generators.
20
Note that this capability is NOT literally required for IBGs because they have no rotor-angle stability issue.

36
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 2.4 List of functionalities of IBGs

Category Functionalities
Control DC Source Control
Control Current Control
Control PLL
Control MPPT
Protection Reduction of maximum inverter current when the DC voltage overcome a certain limit
Protection Limitation of inverter current’s variation rate after a fault
Protection Current limit
Protection DC Overvoltage Protection
Protection Over voltage/Under voltage Protection
Protection Over frequency/Under frequency Protection
Protection Protection for Detecting Balanced Fault
Protection Protection for Detecting Unbalanced Short-Circuit Fault
Protection Protection for Detecting Single-line-to-ground Fault
ROCOF tripping: monitoring the power frequency variation rate and disconnecting the
Protection
inverter when it reaches a certain limit [Hz/s]
Protection Vector jump
Protection Transfer trip
Protection Anti-islanding active detection method
Capability P(f) control (over/under frequency)
Capability Voltage control by means of reactive power
Capability Voltage control by means of active power [P(V)]
Capability Synthetic Inertia
Capability ROCOF immunity
Capability FRT (LV/HV)
Capability Active behaviour during voltage fast variations
Capability Power oscillation damping (POD)

37
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

38
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

3. NECESSARY FUNCTIONALITIES OF INVERTER BASED

GENERATORS FOR KEY PHENOMENA
3.1 INTRODUCTION
The list of functionalities of inverter-based generators (IBGs) has been described in Chapter 2. It should
be noted that those functionalities were developed in terms of the capability of the inverter and the
control and protections of the IBGs. That means, those functionalities were neither developed from
system phenomena perspectives nor modelling perspectives (RMS or EMT). Illustrating and ranking the
necessary functions for each type of phenomenon becomes important for many power system
studies/analysis because considering all functions in any study requires an extremely heavy
computation burden. Therefore, Chapter 3 lists the relevant phenomena simulated in power system
dynamic studies and provides the functions which should be considered for each type of phenomenon.
The list of the phenomena and the mapping to the type of studies are illustrated in Table 3.1. It is noted
that the type of studies shown in Table 3.1 are based on [30].

Table 3.1 Type of phenomena and type of studies

Sub-Chapter Type of Phenomena Relevant Key Words Type of Studies
3.2 Behaviour in response to ●Device protection ●Frequency regulation
frequency deviations ●System support ●Frequency stability
●Plant level control ●Transient stability
●Synthetic inertia
●Frequency Response
3.3 Behaviour in response to large ●Device protection against damage ●Short-term voltage
voltage deviations ●FRT capability stability
●Grid support ●Transient stability
● (Synthetic inertia) ●Provision of fault
current
●Low/High voltage ride
through
3.4 Behaviour in response to long- ●V/Q control Long-term voltage
term voltage deviations ●<permanent> limits stability
●Plant level control
3.5 Modelling simplifications for Small-disturbance
small-disturbance stability stability
analysis
3.6 Unintentional islanding Unintentional islanding
operation detection
3.7 Other phenomena and studies Low- and high-frequency interaction
of controller, switching transients

Many impact analysis studies of IBGs have been performed around the world [31]-[35] and TB 450 [36],
“Grid integration of wind generation” which introduces how WTGs could influence the system
performance for each type of power system studies. However, according to the questionnaire survey
(which was performed during this CIGRE/CIRED JWG activities), still over 30% of the utilities and the
system operators do not use any IBG models and just rely on the negative load model for their regular
power system dynamic stability studies. The reasons for this may be summarized as follows [4]:
 Lack of model requirements of IBG for specific power system phenomena:
As the penetration level of such IBG technologies increases, various aspects of power system
stability and dynamic performance in the grid may change. Therefore, requirements that address
the necessary functions that need to be modelled of IBG for specific power system phenomena
need to be developed. These functions include various aspects, such as control, protection and
the capability of IBG. Considering these requirements, utilities and system operators can select
specific models for each power system phenomenon.
 Lack of well-validated detailed IBG models:

39
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

In recent years, there has been much effort in the development of validated models for IBG.
This work has been primarily related to wind generation but has extended to other types of IBGs
in North America. Now, further attention is starting to be devoted to PV systems and other
technologies world-wide. In North America, there is one set of approved generic models for
large utility scale IBGs that is starting to gain traction [37], however, in general, there is still a
lack of well-validated and generally accepted dynamic simulation models, particularly for
distributed PV systems, for the use in power system dynamic studies. Even in the case of large
utility scale generic IBG models, there is a continued effort to add more features and to refine
them.
 Lack of widely accepted generic IBG models:
Usually utilities and system operators do not create their own (user-written) models. They
request validated models from manufacturers, either proprietary or adjusted generic models.
This request poses two main disadvantages: 1) the manufacturer wants to keep the
confidentiality of their proprietary user-written model; and 2) the extra effort for the manufacturer
to tune the parameters of the generic model which includes the validation of the simulations
against the field measurements. Thus, the importance of developing reliable and flexible generic
models for different technologies and manufacturers of IBG should not be underestimated. The
advantages of generic models include: vendor and manufacturer independent, grid code
compatible, public model structure (control block diagram), software simulation tool
independent, etc. For some technologies, like wind generation, these models are already being
widely used, however, the latest generic models had only recently been developed at the time
the questionnaire survey was conducted.
 Lack of widely accepted range of IBG model parameters:
Even if widely accepted generic models are provided, the control model parameters are crucial
for power system dynamic studies. Because many grid codes do not yet define the detailed
specification/characteristics of the inverter control, the control model parameters could be
different depending on the manufacturer of the inverter. Even if the control model parameters of
one inverter can be identified through validation, it is almost impossible to identify the
parameters of all inverters connected to the power system. Therefore, a set of realistic control
model parameters need to be provided.
 Lack of specific grid code requirements:
Due to the lack of grid code requirements in the past which specified detailed control
functionalities for the IBG, the approach of using negative load models for power system
dynamic studies was justified. However, with the development of new grid codes, and high
penetration of IBG, certain functions of IBG are required (e.g. voltage control, frequency
response etc.) and therefore, the negative load model is no longer adequate.
 Lack of information about dynamic performance of power system with IBGs:
The aforementioned increased penetration level of IBGs also makes system operation, both for
TSOs and DSOs, more challenging than in the past. Already, in some areas the consumers’
demand is mostly covered by generation which is connected directly to the distribution system.
TSOs routinely run time-domain simulations to assess the stability of the power system. Models,
which are currently used to represent distribution systems, are only based on a limited amount
of information, generally related to the high voltage network.
 Lack of agreed or well documented methodology for the aggregation of IBG:
Present trends towards the integration of an increasing range of IBG technologies, widely
differing in size and number, poses serious concerns in the industry on how to represent these
new technologies in power system dynamic studies. There is not only a lack of validated
dynamic computer models of individual distributed generating technologies, such as distributed
PV systems, fuel cells, micro turbines etc., but also there is no agreed methodology on how to
represent or aggregate the enormous number of embedded LV distributed generation for power
system dynamic studies, focusing on both, local (distribution level) and widespread
(transmission level) studies.
The high penetration level of IBGs has resulted in the displacement of conventional synchronous
generators. Therefore, the impact of IBGs on the dynamic performance of the system increases. The
dynamic characteristic of an IBG is different compared to synchronous generators, and with proper
control system design and functionalities of modern IBG technologies, they can provide many of the
same or even better services (e.g. voltage control, frequency response etc.) than a synchronous
generator. However, they do need to be modelled differently and correctly. Therefore, the development

40
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

of the appropriate computer simulation models for IBGs with such additional functions is vital for power
system analyses.
DSOs have, to some extent, a representation of their networks and details about connected consumers
and producers. The limited system and network data is generally not suitable for dynamic simulations
for either the distribution or the transmission system or, at least, has not been used for that purpose in
the past, due to the high level of detail required. From the point of view of the DSOs, time-domain
simulations may also now be necessary to assess protection system behaviour, distribution network
automated operation, unintentional islanding of part of distribution systems including IBG, voltage
issues, etc. For these types of power system dynamic studies detailed IBG models are needed.
Therefore, the necessity of IBG models should be clarified for each type of power system dynamic study.
Although many key descriptions shown in TB 450 are cited, Chapter 3 focuses more on providing the
dynamic response of IBGs to each phenomenon (instead of each type of study) and the fundamental
impact on the system performance, which is the distinctive contribution of this JWG.

3.2 BEHAVIOUR IN RESPONSE TO FREQUENCY DEVIATIONS

3.2.1 Description of phenomena
Frequency stability refers to the power system’s ability to cope with the imbalance between generation
and load. Short-term phenomena in the range of a few seconds and below is related especially to the
ROCOF, inertia (including synthetic inertia) and fast frequency regulation response. Long-term
phenomena with a time frame up to several minutes includes frequency recovery.
Frequency control is the capability of a power generating module – direct or inverter connected - to
control the active power output in order to maintain the system frequency.
Frequency deviation in large, interconnected systems and in islanded systems could be significantly
different. The latter might arise as a frequency containment issue or a frequency stability issue even in
normal operating condition in autonomous micro grids or during conditions outside of normal operation
(i.e. grid restoration after black out).
The frequency stability phenomenon is gaining importance with large penetration of dispersed
potentially RES, including IBGs. With high penetration of IBGs, the system inertia tends to reduce,
resulting in larger frequency excursions and higher ROCOF [31], if the low inertia is not compensated
by fast response times. The increase in ROCOF causes a lower frequency nadir, which could result in
the operation of the under-frequency relay and a lower settled frequency which would result in the lower
frequency containment. Thus, this leads to the new requirements to frequency response of IBGs.

3.2.2 Relationship between behaviour in response to frequency deviation to relevant study

3.2.2.1 Frequency stability
The frequency stability within a power system represents the capability of that particular power system
to maintain the frequency within the predefined limits by assuring the equilibrium between power
generation and power consumption. The frequency stability issue can be split into two main categories:
medium-term frequency stability and long-term frequency stability. The medium-term frequency stability
is often characterized by disturbances causing large power imbalance like the unplanned loss of large
generating units resulting in large frequency variations and usually the studied time interval is up to
several minutes. System separation is more likely to occur during severe system fault. The long-term
frequency stability is characterized by small perturbations within the system such as power variation of
a generating unit/units or an equipment disconnection without the occurrence of a large disturbance. In
this case the system frequency does not deviate significantly in all the system (no networks separation
occurs) and its behaviour follows the slow dynamic process within the power plants and usually the
studied time interval is up to tens of minutes.
A frequency stability study is therefore classified into two types. The first type of frequency instability is
provoked by a large disturbance within the analysed system such as system faults and/or unplanned
power plant tripping which leads of the power plant or by system faults near the power plant. The system
separation is more likely to be formed when the system fault occurs. The second type of frequency
instability may be provoked by a small system disturbance (no system separation involved) which may
lead to a long-term frequency stability problem due to the slow dynamic phenomena within power plants
operation.

41
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

3.2.2.2 Rotor angle stability (transient stability) linked to frequency stability aspect
Rotor angle stability within a power system represents the capability of that particular power system to
maintain grid synchronous operation. The rotor angle stability issue can be as well split into two main
categories: small signal rotor angle stability and transient stability (in case of large disturbances). A
transient stability (rotor angle stability) can also involve a frequency deviation as shown in Table 3.1.
Some rotor angle stability studies include disconnection of a power plant with a system fault. Due to the
system fault, all synchronous generators will accelerate. After the fault is cleared, the disconnection of
the power plant occurs (See Figure 3.1), which causes the increase in the tie line power flow in the bulk
power system and makes the rotor angle stability worse. Because the power plant is disconnected, the
frequency drop also occurs. If the system size is large, the frequency deviation is not a critical issue,
while the transient stability can be a critical issue. When both the transient stability and the frequency
stability are critical issues, the power swing oscillation is overlapped with the frequency response.

G1 G2 G3 G1 G2 G3

G1 G2 G3 G1 G2 G3 G1 G2 G3 G1 G2 G3

Figure 3.1 Transient stability scenario due to loss of generation.

The primary frequency control and secondary frequency control play an important role for frequency
stability studies. The frequency relays for disconnection of generators or loads also play an important
role for such studies. Because the aforementioned frequency controls and the frequency relays
generally operate in the order of tens of milliseconds, those may be represented using RMS models.
Therefore, the above two studies are typically performed using RMS simulation models and tools.

3.2.3 Response of IBG to frequency deviations and impact on system performance

The IBGs can be divided into three groups depending on its frequency response:
 Non-rotating generators without energy storage (PV and fuel cells)
 Rotating generators (inherent energy storage in rotating inertia)
 micro-turbines,
 wind
 variable speed hydro plants (VSC or cyclo-converter)
 Energy storage (batteries)
The frequency control of IBGs can be split in four major parts:
 Tripping due to over/under frequency protection and ROCOF protection (see Sub-Chapter 3.6)
 Continuous activation/deactivation due to frequency deviation (droop characteristics, frequency
sensitive mode)
 Emergency activation due to large frequency deviation, i.e. limited frequency sensitive modes for
over frequency (LFSM-O) and for under frequency (LFSM-U)
 Continuous activation/deactivation due to ROCOF (synthetic inertia)
A common approach is to realize the droop characteristics and the synthetic inertia as separate blocks,
acting on the corresponding primary energy source (for instance adapting the operating point of PV via
the MPPT) or an optional storage device [38], [39]
The main strategies which have the ability to participate in frequency control, are the following:
 Primary energy source controllability characteristics (including limits)
 Energy storage device capacity and state of charge

42
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 Controller structure and parameter settings (droop, dead-band, synthetic inertia)

 Protection settings, over/under frequency, ROCOF

3.2.3.1 Increase in ROCOF [31]

The initial ROCOF that occurs is purely related to the system inertia at the time. IBG technologies
behave differently from conventional synchronous generators hence the system inertia in small networks
or islanding areas of a power system can significantly be reduced in case of a large penetration of IBGs.
This will have an influence on the ROCOF in case of generation-demand imbalance.
The presence of synthetic inertia in IBGs could contribute to limit ROCOF, but currently is very limited
and not as fast as the initial and inertial response from a synchronous generator (See Sub-Chapter 2.1,
point 1 Rotating mass/inertia). As shown in some recent studies, so-called inertia-based fast-frequency
response, or “synthetic inertia” may offer a means of limiting ROCOF as the penetration of IBGs
increases, This, however, needs to be studied on a case by case basis to identify if the particular design
will meet the needs of the particular system. Consequently, the frequency might drop significantly in the
first seconds before primary control will start to operate. This can lead to high frequency variations after
disturbances and hence increasing amount of load shedding. Similar results are shown in Figure. 3-2
and can be derived from the studies performed in [40], [41]. Many small power systems rely upon
ROCOF to detect power islanding. With increased ROCOF for the trip of a unit, such systems are likely
to falsely detect islanding. In such a case, the protection relays settings will have to be adjusted.

Figure 3.2 WAMS recording (50 Hz sampling).

3.2.3.2 Increasing over frequency transient and steady state frequency excursion
3.2.3.2.1 Over frequency transients
Generally, over frequency transients are a consequence of generation exceeding load demand; some
possible causes are:
 Complete trip of a line/lines connecting an electrical area that previous was exporting energy
 Formation of a small/large island / system splitting
 Loss of a large load or trip/temporary interruption (i.e. commutation failure) of a HVDC link which
was exporting energy in interconnected or islanded systems
The area with excess generation suffers a fast over frequency transient, that must be controlled by
conventional power plants and IBGs by means of an over frequency control droop (i.e. LFSM-O) as an

43
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

emergency solution. Depending on the system configuration, staged disconnection is also a valid
solution, instead of the LFSM-O capability referred to. The system operator may choose to allow within
its control area automatic disconnection and reconnection of generators at randomized times, ideally
uniformly distributed, to keep frequency within the threshold. In case of staged disconnection
applications, it must be verified that shedding blocks of generation is sustainable for the system.
3.2.3.2.2 Under frequency transients
Fast primary frequency response (delivered in the first few seconds of an event) may be required if the
nadir is likely to be reached inside the time frame required for governor action to be effective.
The main causes are likely to be:
 Complete trip of line/lines connecting an electrical area that previously was importing energy
 Formation of an electrical island
 Loss of a large generation units or trip/temporary interruption (i.e. commutation failure) of a HVDC
link which was importing energy in interconnected or islanded systems.
Figure 3.2 is a reported real recording showing the effect of a trip of dispersed generation during an
under frequency transient; the frequency decrease caused by loss of conventional generation reaching
the typical trip value of dispersed generation (49.7 Hz). The system was managed in agreement with N-
1 criteria but the loss of dispersed generation drove the frequency behaviour to the intervention of load
shedding relays. This practical case demonstrates the importance of proper settings of IBGs and stability
of the protection devices.

Figure 3.3 System response after the normative contingency in interconnected operation [42].
When the fraction of dispersed generators (including IBGs) was small, the frequency response was
limited by the fault detection relays, which disconnected the generator from power system after each
fault. Due to large penetration of dispersed generators including IBGs the disconnection approach can
lead to large frequency oscillations (See Figure 3.3). It means if the frequency exceeds a certain
threshold value (e.g. 50.2Hz in Germany [43]) the RES including IBGs were required to disconnect from
the grid. In the case of a small grid with some PV installations (an island system), a large power plant
disconnecting from the grid maybe disconnect some PV installations from the grid which can have a
large influence on any load shedding activation.

44
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.2 Size of grid and key frequency control

Small Isolated Grid Large Network

Main generator ●Diesel engine ●Thermal power (oil and coal,
●Gas turbine natural gas, etc.)
●Small hydro ●Hydro power
●Nuclear power
Potential penetration rate of IBGs Highly Likely Over 50% Not likely over 50%
System inertia Rather small with inverter in case Inertia constant: in range of 3s
of high share of IBGs (H < 3s) and 6s

Critical load/generation change Load/generation change in Load/generation change in

which causes maximum frequency several seconds to a dozen several minutes to a dozen
deviation seconds minutes.
Relevant frequency control for Governor response/primary AGC/secondary frequency
maximum frequency deviation frequency control control

3.2.3.3 Critical dynamic behaviour with respect to system size

It should be noted that the impact of dispersed generators including IBGs in small isolated islands can
be different from the impact of dispersed generators including IBGs in large island or in interconnected
grid. This difference could come from 1) Type of source, 2) Potential penetration rate, 3) Critical
deviation in terms of frequency regulation, as shown in Table 3.2.
In the case of small isolated grids, such as a few MW scale grids, new integration of one utility scale
IBG into the grid could result in over 50% of penetration rate of renewables. As shown in Figure 3.4, a
steep change in IBG output such as a 70 - 80% (equal to 8 kW / 12 kW) change, relative to rated power
output of RES, within 10 seconds can impact in an entire small island grid because the smoothing effect
of IBG output cannot always be expected. In addition, not all generators in the small isolated islands are
equipped with the Automatic Generation Control (AGC), i.e. the secondary frequency control. Such a
steep change in IBG output with high penetration of IBGs can cause a significant system frequency
change. Therefore, the primary frequency control which is expected to reduce the frequency deviation
caused by the steep change in IBG output is the key for the impact analysis of the IBGs in the small
island grid from frequency stability point of view.

Figure 3.4 Example of measured PV output (12 kW three phase inverter).

45
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Conversely, in the case of large (interconnected) grids, in which the penetration rate is unlikely to be
over 50%, a steep change in IBG output as in Figure 3.4 is not generally observed in the large-scale
network because a smoothing effect of the IBG output can be expected even for the high frequency
component of IBG output (corresponding to the primary frequency control range) due to the widespread
location of the IBGs. Thus, a relatively long trend of increase/decrease in IBG output (with the ramping
speed of a few % per hour), such as a small % change relative to the total demand in an hour, becomes
the dominant dynamic behaviour (See Figure 3.5). Because the large capacity thermal power plant such
as coal fired units and AGCC units has relatively low ramp rate limit (normally 1-5 percentage per min)
for the change in active power output, the secondary frequency control which is expected to reduce the
frequency deviation caused by the slow and long change in IBG output is the key for the impact analysis
of the IBGs in large-scale network from frequency stability point of view.

Figure 3.5 July 2002 peak week wind and load in NYISO [44].

In large systems, a steep change in IBG output may still present a challenge when penetration of IBG
is high. The solar eclipse that occurred on August 21, 2017 in the USA, particularly in California, may
serve as an example. The eclipse had obscuration ranging from 58% in Southern California to 76% in
Northern California and lasted for about two hours from 9:11 am to 11:17 am. In the California ISO, the
solar eclipse resulted in a loss of approximately 6 GW of utility based solar generation and 1.460 GW
of behind the meter solar generation. In addition, there was reduction of load due to cooler temperatures.
Half of the reduction in solar generation during the solar eclipse was replaced with an increase in
electricity imports. During the solar eclipse, the system absorbed a downward ramp of 48 MW a minute;
after the sun emerged from the solar eclipse, solar production ramped upward rapidly as much as 150
MW per minute.
Since the solar eclipse was expected, the California ISO developed a readiness plan and procured
additional reserves for regulation to help control frequency swings and meet required control
performance metrics. In addition to preparing for the loss of significant solar production, the California
ISO also prepared for the near-immediate reversal and rapid increase in solar production coming out of
the eclipse, which could also create operational challenges. It was also anticipated that the rapid
increases in large amounts of generation following maximum obscuration could cause oversupply
conditions and system frequency management issues. Having Energy Imbalanced Market (EIM) that
gives western states utilities access to a real-time trading market and sophisticated optimization
technology, facilitated system operation during and after the solar eclipse.
Figure 3.6 shows solar energy production on the day of the solar eclipse and on the adjacent days.
Figure 3.7 shows solar production during the solar eclipse compared to the system frequency. More
detailed description of the eclipse and the system operation during that time can be found at [45].

46
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3.6 Solar energy production during the solar eclipse and on the adjacent days in CAISO [45].

Figure 3.7 Solar energy production during the solar eclipse and the system frequency [45].

3.2.4 Real life examples

The California ISO (CAISO) is concerned that during periods of light load and high renewable production,
the system may require reserving headroom on governor responsive resources or primary frequency
control capability to meet frequency response obligations. Initial study results indicated acceptable
frequency performance within WECC.
A dynamic stability data file was created to match the power flow case. The latest WECC Master
Dynamic File was used as a starting dataset. Missing dynamic stability models for the new renewable
projects were added to the dynamic file by using typical models according to the type and capacity of
the projects. The latest models for IBGs recently approved by WECC were utilized. For the new wind
projects, the models for double-fed induction generators or full converters were used depending on the
type and size of the project. For the solar PV projects, three types of models were used: large PV plant,
small PV plant and distributed PV generation. It was assumed that the large plants (20 MW and higher)
have centralized plant control. The most severe credible contingency impacting frequency was studied:
an outage of two units of a large nuclear power plant with a loss of approximately 2.6 GW of generation.
As shown in Figure 3.8, the CAISO’s frequency response was lower than the CAISO frequency response
obligation. Compared to the CAISO’s actual system performance during disturbances, the study results

47
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

seem optimistic because actual frequency responses for some contingencies were lower than the
dynamic model indicated. Optimistic results were partly due to large headroom of responsive generation
modelled in the study case. Amount of headroom on responsive governors is a good indicator of the
frequency response metric, but it is not the only indicator. Higher available headroom on a smaller
number of governor responsive resources can result in less frequency control capability as well as the
lower available headroom on a larger number of governor responsive resources for the same
contingency.
Requirement to provide frequency response from IBGs may be a good solution to the problem of
insufficient frequency response. Simulations of a grid restoration study in continental Europe after a
blackout revealed possible problems with distributed generation. Starting grid restoration with a
frequency above nominal frequency with 51.5 Hz leaves a sufficient margin to avoid load shedding at
49 Hz due to unintentional frequency fluctuations. However, a system will have to reduce frequency
close to nominal frequency in order to synchronise with neighbouring systems. According to older
standards, distributed generation was intended to reconnect at 50.2 Hz automatically. This leads to an
increase of system frequency, decrease of generation from conventional units participating in primary
frequency control and possible tripping of distributed generation in case 50.2 Hz is exceeded. Depending
on the time constant of primary frequency control and switching hysteresis as well as accuracy of
frequency measurement of distributed generation, the system might get unstable. The example shown
in Figure 3.9 emphasizes the need of taking over/under frequency protection and automated
reconnection of distributed generation into account for frequency stability studies.

Figure 3.8 Frequency on a 500 kV bus in central California with an outage of two nuclear units in
the cases with high and low headroom and inertia.

48
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

50.5 0.6

Hz p.u.

50.2 Hz

50.0 0.4
frequency

frequency
49.5 0.2

49.0 0.0
0 5 10 15 min
frequency TSO1 frequency TSO2 PV TSO1a PV TSO1b

Figure 3.9 Simulation of 50.2 Hz effect, tripping and automatic reconnection of PV in Europe [46].

3.2.5 Recommended functionalities of IBGs for frequency deviations

The recommended functionalities of IBGs shown in Chapter 2 are examined in terms of the frequency
deviations as shown in Table 3.3. The three categories are formed based on the following criteria:
1) Control category: The internal control of the inverter which is performed at a local level.
2) Protection: The protection relay for the inverter. The grid protection relay for the inverter is also
included. The control for protecting the internal devices is also categorised as the protection.
3) Capability: The grid-friendly control or the control which owns an ability for improving the grid stability.

It is noted that the necessity ranking in Table 3.3 is based on general power system dynamic studies
and there could be an exception when a specific study is performed relating to the frequency deviation
or with special controls / special system conditions is performed.
The major functionalities which should be considered for frequency deviations are the following:
(1) MPPT: The time for running the simulation also known as “stop time for simulation” for frequency
control and stability study can be split to two: 1) less than 30 seconds 2) longer than 5 min. The first one
is to examine the frequency nadir and the settled frequency in the case of a generator tripping. The
second is to examine the peak-to-peak value of the frequency fluctuation especially in the small isolated
grid. When 2) needs to be examined, the change in MPPT signal needs to be considered for this type
of phenomenon. This is because the solar radiation cannot be assumed to be constant for the
aforementioned time frame. However, the use of the MPPT model element is not always necessary
because the control response speed is fast enough for the long-term dynamics. Therefore, the change
in MPPT signal may be modelled as the change in the active power reference (Pref) itself.
(2) Current limit: When the frequency drops, it is desirable for any generator to increase the active
power output to mitigate the frequency nadir. On the other hand, the active and reactive power output
of the IBG is limited by the maximum current. Therefore, the increasing amount of the active power
output of the IBG during the frequency drop might be limited not only because there is no available
headroom, but also because the allowable current from the IBG is limited. That means, the different
logic for the limitation of the current could lead to the different increasing amount of the active power
output. It will also result in a different dynamic frequency response. Therefore, the current limit needs to
be modelled for the frequency deviation.

49
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.3 Necessary functionalities for frequency deviations

Category Functionalities Necessity Remark
Control DC Source control No
Control Current control No
Control PLL No
Yes (only for
Control MPPT small isolated See (1) above
grid)
Reduction of maximum inverter
Protection current when the DC voltage No
overcome a certain limit
Limitation of inverter current’s
Protection No
variation rate after a fault
Protection Current limit Yes See (2) above
Protection DC overvoltage protection No
Overvoltage/Under voltage
Protection No
protection
Over frequency/Under frequency
Protection Yes See (3) below
protection
Protection for detecting balanced Positive sequence overvoltage protection
Protection No
fault may be used.
Protection for detecting Negative sequence overvoltage protection
Protection No
unbalanced short-circuit fault may be used.
Protection for detecting single-line- Zero sequence overvoltage protection may
Protection No
to-ground fault be used.
ROCOF tripping: monitoring the
power frequency variation rate and
Protection Yes See (4) below
disconnecting the inverter when it
reaches a certain limit [Hz/s]
Protection Vector jump Yes
Protection Transfer trip No Yes if massive IBGs are tripped
Anti-islanding active detection Not all the utilities require using this
Protection Yes
method method for IBGs. See (6) below
Capability P(f) control (over/under frequency) Yes See (7) below
Voltage control by means of
Capability Yes See (8) below
reactive power
Voltage control by means of active
Capability No
power [P(V)]
Capability Synthetic inertia Yes See (9) below
Capability ROCOF immunity Yes See (10) below
Capability FRT (LV/HV) No
Zero current injection, Reactive current
control calculated by mean of power factor
Active behaviour during voltage
Capability No input, Maximum reactive current injection,
fast variations
Reactive current level depending on
voltage depth
If the damping controller is designed to
Capability Power oscillation damping (POD) Yes damp the common mode of frequency
evolution See (11) below
Note: Control for grid support functions are examined in the category of the capability.

In case of small isolated systems, the number of synchronous operating units could dramatically
decrease due to the integration of IBGs. For example, consider the number of synchronous operating
units of two in a small isolated system. When one out of two units is disconnected from the grid, not only
a significant frequency drop occurs, but also a significant voltage drop. If the IBGs generate less than
100% (of the rated current), say 70%, the IBGs active power output may be assumed as constant.

50
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Conversely, if the IBGs are operated at almost 100%, the IBGs active power output cannot increase and
the active power output will decrease when the terminal voltage of the operating unit drops, which
causes a larger frequency drop.
(3) Over frequency/Under frequency protection: As already mentioned in Item 3.2.3.2.2, the large
integration of the IBGs could cause a larger undesired frequency nadir especially in case of a small
electrical island or the loss of a large generation unit. This could lead to the tripping of the over/under
frequency protection. In addition to those relay operation, automatic reconnection should be also
considered.
(4) ROCOF protection: The large integration of IBGs could cause a larger undesired frequency nadir
especially following significant increase/decrease in load or generation.
(5) Vector jump: The vector jump method is a typical passive anti-islanding protection scheme. The
fundamental principle is to detect the islanded condition via a change in voltage phase. Generally, one
of the assumed triggered events for frequency deviations is the generator tripping. Not only the
generator tripping but also the disconnection of IBGs and the interruption of their current injection can
be the triggering events. As shown in Equation 3.1 (See Figure 3.10), when the active power flow
changes dramatically without significant change in the system voltage, the angle between the two buses
can exhibit the step change D2 when the generator trips. Such immediate angle change can cause the
undesired vector jump operation and the disconnection of the PV (including the power electronic device
blocking of the PV). Such loss of IBGs can also cause further frequency drop.

 VV
 P  1 2 sin  2 - 1 
 X VV
  DP 1 2
D 2
 P - DP  VV
1 2
sin  2 - D 2 - 1 
X

 X
Equation 3.1

P-DP V2 2-D2 Large

jX Grid
HV
V1 1
HV MV/LV
G G
2G 1G Load IBG
Figure 3.10 Voltage phrase jump mechanism by means of active power [47].
(6) Anti-islanding active detection method: Anti-islanding active detection methods intentionally inject
negative/positive reactive power into the network, which assists the growing frequency change/deviation
when the designated system is isolated from the main grid. A small portion of anti-islanding active
detection methods utilize the GPS signal or the system frequency for determining the control signal for
the positive/negative reactive power injection. The IBGs which are equipped with such anti-islanding
active detection methods could inject positive/negative reactive power in synchronization with each
other. Due to the voltage sensitivity of loads, an increase in voltage caused by an increase in reactive
power via the anti-islanding protection method, will lead to an increase in active power consumption of
loads and to further system frequency drop (See Figure 3.11).
Moreover, the frequency change which is derived from the bus voltage can also occur when the voltage
magnitude changes due to the reactive power injection, no matter how responsive the load is to voltage.
As shown in Equation 3.2, when the small reactive power injection from the IBG does not have ability to
change the connected bus voltage, a voltage phase change occurs (See Figure 3.12), which leads to
the change in the active power i.e. the frequency (See Equation 3.1).
Therefore, if a positive reactive current is injected in the case of a frequency drop, a further system
frequency drop could occur.
It should be noted that the anti-islanding active detection method is not widely used around the world
and such a model is not needed to be modelled for an IBG which is not equipped with the anti-islanding
active detection method.

51
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Frequency
Variation

Inertia

Active Power
PV Array
Variation 0.0kW
PV Inverter
Load Voltage
Controller
Characteristics
Protection

Voltage
Variation

PV Capacity
Reactive Power Active
Electric Distance
Variation Anti-islanding
Control Signal

Figure 3.11 One example mechanism how frequency is forced to be oscillated

via Q variation type active anti-islanding control signal.

 V1V2 V22


Q  cos   2 -  1  -
V1V2

X X
2
 DQ  2 - 1  D 2
Q - DQ  1 2 cos  + D -   - 2
V V V X

 X
2 2 1
X
Equation 3.2

Figure 3.12 Voltage phase jump mechanism by means of reactive power [47].
(7) P(f) control [35]: The increased utilization of tie-lines is encouraged in terms of increased integration
of RES. The increased tie-line power flow could lead to more significant frequency increases/decreases
after a power system is split into two sub-systems due to tie-line tripping. A frequency rise in the system
could lead to a disconnection of a large amounts IBGs due to the over frequency protection. On the
other hand, the tripping of this large amount of generation can lead to a frequency drop and therefore a
disconnection of load due to the under frequency load shedding scheme. To overcome this problem,
IBGs can control frequency by means of active power, i.e., reduce their active power feed-in during over
frequency, or increase their active power feed-in during under frequency. Hence, this function can play
an important role when the power system experiences a significant frequency change. In this context,
the following topics are addressed:
 Modelling the primary power source control (if existing) including maximum range and gradients
(for frequency rise/drop)
 Modelling the optional storage devices including charging management controls (for frequency
rise/drop)
 Modelling the power reserve in the case, where the IBG is operated at a point below the maximum
power point in order to allow primary frequency control (for frequency drop only).
(8) Voltage control by means of reactive power: Although the dynamic behaviour is not directly
related to this capability, the pre-fault active and reactive power feed-in of the IBG can affect the fault-
on and post-fault dynamic behaviour of the IBG output. This could indirectly change the dynamic
frequency response of the system.
(9) Synthetic Inertia: The displacement of synchronous generators by IBGs could cause a reduction of
the system inertia. According to the power swing equation, the lower inertia will increase the change in

52
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

rotor angle speed (df/dt) for the same mismatch between the mechanical power and the electrical power.
The increase in the ROCOF could endanger frequency stability.
IBGs will not exhibit an inertia-like behaviour per se. While some of the prime movers could deliver the
inertia, it is not certain that the IBG is able to provide it, because it requires a modification of the control
(See Sub-Chapter 2.1 (1)). However, assuming an appropriate control scheme of the inverter and a
(limited) energy storage capability, a “synthetic inertia” can be provided to the grid.
Potential solutions that can be used are
 Installation of energy storage devices, e.g., large flywheel generators or flywheel coupled
synchronous compensators.
 Adopt alternative approaches to detect electrical islands and ensure that the main equipment is
robust against ROCOF.
 Operate PV systems below the maximum power point.
 Modify control of IBGs to provide synthetic inertia21, which can be done by a change of active power
output during frequency deviations. This is the approach which has been followed by Hydro Quebec
[48], [49]. It is noted that this technology is recognized as the fast frequency response (FFR) and
the closed-loop inertia-based FFR has been commercially available for many years. However, the
practical use of FFR has still being discussed and needs to be carefully examined and implemented.
(10) ROCOF immunity: ROCOF immunity is equivalent to frequency ride-through, i.e., low and high
frequency ride-through. If the ROCOF setting for the ROCOF immunity is the same as the relay setting
for the ROCOF protection, the ROCOF immunity model is not necessary. On the other hand, it is more
likely that the relay setting of ROCOF protections could change depending on the location. In such a
case, the ROCOF immunity needs to be modelled independently.
(11) POD: If the loss of generation causes poorly damped power swing oscillations, POD plays an
important role to ensure transient stability. Therefore, in the aforementioned case and if the POD is
assumed to be implemented into IBGs, IBGs with POD should be modelled, otherwise, there is no need
to model POD.
3.3 BEHAVIOUR IN RESPONSE TO LARGE VOLTAGE DEVIATIONS
A brief introduction of stability studies, with the key responses relevant to system faults are examined in
this sub-chapter. The real-life examples with and without the IBGs are illustrated. In addition, the
important functionalities of IBGs for their behaviour in response to large voltage deviations are
discussed.
3.3.1 Description of phenomena
Large voltage deviations may be triggered by system faults such as three-phase faults, single-line to
ground faults and short-circuit faults. Unless a great fraction of generation or loads compared to total
rated power output change, frequency stability can be omitted. Severe system faults could cause the
out-of-step phenomenon and possibly lead to large-scale blackouts. This issue has generated great
interest in knowing the system response behaviour to large voltage deviation for the power system
planners and operators.
Several phenomena can be linked to the system faults, such as:
Out-of-step phenomenon: This phenomenon refers to losing synchronism of a synchronous generator
connected to the system following large voltage deviations. When a synchronous generator falls into the
Out-Of-Step (OOS) condition, at least one voltage angle difference between two buses including the
machine internal (fictional) bus behind its transient reactance becomes equal to 180 degrees. When the
angle difference is 180 degrees, there is a point where voltage is zero between the two generators
showing this difference. The point is known as the “electrical centre (See point Ec in Figure 3.13)”. The
impact of the OOS condition at the electrical centre is the same as a short-circuit at this point. However,
the difference between a short-circuit and the OOS phenomenon at the electrical centre is the duration
and the repetition of the zero voltage, as shown in Table 3.4. In practice, OOS is generally caused by
more than two synchronous generators and it seems that OOS is more important when the network size
is larger and the voltage level is higher.

21
The model of the wind turbine aerodynamics is also needed when simulating the synthetic inertia.

53
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.4 Difference between normal short-circuit and out-of-step phenomenon

Short-circuit fault Out-of-step (OOS)
Fault duration from tens of ms to a few hundreds of ms Instantaneous
Repetition of In general, once. a few times per second until OOS relay
zero voltage operates
condition

Electrical
Centre
Voltage
Current

Figure 3.13 Image of dynamic behaviour of voltage and current in case of OOS.
Because PVs have no rotating masses behind their inverters, they are not subject to rotor angle
instability issues. Other types of RES, such as Type 3 and Type 4 WTGs and converter-driven induction
machines are non-synchronously connected; hence, the OOS phenomenon is also out of scope.
However, the massive penetration of IBGs may significantly impact the ability of the power system to
prevent OOS, due to the reduction of system inertia, the reduction of the reactive reserves from
synchronous generation and the reduction in the number of voltage and power system stabilizer devices.
Those changes mainly arise from the difference in characteristics between IBGs and synchronous
generators which were illustrated in Chapter 2. In light of this, various requirements, such as IBGs FRT
capability, have been examined and imposed in some countries.
Short-circuit Current: A short-circuit denotes the accidental or intentional conductive path between two
or more conductive parts (e.g. three-phase short-circuit) forcing the electric potential differences
between these conductive parts to be equal or close to zero. The short-circuit current denotes the large
current resulting from the short-circuit in an electric system [50]. Because the short-circuit current usually
has a DC component as well as an AC component as shown in Figure 3.14, the DC component could
be reduced depending on the X/R ratio in the designated short circuit as time progresses. Many
transmission line protections, transformer protections and generator protections at least implicitly rely
on the difference between the current magnitude during the short-circuit and the loading current (before
the short circuit). The electric quantity used for the relay setting is mostly an RMS value.
The increasing penetration of IBGs tends to reduce the short-circuit current during faults especially when
the IBGs are connected near the fault point because the current provided by an IBG is generally limited
to the nominal value of this equipment. If the IBG is oversized a certain amount of additional short circuit
current may be achieved. IBGs can also, if requested by the responsible entity (TSO, DSO, utility),
effectively inject additional reactive current to support voltage during the fault and help its recovery after
the fault clearing. Such dynamic reactive power (current) support is limited and is not at all comparable
to the short-circuit current provided by synchronous generators especially when the IBGs are connected
near the fault point.

54
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Ik”: initial symmetrical short-circuit current

Ip: peak short-circuit current
Ik: steady-state short-circuit current
iDC: DC component of short-circuit current
A: initial value of DC component, iDC
Figure 3.14 Example short-circuit current (schematic diagram) [50].
A unique property of IBGs is the power electronics interface and the ability of the electronics to control
fault current contributions. This adjustability can thereby optimise the system protection coordination
issues by controlling fault current levels. Typically, IBGs are designed to act as ideal current sources.
They provide minimal fault current contributions and have little effect on overcurrent protection and
coordination strategies for fuse and circuit breakers. Even though the fault contribution from a single,
small IBG unit is not significant, the total contribution of many small units may alter the fault current level
enough to cause lack of coordination for overcurrent protection and wrong fuse operation or hamper
fault detection [51]. For instance, in case of islanding situations for small IBGs, the fault current
contribution can be relevant from the protection coordination point of view in low voltage and/or medium
voltage network although such contribution depends on the system size of the isolated grid.
Short-term dynamic voltage response: This term generally refers to the voltage recovery after a fault
is cleared. The time frame is at most a few seconds after fault is cleared 22 . If a large number of
conventional (i.e. not interfaced to the grid through power electronics) induction motor (IM) loads are
connected to the power system, the voltage recovery may be delayed by the IMs taking time to re-
accelerate, which requires a large amount of reactive power. If the situation becomes severe, the IMs
cannot re-accelerate and stall; then, if the IMs are not disconnected by the under-voltage protection, the
load bus voltage is locked at a very low value before slower (thermal) protections disconnect the motors.
This situation may occur due to the presence of massive IM-based air conditioning load in summer peak
load conditions; those IMs usually have low inertia. Conversely, the motors controlled by drives or
inverter-interfaced IMs, widely used around the world mainly in domestic appliances, such as in air
conditioners and in refrigerators, seem to be less prone to depressing the voltages.
The inverters react in the same time frame and hence it is relevant to consider their impact on short-
term voltage stability/recovery. The reduction of active power output of IBGs can cause an increase in
the net load which is derived from “pure loads minus IBG active power output”. The increase in the net
load can cause a lowering of the voltage level at the load bus. Therefore, IBGs without the dynamic
reactive power support will worsen the voltage recovery, and can eventually deteriorate short-term
voltage stability. It is noted that such dynamic reactive power support provided by IBGs can lead to
excessive voltage recovery (The real event revealed that IBGs were tripped due to the overvoltage),
which is very much likely to require HVRT (See also Figure 5.6 in Chapter 5).
Although both short-term voltage instability and rotor angle instability result in low voltage at certain
buses in the grid, it is appropriate to differentiate those instabilities based on the final behaviour of the
grid such as OOS or voltage collapse as well as the effective remedial actions such as generator tripping
or load shedding. In light of this, various requirements, such as dynamic reactive current (power) support,
have been examined and imposed by grid codes in some countries. It should be noted that different

22
Very similar time frame for transient stability

55
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

dynamic reactive current control schemes including active and reactive power limiter control or dead
band characteristics can have different impacts on short-term voltage stability.
3.3.2 Relationship between behaviour in response to large voltage deviation to relevant
study
3.3.2.1 Short-term voltage stability study and short-term voltage behaviour following faults
Short-term voltage stability problems can occur after a disturbance if the stability limit of the network is
exceeded within a time frame of a few seconds. Thus, the stability limits are analysed with time-domain
simulations, where the system behaviour is tested with critical contingencies including large voltage
deviations. Reactive power reserves can be of minor importance in the short-term range compared to
the long-term range [36]. However, not only SVCs or STATCOMs (and their smaller-size variants
adapted to distribution systems) but also PVs with the advanced capability can prove useful if located
near the load centre exposed to short-term voltage instability [52]. Indeed, they can provide immediate
and full-scale reactive power support following faults.
Comparatively, switchable capacitor banks are less able to support the system in the short-term range
because bank switching times are usually too long, and switching off may be required to avoid
overvoltages once the motors have regain normal speeds. The IBGs can impact voltage stability through
the priority given to reactive over active current in low voltage conditions. On one hand, the temporary
reduction of active current, and even more the disconnection of IBGs without FRT requirement,
increases the net load. On the other hand, the reactive current increased up to the thermal limit of the
IBG can have a significant positive effect on the local voltage. Besides the impact on dynamic reactive
reserves, the impact of the IBGs on short-term voltage stability limits is related to the active power
support of IBGs during low voltages (See TB 450 [36]).
The following representative short-term voltage stability study is performed to gather the following
indicator:
 “Critical Area Exchange [36]”. This value determines the maximum export from one area to another
area at which the system is remaining stable for a specified fault, e.g. a three-phase fault with a
designated fault clearing time from the voltage stability point of view. Usually the dedicated power
system studies are performed considering detailed contingency analysis for N, N-1 and N-2
situations.
 “Loading Level”. This value determines the maximum loading level without voltage collapse. In this
study, the fraction of the induction motor among the load is also an important indicator.
The OOS phenomenon should not be confused with short-term voltage stability. The slow voltage
recovery caused by the disconnection of IBGs without the FRT requirement or the high penetration of
induction motors could lead to shortening the time to the OOS condition. The short-circuit current is also
not so significant in the short-term voltage stability study. However, the decrease of the short-circuit
current level could extend the fault duration and the longer fault duration could make the power system
more vulnerable to voltage instability (See Figure 3.15).

Short-term Rotor Short-circuit

Voltage Angle Fault LVRT
Stability Stability Current Study
Study Study Study

Short-term
Out-of-step Short-circuit Dynamic
Phenomenon Current Voltage
Response*

* Especially short-term voltage recovery

Figure 3.15 Type of studies and phenomena with large voltage deviation.

56
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

3.3.2.2 Rotor angle stability in case of large disturbances (transient stability)

Transient stability is defined as the ability of the power system to maintain synchronism after severe
disturbances such as short-circuits cleared by lines opening or generator trips [30]. It is also referred to
by the term large-disturbance rotor angle stability. Hence this aspect of power system stability is related
to the behaviour of the synchronous generators located near the fault trying to maintain synchronism
with other synchronous generators during and after the disturbance. The system behaviour is highly
dependent on the type and duration of the disturbance. Hence, to ensure rotor angle stability, a number
of critical contingencies have to be simulated at different locations. For evaluating rotor angle stability,
the following three indices are used [36], [40] alone or combined:
 CCT: The fault Critical Clearing Time (CCT) is calculated for various critical contingencies, i.e.
different fault conditions. The CCT represents a useful measure for characterizing the security
margin with respect to rotor angle instability in a given dispatch scenario.
 Critical area exchange: determines the maximum export from one area to another at which the
system still remains stable after a specified fault with a designated fault clearing sequence.
 Maximum power transfer: This value determines the maximum export from one power station to
another area without loads at which the system is remaining stable for a specified fault, e.g. a three-
phase fault with a designated fault clearing time from the rotor angle stability point of view.

Although the CCT is a very common indicator for showing the degree of rotor angle stability, it is not
frequently used in large system rotor angle stability studies performed by the system operators in many
countries. The major reason is that the primary and back-up relay operation times are always fixed /
constant for the designated fault points (usually these times are derived from previous CCT studies). It
is likely that in the EHV and HV networks the shortest fault clearing time is around 3 cycles and the
longest in MV network is in the range of a few cycles and less than several hundreds of milliseconds.
Therefore, the derivation of the critical area exchange or the maximum power transfer are usually the
studies preferred by system operators as regards rotor angle stability point of view.
If the power system cannot maintain synchronism, OOS occurs, which could trigger a cascaded failure
and a large blackout. If a first-swing OOS occurs, a resulting short-term swing of voltage magnitude can
be observed near the electrical centre of the system. On the other hand, if a multi-swing OOS occurs,
such a short-term voltage response is not likely to be observed since its short-circuit current is not
directly linked to rotor angle stability. However, if the protection cannot detect the fault properly due to
insufficient short-circuit current from the grid, the longer fault clearing time could endanger rotor angle
stability. In this respect, the decrease in short-circuit current level caused by the replacement of
synchronous generators by IBGs can be of interest in a rotor angle stability study.
3.3.2.3 Provision of short-circuit fault current (short-circuit fault current)
Provision of fault current by IBGs is often examined in the case of the grid interconnection study. In
some countries, the fault current level is assumed to be 150%of the rated current of the inverter 23. Then,
the short-circuit current which flows through the circuit breaker (CB) is calculated based on a simple
electric circuit calculation and is checked to see if it is less than the maximum current for which the CB
can properly operate to clear the fault. However, due to the dynamic behaviour of the inverter, the short-
circuit current could vary depending on controller action such as dynamic reactive power support. In
addition, the current that is to be examined is not only the peak fault current [50] but also the breaking
current [50] which is interrupted when the CB operates. This means that the fault current 70 - 100
milliseconds after the fault occurrence is determined and examined. It is noted that [50] provides the
way how to calculate the fault current provided by IBGs focusing on DFIGs. On the other hand, the
infeed of the fault current very much changes on the design of the inverter control including the FRT
function. Therefore, the importance of the simulation-based approach would be getting higher and higher
as the penetration rate of IBGs increases.
Another typical study related to the provision of short-circuit current is aimed at confirming the proper
operation of protection systems such as the line protection. The additional current could cause an error
in some line protection schemes, which could result in an unwanted operation or a failure to operate.
Usually under this scenario study detailed short-circuit computations are performed both for peak short-
circuit conditions (i.e. the power system operating condition which gives the largest transient short-circuit

23
For example, the fault current provision of IBGs is assumed to be in the range of 1.1 to 1.5 times of the rated current of
IBGs according to the results of the laboratory test using the actual IBGs in Japan. Those quantities have been used for
countermeasure of electric facilities (e.g. examination of thermal limits) and for the analysis of the relay operation.

57
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

power) and off-peak short-circuit conditions (i.e. the power system operating condition which gives the
smallest transient short-circuit power) which are compared with the normal conditions currents.
Because this type of study deals with the situation during a fault, the short-term dynamic voltage
response and the OOS phenomenon, which are both related to the dynamic behaviour after the fault is
cleared, are usually insignificant.
3.3.2.4 LVRT study
The grid code of most countries requires that the IBGs stay connected in case of network faults leading
the voltage to remain above some limit curve (See Figure 5.8 in Chapter 5). This is referred to as Low
Voltage Ride-Through or Fault Ride-Through (LVRT or FRT) capability. It is indeed important that IBGs
stay connected in case of major transmission faults, if they produce a significant amount of power. LVRT
capability is a definite requirement for all larger IBGs in most countries.
The LVRT study may be classified into two types:
 LVRT studies for obtaining better LVRT capability from the grid stability point of view.
 LVRT studies for impact analysis on the dynamic stability studies with and without LVRT capability
for the IBGs.
Although it is not likely that this type of study is performed by utilities and system operators on a regular
basis, the desired LVRT capability could evolve as the penetration level of the IBG increases and,
therefore, the need for these studies becomes important.
Another dynamic simulation study requested by system operators deals with the impact of having or not
low voltage ride-through capability on rotor angle stability, on short-term voltage stability and on
frequency stability. Therefore, the OOS phenomenon and the short-term dynamic voltage response can
be significant in the LVRT study.
Besides the above LVRT study, the LVRT capability itself, i.e. the ability of continuous operation of IBGs
following faults, needs to be carefully examined by the manufacturers. Usually this type of study is
included in the overall grid connection study which highlights the impact of a new IBGs site on the
existing grid. LVRT capability is not easily achieved when the IBG is weakly connected to the main grid.
In such a case the LVRT capability is analysed taking into account the short-circuit current during the
fault. Therefore, the short-circuit current assessment can be significant in the LVRT capability analysis.
3.3.3 Response of IBGs to large voltage deviations and impact on system performance
3.3.3.1 Short-term voltage stability
The short-term voltage instability with large integration of IBGs may often occur after a significant system
configuration change such as high penetration of IBGs. Figure 3.16 shows an example of voltage
response of the real single phase PV inverters following faults obtained in a power system simulator.
The real residential single-phase PVs are connected to the test system. After the system fault at 3.4
seconds, the faulted line is promptly removed from the network and is reconnected to the network at 7.4
seconds as shown in Figure 3.16. The entire network looks stable at 30 seconds after the fault occurs.
However, the single-phase PVs which are connected to the line-to-line voltage Vca, fail to restart and
keep being disconnected from the network. Because of the disconnection of the PVs (Note that the PVs
do not meet the FRT requirement), the net load increases and the load bus voltage significantly
decreases.
To have 70% of the rated voltage is quite abnormal and the system cannot be operated for a long time
at such low voltage. Resistive loads can operate at a lower voltage of a P-V curve (also known as a
nose curve) as shown in Figure 3.16, however, the constant power load cannot and such load must be
disconnected due to voltage instability. Thus, high penetration of IBGs could cause the significant
voltage drop following system faults, which also could cause the short-term voltage instability including
the slower voltage recovery with induction motors.

58
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

(a) Measured response of synchronous generator (Rated Voltage of PV is 200 V)

(b) Measured response of PVs (Rated Voltage of PV is 200 V)

Figure 3.16 Measured response obtained in power system simulator [53].

59
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

3.3.3.2 Short-term dynamic voltage response [54]

The short-term voltage instability is indirectly linked to the slow voltage recovery after the system fault
is cleared. Therefore, the study of the dynamic voltage response following faults is important in terms of
the short-term voltage stability. WECC performed a study to investigate the need of the reactive power
control function in IBGs. The example test system included 800 MW PV plants and a three-phase fault
was applied to the Point Of Interconnection (POI) at the 230 kV Switching Station as shown in Figure
3.17. The conditions for the study, are listed in Table 3.5.
The example of the simulation results are summarised as Table 3.6 and Figures 3.18 and 3.19. The
detailed simulation results including the provided models are described in Appendix 3-E. The major
findings are the following:
 Slightly lower voltage without voltage control from the solar PV plant (at 230kV)
 Better damping and faster voltage recovery with the solar PV plant (at 230 kV)
 Overshoot in voltage control scenario (at 230 kV)
 No impact from motor load stalling (at 230 kV)
 Delayed voltage recovery with stalling of single-phase induction motor loads (at the load bus)
 Voltage control of the solar PV plant has no impact (at the load bus)

Other than those above findings, the following bullet points also need to be considered for the study that
includes short-term dynamic response. It is noted that
 Positive impact on transient voltage performance from IBGs – better damping and faster voltage
recovery due to low inertia and faster controls (if a PV plant regulates voltage).
 For large system studies, very detailed models of the generators are not necessary.
 Without LVRT Capability (for example, with large amount of distributed solar PV), loss of large
amount of generation with a fault may be a concern
 High voltages under normal system conditions may be a concern with large penetration of
distributed solar PV generation

Figure 3.17 WECC test system with large-scale PV plants [54].

60
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.5 Condition of voltage and reactive power control [54]

Scenario 1 2 3 4 5 6
Generator power factor +/-0.95 +/-0.95 1 1 N/A N/A
Generator Q and V control Yes Yes No No N/A N/A
Network Motor Stalling No Yes No Yes No Yes
Inverter Q priority Yes Yes No No N/A N/A
Inverter P priority No No Yes Yes N/A N/A
Source PV PV PV PV Thermal Thermal

Table 3.6 Summary of short-term dynamic voltage response [54]

Transient voltage Post-fault Slow voltage recovery
Scenario on generator steady state on adjacent load
terminals voltage buses
1 Solar PV, voltage control, induction
High Normal No
motors don’t stall
2 Solar PV, voltage control, induction
High Normal Yes
motors stall
3 Solar PV, no volt and Q control,
Normal Low No
induction motors don’t stall
4 Solar PV, no volt and Q control,
Normal Low Yes
induction motors stall
5 Thermal, induction motors don’t stall Low Normal No
6 Thermal, induction motors stall Low Normal No

61
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Scenario 1 – blue, Scenario 2 – pink, Scenario 3 – green

Scenario 4 – brown, Scenario 5 – purple, Scenario 6 – dark green
Figure 3.18 Voltage at PCC (230 kV) [54].

Scenario 1 – blue, Scenario 2 – red, Scenario 3 – light green

Scenario 4 – brown, Scenario 5 – purple, Scenario 6 – black
Figure 3.19 Voltage at load bus [54].

62
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

3.3.3.3 Short-circuit current provision

The inverter interface between the IBG and the utility system connection can either use a voltage control
scheme or a current control scheme. The DC link capacitor between the DC/AC converter and the IBG
unit holds the voltage near constant during transient conditions. The voltage control scheme has higher
initial current overshoot, while the current control scheme has a much slower increase and decreases
back to steady-state values. The fault contribution will be higher during the transient period if the IBG is
under a voltage control scheme [51].
A unique property of a power electronics interface is the ability to program the fault characteristics from
the inverter and the response time to fault condition, thereby allowing negligible impacts on protection
coordination. Research has indicated that power electronics can optimally regulate and limit IBG fault
current, improve power quality, and provide the utility with reactive power control and voltage regulation
at the PCC. Continued testing of actual inverter fault characteristics is needed to develop information
that could be used in modelling (detailed and simplified) and fault analysis programs [18], [55], [56].
However, short circuit current from IBG cannot be delivered in the first moment after the event.
Dependent on the PLL-behaviour of the IBG [57], it will take a while until the control acts to deliver
reactive power.
Some inverters have specifications for the short-circuit contribution, defined for symmetrical and
asymmetrical currents. Figure 3.20 shows two real life examples of the recorded short-circuit current
provided by a large-scale PV plant connected to a weak AC grid with constant power factor control and
active power priority injection.
The figure indicates that the three-phase PV inverters can have a potential capability of providing a
marginal contribution to the network fault current. The amount of peak fault current is determined by the
inverter overload capability but it does not generally exceed the nominal rated current by more than 30%.
The peak fault current decays quickly as the inverter control will act to maintain the fault current at
approximately the rated fault i.e. the RMS fault current contribution of the inverter is approximately 1 p.u.
Limitation of fault current shortly after the fault occurrence is essential to maintain the integrity of IGBT
switching devices used in the inverter. Most PV inverters are designed such that the inverter control will
inject the reactive current of up to 1 p.u. shortly after the fault occurrence so as to assist the network
voltage recovery.
This control generally supersedes the pre-event control strategy of the inverter, and whether the inverter
is controlling voltage, power factor or reactive power will make very little difference in its fault current
contribution. The fault current contribution of solar inverter is very similar to that of a type 4 (i.e. full
converter type) wind turbine, determined by controls in the power electronic converter rather than
contribution from a generation source.

63
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3.20 Example of short-circuit current provision measured in PV plant.

64
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

3.3.3.4 Transient stability with and without LVRT [16]

The angle stability is strongly related to angle difference between two or more synchronous generators
(See Figure 3.21). Especially as the synchronizing torque component, which is one of the two important
aspects of stability, is directly linked to the angle difference between them. From a qualitative point of
view, the effect of increasing or decreasing amplitude of power swing (active power oscillation) following
faults can be the cause of angle stability or instability.

V1 ∠ P
V2 ∠ 0
G G

X
Figure 3.21 Two equivalent machine system and example power swing oscillation.
It is well known that the amplitude of the power swing depends on the dynamic behaviour of loads
following faults and their location in the network. For example, if loads are connected near the power
station generators which are accelerated following faults, the constant power load gives better damping
effect compared to the constant impedance load, because the dynamically larger loads can play a role
of a dynamic braking system and because the constant power load, during voltage sag, consumes more
active power compared to the constant impedance load case. There also may be IBGs in the vicinity,
residential PV in particular. The most critical case is the disconnection of PVs right after the fault occurs.
Although LVRT capability has been regulated in many countries, a large number of installed PVs, do
not meet LVRT requirement. The disconnection of PVs is equivalent to increase in net loads. The
momentary cessation of power output for the duration of the voltage sag also results in increase in the
net loads. If PVs are connected near a huge power station, the power transfer over the transmission line
from G1 to G2 (See pink arrow and dotted red line in Figure 3.22 (a)) after the disconnection of PVs
could decrease because the increased net loads near the power station consume more active power
coming from the power station as shown in Figure 3.22 (a). Conversely, if PVs are connected near a
load centre, the power transfer over the transmission line from G1 to G2 (See pink arrow and dotted red
line in Figure 3.22 (b)) could increase after the disconnection of PVs because the increased net loads
near the power station require more active power coming from the power station as shown in Figure
3.22 (b). The increase in power transfer over the transmission line from G1 to G2 could lead to a growing
or poorly damped power swing oscillation. Therefore, the dynamic behaviour of active power output of
IBGs is a key factor in the case of transient stability studies and their network location is another key
factor in these studies.

PG[MW] PG-0.5PL PG-0.5PL PT

G1 G2 G G

Before Load Before Load

stoppage of PV 0.5PL stoppage of PV 0.5PL

PG[MW] PG-PL[MW] PG[MW] PT[MW]

G1 G2 G G

After stoppage Load After stoppage PL=PG+PT Load

of PV PL[MW] of PV PL[MW]

(a) PV is located near power source (b) PV is located near power sink
Figure 3.22 Change in power flow over transmission lines before and after disconnection of PVs at
different locations [16].

65
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

As shown in Figure 3.23, the LVRT requirement improves transient stability when IBGs are connected
near the power sink (See solid line in Figure 3.23 (b)), while it deteriorates the transient stability when
IBGs are connected near power source (See solid line in Figure 3.23 (a)). In other words, IBGs without
the LVRT requirement near power source could improve the transient stability (It is noted that IBGs
which momentarily cease the current injection also show equal efficacy). It is noted that PVs are equally
distributed throughout the network and only PVs near the fault point mainly influence the transient
stability. Therefore, the influence of the fault location on transient stability is the same as the influence
of the IBGs location on transient stability.
200 75

150 70
内部相差角（deg）

内部相差角（deg）
Full
脱落なし F RT
angle [deg]

angle [deg]
100 65 Full F RT
FRT2
2016年 脱落なし
FRT2
2016年
50 FRT1
2012年 60
FRT1
2012年
w/o FRT
現在ﾓﾃﾞﾙ
Rotor

Rotor
0 55 w/o FRT
現在ﾓﾃﾞﾙ

-50 50

-100 45
0 5 10 0 5 10
時間（秒）
Tim e [s] 時間（秒）
Tim e [s]

(a) Fault point is located near power source (b) Fault point is located near power sink
Figure 3.23 Rotor angle of G1 with relative to centre of inertia with various level of LVRT
requirements [16].

3.3.4 Real life example - Loss of 1200 MW of solar PV caused by faults in southern
California [58]
On August 16, 2016 there was a fire (Blue Cut fire) in the mountains in Southern California that quickly
moved to the transmission line corridor that is comprised of three 500 kV lines owned by Southern
California Edison (SCE) and two 287 kV lines owned by Los Angeles Department of Water and Power
(LADWP). By the end of the day, the SCE transmission system experienced thirteen 500 kV line faults,
and the LADWP system experienced two 287 kV faults as a result of the fire. Four of these fault events
resulted in the loss of a significant amount of solar photovoltaic (PV) generation. The most significant
event related to the solar PV generation loss resulted in the loss of nearly 1,200 MW. The value of the
generation loss was determined by the SCADA system. There were no solar PV facilities de-energized
as a direct consequence of the fault event; rather, the facilities ceased output as a response to the fault
on the system.
The Western Interconnection frequency reached its lowest point of 59.867 Hz, shown in Figure 3.24.

Figure 3.24 Western interconnection frequency during fault [58].

66
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The frequency recovered about seven minutes later (not shown). Notice the second frequency graph is
of a smaller time frame to accent the primary frequency response characteristics.
The four faults that caused loss of generation were: one fault line-to-line and three faults line to ground.
They were cleared normally with approximately the same fault clearing time and fault magnitude. The
largest loss of generation (1178 MW) was with the line-to-line fault. Its clearing time was 2.49 cycles.
Approximately 66 percent of the generation lost with that fault recovered within about five minutes. Three
PV plants had a sustained loss of 400 MW that did not return until the following day, reportedly due to
curtailment orders from the Balancing Authority. The solar production level did not return to its pre-
disturbance level.
The August 16, 2016 event drew attention to the issue of inverter disconnects during faults. It appeared
that this was not an isolated incident. Including the August 16 events, SCE/CAISO determined that this
type of inverter disconnect has occurred eleven times between August 16, 2016, and February 6, 2017.
Knowing that this was not an isolated incident and considering the rapid increase in solar installations
in the CAISO Balancing Authority area (BAA), it was determined that these types of inverter disconnect
events could be a potential reliability risk that need to be analysed and mitigated. A joint Task Force
from NERC and WECC was assembled by the NERC Operating Committee to analyse this disturbance,
determine the causes of inverter disconnect and develop findings and recommendations to mitigate
such events in the future.
By analysing the event, it was determined that the largest percentage of the resource loss (~700 MW)
was attributed to a perceived, though incorrect, low system frequency condition that the inverters
responded to by tripping. The perceived low frequency was due to a distorted voltage waveform caused
by the transients generated by the transmission line fault. The inverter phase lock loop (PLL) control
detected a frequency less than 57 Hz and initiated an instantaneous inverter trip. Frequency measuring
network (FNET) data from this disturbance showed that the Western Interconnection frequency did not
actually reach 57 Hz. The Curve Data Points section of the NERC Standard PRC-024-24 indicates an
instantaneous trip for frequencies less than or equal to 57 Hz for the Western Interconnection. Thus, the
inverters were set to trip instantaneously for that level of frequency.
The second largest significant contributor (~450 MW) was determined to be inverter momentary
cessation due to system voltage reaching the low voltage ride-through setting of the inverters.
Momentary cessation is when the inverter control ceases to inject current into the grid while the voltage
is outside the continuous operating voltage range of the inverter. The inverter remains connected to the
grid but temporarily suspends current injection. When the system voltage returns within the continuous
operating range, the inverter will resume current injection after a short delay (typically from 50
milliseconds to one second) and at a defined ramp rate. In the August 16, 2016 1,200 MW loss event,
many inverters momentarily ceased current injection. The time to return to pre-disturbance values
(restoration of output) was at a ramp of approximately two minutes. Figure 3.23 shows this as the
percentage increases gradually after the initial event.

Figure 3.25 SCE solar resonance output SCADA graph [58].

Some inverter manufacturers and generator owners have interpreted the no-trip area of the PRC-024-2
Standard curves to allow momentary cessation, since the inverters are connected to the grid during
these conditions. Thus, in some generator interconnection agreements momentary cessation during

67
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

voltages below than 0.9 per unit or above 1.1 per unit was allowed, which is according to the NERC
Standard PRC-024-2 is a no trip area.
The third largest amount of loss was approximately 100 MW that tripped by inverter DC overcurrent
protection after starting the momentary cessation operation. The exact cause of these inverters tripping
has not been determined and is still under investigation by the manufacturers.
Of the two types of interruption, tripping and momentary cessation, tripping is the most impactful as it
removes the resource from the interconnection for approximately five minutes. If momentary cessation
is restored quickly, the frequency decline is less severe than an equivalent MW amount of tripping.
The California ISO (CAISO) balancing area has experienced a rapid growth of solar photovoltaic (PV)
resources in the recent past. CAISO has recorded a peak of 9,800 MW of utility scale solar PV
generation. During light load days, they have experienced 47 percent of the area load served by utility
scale solar. This widespread disconnection of inverter-connected resources is a significant concern for
CAISO. Additionally, with the proliferation of solar in many balancing areas across the North America,
this issue needs to be resolved to ensure interconnection reliability.
As the result of the investigation of the solar PV generation loss on August 16, 2016, the following
recommendations were made.
Inverter manufacturers that experienced tripping during the Blue Cut fire event have recommended
changes to their inverter settings to avoid the erroneous tripping due to the distorted measurement of
the frequency; this change will add a time delay to inverter frequency tripping that will allow the inverter
to “ride through” the transient/distorted waveform period without tripping. Solar development owners and
operators involved in this event are working with their inverter manufacturers, CAISO and SCE to
develop a corrective action plan for implementation of changes to inverter parameters.
Inverters that momentarily cease output for voltages outside their continuous operating range should be
configured to restore output with a delay no greater than five seconds. NERC should review PRC-024-
2 to determine if it needs to be revised to indicate that momentary cessation of inverter connected
resources is not allowed within the no-trip area of the voltage curves.
Additional recommendations include that a NERC alert should be issued to the NERC registered
Generator Owners (GOs) and Generator Operators (GOPs) to ensure they are aware of the
recommended changes to inverter settings and alert them of the risk of unintended loss of resources.
This alert should include a recommendation for Balancing Authorities (BAs) and Reliability Coordinators
(RCs) to assess the reliability risk of solar PV momentary cessation and take appropriate measures.
NERC should review PRC-024-2 to determine if it needs to be revised to add clarity that outside the
frequency curves is a “may-trip” area (if needed to protect equipment) and not a must-trip area and to
determine if there should be a required delay for the lowest levels of frequency to ensure
transient/distorted waveform ride through.
Additional analysis is needed to assess the risk and consequences of the momentary cessation with
higher penetration of the IBGs. These studies are currently underway in NERC.
3.3.5 Recommended functionalities of IBGs for large voltage deviations
The necessary functionalities of IBGs shown in Chapter 2 are examined in terms of the large voltage
deviations as shown in Table 3.7. It is noted that the “necessity” in Table 3.7 is based on general power
system dynamic studies and there could be exceptions especially when a specific study related to the
large voltage deviation or the study with special controls or special system conditions is performed. The
necessary major functionalities for large voltage deviation are the following:
(1) (4) and (7) DC source control, DC overvoltage protection, reduction of maximum inverter
current when the DC voltage exceeds a certain limit: If a system fault occurs and if the voltage level
is extremely low, the power electronic device in the inverter could be blocked, which is known as the
momentary cessation of the interruption of the current injection [58]. In such a case, the DC voltage will
rise because the DC source continues to provide DC current, while the AC current is controlled to zero.
That could cause an over voltage in the DC link and would trigger the DC over voltage protection.
Without this DC overvoltage protection model, the simulated response of the active and reactive power
can be different from the actual measured response.

68
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.7 Necessary functionalities for large voltage deviations

Category Functionalities Necessity Remark
Yes, if DC link
Control DC source control See (1) above
model is included.
Yes, if DC link Yes for EMT
Control Current control
model is included. See (2) below
Control PLL Yes See (3) below
Control MPPT No
Reduction of maximum inverter current when Yes, if DC link
Protection See (4) below
the DC voltage overcome a certain limit model is included.
Limitation of inverter current’s variation rate
Protection Yes See (5) below
after a fault
Protection Current limit Yes See (6) below
Yes, if DC link
Protection DC overvoltage protection See (7) above
model is included.
Protection Overvoltage/under voltage protection Yes See (8) below
Protection Over frequency/under frequency protection Yes See (9) below
Positive sequence
Protection Protection for detecting balanced fault N/A overvoltage protection
may be used.
Negative sequence
Protection for detecting unbalanced short-circuit
Protection N/A overvoltage protection
fault
may be used.
Zero sequence
Protection for detecting single-line-to-ground
Protection N/A overvoltage protection
fault
may be used.
ROCOF tripping: monitoring the power
Protection frequency variation rate and disconnecting the Yes See (10) below
inverter when it reaches a certain limit [Hz/s]
Protection Vector jump Yes See (11) below
Protection Transfer trip No
Not all the utilities
require to use this
Protection Anti-islanding active detection method Yes
method for IBGs.
See (12) below
Capability P(f) control (over/under frequency) No
Capability Voltage control by means of reactive power Yes See (13) below
Capability Voltage control by means of active power [P(V)] Yes See (14) below
Capability Synthetic inertia Yes See (15) below
Capability ROCOF immunity No
Capability FRT (LV/HV) Yes See (16) below
Zero current injection,
Reactive current
control calculated by
mean of power factor
input, Maximum
Capability Active behaviour during voltage fast variations Yes reactive current
injection, Reactive
current level
depending on voltage
depth
See (17) below
Capability Power oscillation damping Yes See (18) below
Note: Control for grid support functions are examined in the category of the capability.

69
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

(2) Current Control (Inner current control loop): The current control is necessary for large voltage
deviations at the DC link. The converter model is simplified to a current controller model that generates
the desired currents. The electrical controller of this type of model only extracts the essential component
of a more detailed electrical controller, i.e. a decoupled active and reactive power controller via current
control. The controller has a double-loop structure. The outer loop uses the active power and reactive
power references to generate the respective d-axis and q-axis current references for the inner loop. The
inner loop regulates its current output to the respective current reference. In both control loops, PI
controllers are usually used. It is often the case that the inner current control loop is also omitted in the
RMS model mainly because the current control is most likely to be completed in the time step of the
RMS simulation. Such a model is less complex when compared to the detailed generator/converter and
electrical control model, however, it focuses on the fundamental characteristics of the converter outputs
and has been widely applied in many engineering projects for bulk power system dynamic studies.
(3) PLL: When a large voltage deviation occurs, the bus voltage angle could quickly and significantly
change. Such an immediate jump of the voltage angle could delay in tracking the voltage angle change
at the PCC. Such time delay, coming from the response speed of the PLL could result in undesirable
transient positive or negative current (active or reactive power) variations. To maintain stable operation
of a converter, the PLL needs to track the voltage during a network fault, as shown in Figure 3.26 (a). If
the PLL loses track of the voltage, the converter instability could be induced as shown in Figure 3.26
(b). This can cause damage to network equipment and loss of the generator.[57]

Figure 3.26 System voltage with PLL (a) tracks (b) loses voltage [57].
It is noted that most state-of-the-art IBGs may have very short time delays. The change in active and
reactive power of the IBGs following system faults can affect the critical area exchange or the maximum
power transfer with respect to transient stability or short-term voltage stability. It is also noted that the
detailed PLL control block is usually modelled in the EMT model and the simplified control block which
approximates the PLL behaviour is normally modelled in the RMS model because the assumption that
the IBG model can be represented only by a current source can be justified by the fact that the converter
control is much faster than the dynamics/transients of interest.
(5) Limitation of inverter current’s variation rate after a fault: The rate of current variation (di/dt)
needs to be designed to be below the maximum permissible value of the inverter switching devices. It
is often set by selecting appropriate values for the inductance in the main circuitry and therefore it is by
design a constant. In some systems, an adjustable current variation rate exists, which can be
implemented within the current control loop, in order to protect the switching devices from excessive
current stresses. When a short circuit in the power system occurs, the excessive current will flow through
the switching devices which results in extra stresses on the devices. If the temperature at the junctions
of the switching devices rises higher than a fixed internal threshold with a hysteresis, the current
variation rate will be reduced for a certain period of time to protect the switching devices.
Because the operation of this type of protection will lead to the disconnection of the IBGs, the massive
disconnection of the IBG could affect the critical area exchange or the maximum power transfer with
respect to transient stability or short-term voltage stability.
(6) Current limit: When voltage drop is significant, active or reactive power output of the IBGs can
increase quickly. That means, the current of the IBGs is more likely to hit its limit and the active or
reactive power of the IBGs will be restricted so as to control the current of the IBG within a permissible
range. It should be noted that the dynamic behaviour of the active and reactive power of the IBGs could
be different because the control scheme including the active or reactive current prioritisation can vary

70
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

depending on the designer/manufacturer of IBGs and on the grid code. The possibility of hitting the
current limiter also depends on the initial current of the IBGs, i.e. the pre-fault current of the IBGs. If the
initial current of the IBG is high, the recovery of the active power or the increase in reactive power is
more likely to be limited, which results in a different critical area exchange or a different maximum power
transfer and therefore could affect transient stability and the short-term voltage stability.
(8) Over voltage/under voltage protection: If a system fault occurs and if the fault duration is long,
the under voltage protection for both loads and IBGs is likely to operate. After the fault is cleared the
slow voltage recovery may be of a long duration (e.g. due to the massive induction motor loads) the
under voltage protection is likely to operate. The dynamic reactive power (current) support of the IBGs
could cause a significant voltage rise after the fault is cleared. The self-disconnection of loads with the
active power recovery of the IBGs could also cause a significant voltage rise after the fault is cleared.
Such significant voltage rise is most likely to operate the overvoltage protection, which leads to different
critical area exchange or different maximum power transfer and hence could impact transient or short-
term voltage stability.
(9) Over frequency/under frequency protection: If a large power plant is connected to the main grid
via a transmission line and if a short-circuit occurs on that transmission line, not only a large voltage
deviation but also a significant frequency deviation could occur. When massive amounts of residential
and utility-scale IBGs, with improperly designed LVRT/HVRT capability and/or PLL [31] disconnect
following system faults, significantly frequency drop can be observed. If loads are disconnected due to
the underfrequency load shedding scheme, an over frequency could also occur. Therefore, the over or
under frequency protection might operate in case of the large voltage deviation with high penetration of
IBGs.
(10) ROCOF tripping: The frequency is normally measured using the bus voltage. When a system fault
occurs, the bus voltage angle could immediately shift to a different voltage bus angle [31]. Such jump of
the voltage angle could also cause a significantly large ROCOF value. In order to avoid the undesirable
ROCOF relay operation, a filter may be used. However, this can result in a slower operation of the
ROCOF relay in the case of a real frequency drop. The ROCOF tripping is closely related to frequency
deviations. However, frequency instability could occur together with transient instability through high
penetration of IBGs [58]. Therefore, the ROCOF relay might operate due to the large voltage deviation
with high penetration of IBGs.
(11) Vector jump: Jump method is a typical passive anti-islanding protection scheme (See also Clause
3.2.5 (5)). It is highly possible that the voltage phase jumps to another voltage phase when the network
is split into multiple networks. On the other hand, the voltage phase could jump due to a change in the
network configuration although the network is not separated. A typical change in the system
configuration is during a system fault which has a large voltage excursion, i.e. when the system fault
occurs and when the system fault is cleared. As mentioned earlier, the voltage phase jump can lead to
significant change in frequency seen by the IBG, which could lead to a disconnection of the IBG [58].
Hence, such disconnection of IBGs may be represented by the vector jump protection as well as the
over/under frequency protection.
(12) Anti-islanding active detection method: As mentioned earlier, when the magnitude of bus
voltages changes, i.e., if the small reactive power injection from the IBG does not have ability to change
the connected bus voltage, a voltage phase change occurs. However, if the penetration of IBGs is very
high and if the cumulative reactive current injection from the distributed IBGs has ability to change the
voltage magnitude, a negative reactive current injection during the voltage dip could lead to a further
voltage drop (in short-term and long-term).
(13) Voltage control by means of reactive power: This capability is generally required for the steady-
state condition. Therefore, the dynamic reactive power (current) support is not included in this capability.
As mentioned earlier, although the dynamic behaviour is not directly related to this capability, the pre-
fault active and reactive power feed-in of the IBG can affect the fault-on and post-fault dynamic
behaviour of the IBG output. This could also result in a different critical area exchange or a different
maximum power transfer and therefore affect rotor angle and short-term voltage stability.
(14) Voltage control by means of active power: The voltage control by means of active power is
commonly used for residential PV systems in a few countries. The IBG mainly in LV network reduces
its active power when the voltage exceeds a threshold value, e.g., 1.09 p.u. If a significant voltage dip
occurs in the grid, induction motors might stall and loads could be disconnected from the power system.
A large amount of self-disconnected loads can cause a voltage rise. The control speed of this type of
voltage control is slow and the reduction of active power output from PV systems is in the order of tens

71
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

of seconds to a minute. Therefore, transient stability could be affected by voltage control by means of
active power if the simulation time considers the before mentioned study period.
(15) Synthetic inertia: Rotor angle stability consists of two elements: i) synchronizing torque in phase
with rotor angle deviations, and ii) damping torque inphase with speed deviations. The damping torque
can be expressed as Equation 3.3.
2M
D
Td
-D wK D2
s j 0 -
2M M 4M 2
M 2 D
s D + sD + K D  DTm
w0 w0
Equation 3.3
where: M denotes system inertia,
D denotes damping torque,
K denotes synchronizing torque,
S denotes Laplace operator,
Td denotes decay time constant,
Tm denotes mechanical torque.
Equation 3.3, includes the variable of the system inertia, M, which means the system inertia can affect
the damping torque and therefore rotor angle stability. Because the power swing oscillations caused by
large voltage deviations can change with and without synthetic inertia, it could also influence the critical
area exchange or the maximum power transfer with respect to rotor angle and short-term voltage
stability.
(16) FRT (LV/HV): The Low Voltage Ride-Through (LVRT) characteristic clarifies a zone where IBGs
have to stay connected to the grid during grid disturbances depending on the duration and depth of the
voltage sag. IBGs may trip if the voltage is outside the zone. The most conservative way from the stability
point of view is, to trip the IBGs when the voltage is outside the zone. Thus, the LVRT characteristic is
most likely to be represented as an undervoltage protection with different settings for time delays and
voltage thresholds. These protection settings ensure the LVRT and HVRT capability. Although, different
manufacturers have different control strategies for active and reactive power during the fault-on period,
there are common principles. In general, when the voltage drops, the IBGs will reduce its active power
output. In some countries, the active power output is not allowed to be zero during the fault unless the
residual voltage is below a threshold value, e.g., 0.2 p.u. Hence, the IBG is allowed to temporarily stop
the current injection when the residual voltage is extremely low. In other countries, the active power
output is allowed to be zero, while the reactive power output is required to increase. When modelling
the FRT characteristics, it is important to set the up-to-date under voltage limits according to the
information provided by the manufacturer or the grid code requirements.
Because the dynamic behaviour of the IBG output following faults can significantly change with and
without the LVRT/HVRT characteristics, it could also affect the critical area exchange or the maximum
power transfer and therefore transient and short-term voltage stability.
(17) Active behaviour during voltage fast variations: This capability is typically known as dynamic
reactive power (current) support or dynamic voltage support. As already mentioned in the previous
clause, this capability supports the voltage during and following faults by injecting reactive current and
reducing active current. Such dynamic voltage support can change the critical area exchange or the
maximum power transfer and therefore influence transient and short-term voltage stability. It is noted
that the current limit function is extremely related to this function during large voltage variations because
the active and/or reactive currents are more likely to hit their limitation during large voltage deviations.
In such a case, the inverter control modes such as active power priority and reactive power priority play
an important role how the active and/or reactive current hit the current limit, through which the different
dynamic behaviour of active power and reactive power of IBGs can affect power system dynamic
stability.
(18) Power oscillation damping: Although it is not currently implemented into IBGs except HVDC, this
capability can contribute to the mitigation of undamped power swing oscillations or acts to shorten the
decay time of damped power swing oscillations. Such mitigation of power swing oscillations following

72
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

faults can change the critical area exchange or the maximum power transfer and therefore influence
transient and short-term voltage stability.
(19) Misc.: To make a compromise between simulation accuracy and computational efficiency for the
bulk power system studies, while at the same time considering the requirement for electromechanical
transient analysis, a current-source based model has been proposed and is widely applied to the RMS
model. The current source model neglects the dynamics in the prime mover/converter model block. So,
there are no explicit prime mover models in these types of IBG models. Consequently, not only all the
inductances connecting the converter to the power grid but also the detailed PLL block are unlikely to
be modelled. The assumption that a generator/converter model can be represented only by a current
source can be justified by the fact that the converter control is much faster than the transients of interest
and in its steady state (i.e. the end of the control transients) it injects the specified active and reactive
currents. It should be noted that the current source model causes a numerical problem when the voltage
magnitude is very low and the active/reactive current cannot be injected independently. If the injection
of the active current, i.e. active power net load exceeds the voltage stability limit of the system for a low
voltage condition, the model fails to converge.
3.4 BEHAVIOUR IN RESPONSE TO SMALL AND LONG-TERM VOLTAGE DEVIATIONS
A brief introduction of stability studies with the key response relevant to the system faults are
overviewed. Real-life examples with and without the IBGs are illustrated. In addition, the necessary
functionalities of IBGs for the behaviour in response to small and long-term voltage deviation are
discussed here.
3.4.1 Description of phenomena
Long-term voltage instability is mainly driven by automatic load tap changers and over-excitation limiters
and develops from a few seconds to tens of minutes after a disturbance. Most long-term voltage
instability incidents are as a result of transmission and/or generation equipment outages, regardless of
the severity of the initial fault. The disturbance could also be a sustained load build up, like morning load
increase, and to avoid islanding, one of the possible defensive actions is to introduce a delay on the
Q(V) regulation rules of IBGs; to avoid interferences and, as a consequence, further long-term voltage
stability problems, this delay must be limited to a maximum value, so it does not interfere with on load
tap-changers operation times.
In many cases, static analysis can be used to estimate stability margins, to identify factors influencing
stability, and to screen a wide range of system conditions and a large number of other scenarios. Where
timing of control actions is important, this analysis should be complemented by quasi-steady-state time-
domain resolved simulations [30]. It does not mean the problem is of static nature; it is a dynamic
problem but it can be assessed in some systems using static power flow calculations. This condition is
no longer valid when automatic controls are activated in response to the system evolution or when these
controls may be beneficial for voltage stability (e.g. automatic shunt compensation switching).
With more and more systems relying on post-disturbance controls (or system integrity protection
schemes), the need for dynamic simulations will increase, particularly for real-time dynamic security
assessments.
The computational power and tools available today allow the simulation of long-term dynamics (in the
range of 5 to 10 minutes simulation time) performed efficiently.
Voltage stability refers to the ability of a power system to maintain steady-state voltages at all buses in
the system after being subjected to a disturbance from a given initial operating condition or in the case
of gradual increase in loads (a typical scenario to cause voltage instability). It depends on the ability to
maintain/restore equilibrium between load demand and generation supply from the power system.
Instability that may result occurs in the form of a progressive fall or rise of voltages of some buses. A
possible outcome of voltage instability is loss of load in an area, or tripping of transmission lines and
other elements by their protective systems leading to cascading outages. Loss of synchronism of some
generators may result from these outages or from violation of field current limit to out-of- step conditions
[30], [59].

73
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The term voltage collapse is also often used 24 as the process by which the sequence of events
accompanying voltage instability leads to a blackout or abnormal low voltages in a significant part of the
power system [11], [60], [61].
Voltage stability problems normally occur in heavily stressed systems. While the disturbance leading to
voltage collapse may be initiated by a variety of causes, the underlying problem is deficit(or excess) of
reactive power in the power system. In addition to the strength of transmission network and power
transfer levels, the principal factors contributing to voltage collapse are the generator reactive
power/voltage control limits, load characteristics, characteristics of reactive compensation devices, and
undesired action of voltage control devices such as transformer on-load tap changers (OLTCs) [11].
For a load demand, higher than the maximum generating power, control of power by varying the load
would be unstable, i.e., an increase on load admittance would reduce power consumed. In this region,
the load voltage may or may not progressively decrease depending on the load-voltage characteristic.
With a constant-admittance load characteristic, the system condition stabilises at a voltage level that is
lower than normal. On the other hand, if the load is supplied by a transformer with OLTC, the tap-
changer action will try to raise the load voltage, which has effect of reducing effective impedance of the
load. This lowers the voltage regulator still further and leads to a progressive reduction of voltage. This
process leads to the phenomenon of voltage instability.
The analysis of voltage stability for a given system state involves the examination of two aspects [11],
[62]:
(a) Proximity to voltage instability: How close is the system to voltage instability?
Distance to instability may be measured in terms of physical quantities, and reactive power reserve. The
most appropriate measure for any given situation depends on the specific system and the intended use
of the margin; for example, planning versus operating decisions. Consideration must be given to
possible contingencies (line outages, loss of a generating unit or a reactive power source, etc.).
(b) Mechanism of voltage instability: How and why does instability occur? What are the key factors
contributing to instability? What are the voltage-weak areas? What measures are most effective in
improving voltage stability?
Time domain, in which appropriate modelling is included, can capture the events and their chronology
leading to instability. However, such simulations are time-consuming and do not readily provide
sensitivity information and the degree of stability.
System dynamics influencing long-term voltage stability are usually slow. Therefore, many aspects of
the problem can be effectively analysed by using static methods, which examine the viability of the
equilibrium point represented by a specified operating condition of the power system. The static analysis
techniques allow examination of a wide range of system conditions and, if appropriately used, can
provide much insight into the nature of the problem and identify the key contributing factor. Dynamic
analysis, on the other hand, is useful for detailed studies of specific voltage collapse situations,
coordination of protection and controls, and testing of remedial measures. Dynamic simulations also
examine whether and how the steady-state equilibrium point will be reached.
3.4.2 Response of IBGs to long-term voltage stability triggered by large voltage deviation
and impact on system performance
3.4.2.1 Long-term voltage response with and without LVRT [63]
An example of long-term voltage stability dynamic analysis is shown in a large-scale combined
transmission and distribution network model based on the Nordic system [62]. It is noted that the
phenomena being studied in this example is the combination of “large voltage deviation” and “small and
long-term voltage deviation” rather than “small and long-term voltage deviation”. The transmission
network (TN) model (presented in Figure 3.27) is expanded with 146 distribution networks (DNs) (shown
in Figure 3.28) that replace the aggregated distribution loads.

24
Voltage instability can arise: on fault event or started by a “normal” event like opening a line or spontaneously by very slow
phenomenon such as load increase or decrease

74
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3.27 Expanded Nordic system [65].

Figure 3.28 Detailed distribution network model [63].

75
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Each one of the 146 DNs is connected to the TN through two parallel transformers equipped with On
Load Tap Changing (OLTC) devices. Each DN includes 100 buses, one distribution voltage regulator
equipped with OLTC, three PV units, three type-3 WTs, and 133 dynamically modelled loads (such as
combination of small induction machines and exponential loads). In addition, each DGs complies with
the LVRT requirements sketched in Figure 3.29. More information on the models and data used, can be
found in [63].
To test the long-term voltage stability of the system, a 5-cycle 3-phase fault near the TN bus 4032
cleared by the opening the faulted line 4032-4042 was considered. The system was simulated for 180
seconds with a time-step size of half cycle (10 milliseconds in a 50 Hz system). The simulation was
performed twice. Firstly, when the LVRT curve of an IBG was violated, automatically disconnecting the
IBGs; and secondly, with all the IBGs remaining connected throughout the entire simulation.

Figure 3.29 LVRT capability curve of RES [65].

Figure 3.30 shows the voltage at the TN bus 4044 for both simulations.

Figure 3.30 Voltage on TN bus 4044 [65].

Both cases are short-term stable. After the electromechanical oscillations have died out, the system
evolves in the longer-term under the effect of OLTCs acting to restore distribution voltages, and the
over excitation limiters on the generators. It can be seen that, while second case is long-term stable,
the first one is unstable with the system collapsing at t ≈ 150 s.
Figure 3.31 shows the voltage evolution of the terminal voltage of two IBGs in DN1 and DN2, where in
the first case the LVRT is violated and in the second not. In the first case, the successive disconnection
of IBGs in accordance with the LVRT is reflected on the voltages and leads to the voltage collapse
shown in Figure 3.30.

76
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3.31 Terminal voltages of two RES units in two different DNs
(The one in DN1 violates the LVFRT curve while the other not) [63].
Figure 3.32 shows the total active power generated by IBGs. It can be seen that the IBG disconnection
leads to losing approximately 140 MW. As the IBGs disconnect, the DNs import the lost power from the
TN and this increased TN-DN power transfer leads to depressed TN voltages. Moreover, the OLTCs act
to restore distribution voltages and consequently the consumption of voltage dependent loads. This
leads to a further voltage depression at the TN level until the system collapses. On the other hand, in
second case the IBGs remain connected to the DNs throughout the simulation, thus supporting the
system and the long-term voltage collapse is avoided.

Figure 3.32 Total active power generation by RES in all the DNs [65].
This complex interaction mechanism shows the necessity for detailed representation and validation of
the protection mechanisms in dynamic simulations. The sequence of discrete events, like IBG
disconnections, OLTC actions, the behaviour of DN components and controls, and the interactions of
DNs with the TN or between them, dictate the system evolution.
The simulations were performed with RAMSES, a dynamic simulator developed at the University of
Liège [66].
The long-term voltage stability often follows the short-term studies used to determine the post-fault
system topology (including the disconnection of IBGs due to FRT or other protections and the final
control settings). The post-fault system is then used in the long-term studies, making it unnecessary to
include the short-term protections and fast acting controls. If engineers want to combine both short- and
long-term studies, then the features included in Table 3.7 and Table 3.8 shown in Clause 3.4.3 should
be combined to capture all the necessary dynamics. Failure to do so could lead to the optimistic post-
fault being used and incorrect results.
3.4.3 Recommended functionalities of IBGs for small and long-term voltage deviations
The necessary functionalities of IBGs shown in Chapter 2 are examined in terms of the small and long-
term voltage deviations as shown in Table 3.8. The necessary functionalities of IBGs are examined in
terms of the small and long-term voltage deviations concerning a post-fault phenomenon. For this
reason, system fault reactions (FRT) of IBGs are not considered in this chapter.

77
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.8 Necessary functionalities for small and long-term voltage deviations
Category Functionalities Necessity Remark
Control DC source control No
Control Current control No
Control PLL No
Control MPPT Yes See (1) below
Reduction of maximum inverter current when
Protection No
the DC voltage overcome a certain limit
Limitation of inverter current’s variation rate
Protection No
after a fault
Protection Current limit Yes See (2) below
Protection DC overvoltage protection No
Protection Overvoltage/under voltage protection Yes See (3) below
Protection Over frequency/under frequency protection No
Positive sequence
Protection Protection for detecting balanced fault No overvoltage protection
may be used.
Negative sequence
Protection for detecting unbalanced short-
Protection No overvoltage protection
circuit fault
may be used.
Zero sequence
Protection for detecting single-line-to-ground
Protection No overvoltage protection
fault
may be used.
ROCOF tripping: monitoring the power
Protection frequency variation rate and disconnecting the No
inverter when it reaches a certain limit [Hz/s]
Protection Vector jump No
Protection Transfer trip No
Not all the utilities
require to use this
Protection Anti-islanding active detection method Yes
method for IBGs. See (4)
below
Capability P(f) control (over/under frequency) No
Capability Voltage control by means of reactive power Yes See (5) below
Voltage control by means of active power
Capability Yes See (6) below
[P(V)]
Capability Synthetic inertia No
Capability ROCOF immunity No
Not used for long-term
Capability FRT (LV/HV) No
post fault analysis.
Zero current injection and
Reactive current control
calculated by mean of
power factor input.
Capability Active behaviour during voltage fast variations No Maximum reactive
current injection and
Reactive current level
depending on voltage
depth.
Capability Power oscillation damping No
Note: Control for grid support functions are examined in the category of the capability.

The major necessary functionalities for small and long-term voltage deviation are the following:
1) MPPT: “stop time for simulation” could be set as over 10 min when the small and long-term voltage
deviation is examined. As mentioned earlier, because the solar radiation cannot be assumed to be
constant for the aforementioned time frame, the change in MPPT signal needs to be considered for this

78
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

type of phenomenon. The change in MPPT signal may be modelled as the change in the active power
reference (Pref) itself.
2) Current limit: When the power system is about to experience long-term voltage instability, any
significant low voltage at a system bus must be observed. Extremely low voltages could lead to a current
limitation of the IBG due to a high current injection.
3) Over voltage and under voltage protection: When the power system is about to experience long-
term voltage instability, any significant low voltage at a system bus must be observed. Extremely low
voltages could trip the under voltage protection.
4) Anti-islanding active detection method: Same as Clause 3.3.5-(12).
5) Voltage control by means of reactive power: This voltage control is in general intentionally set as
slow as possible in order to coordinate with the other voltage controllers, such as tap changers and
reactive power compensators. Therefore, the operation of the voltage control by means of reactive
power of IBGs can play an important role when the power system experiences long-term voltage
instability. It should be highlighted that the voltage control by means of reactive power is required not
only for the inverter control but for the plant control (See Chapter 5).
6) Voltage control by means of active power: Without the large voltage deviation, the voltage control
by means of active power can be activated when the load decreases and/or when the PV output
increases in steady-state. Once voltage control by means of active power is activated, the post-fault
steady-state operating point keeps moving for a long time. Therefore, the voltage control by means of
active power can play an important role when the power system experiences long-term voltage instability.
3.5 SMALL DISTURBANCE ANALYSIS
3.5.1 Description of phenomena
Small signal stability is defined as the ability of the system to maintain synchronism when it is subjected
to small disturbances. The definition of small disturbance is limited to the case when the response of
the system to such a disturbance can be analysed with linearized equations. The dynamics is generally
analysed in the frequency domain. Small-signal analysis uses linear techniques, based on
eigenvalues/eigenvectors calculations, and provides valuable information about the inherent dynamic
characteristics of power systems and assists in their design [30],[67]. This analysis can be performed
by the linearization of the non-linear power system model at a certain operating state. A linearized power
system model at a given operating point can be represented by Equation 3.4.

Dx  ADx + BDu


Dy  CDx + DDu
Equation 3.4
where, x is the vector of state variables, y is the vector of output variables, u is the vector of input
variables, A is the state matrix, B is the input matrix, C is the output matrix, and D is the matrix which
defines the proportion of input that appears directly in the output variables.
Stability of a non-singular linear system is determined by the eigenvalues of the matrix A. A real
eigenvalue corresponds to a non-oscillatory mode. A pair of complex eigenvalues λ=σ ± jω with a
positive real part corresponds to an undamped oscillatory mode but with a negative real part, the original
system is asymptotically stable. The frequency of the oscillatory mode is given by f= ω/2π.
The damping ratio of the oscillatory mode is given by Equation 3.5.
𝜎
𝜁=
√𝜎 2 + 𝜔 2
Equation 3.5
In addition, the right and left eigenvectors corresponding to the eigenvalues of interest, are the basis of
sensitivity tools to identify the relationships between variables and eigenvalues.
The relationship between the i-th mode and the j-th mode is given by their right eigenvectors. The
magnitude of the right eigenvector element indicates how much the mode excites the specific frequency
component of the power swing oscillation from the viewpoint of the amplitude of the frequency
component. The angle of the right eigenvector element indicates how much the i-th mode and the j-th

79
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

mode are synchronised during the power swing oscillation from the viewpoint of the phase of the
frequency component, which can be used to group the generators to two or more subgroups.
In a power system with high IBG penetration, some of the synchronous generators are replaced with
IBGs, e.g. PV units. The addition of the PV and other inverter-based resources are certain to occur in
the future, hence the exact effect of these systems, particularly on the transmission systems, must be
adequately studied from the small signal stability perspective [68].
When a system is no longer small signal stable, a power swing oscillation is initiated even by the natural
change in loads and the growing power swing oscillation cannot be damped without the power system
oscillation damping devices, such as PSS. The small signal stability is theoretically violated when the
angle difference between the two buses exceeds 90 degrees. The power-angle curve is often used for
understanding the relationship between the synchronizing torque coefficient and the operating point on
the power-angle curve. The synchronizing torque coefficient is defined as the slope of the power-angle
curve. Because the slope of the power-angle curve becomes zero when the angle difference becomes
90 degrees, a positive synchronizing torque coefficient is the condition for which the small signal stability
is not violated.
The system faults are not normally considered as small disturbances. However, even the single-line-to
ground fault in the bulk power system could be treated as an equivalent small disturbance. Therefore,
the definition of a small disturbance very much depends on how little an impact the disturbance provides
to the power system.
Once the growing power swing oscillation occurs, it is difficult to initially detect such oscillation, because
the power swing oscillation could be almost imperceptible due to the existence of steady-state
oscillations caused by natural load changes. When the growing power swing oscillation is detected, it
may be too late for any corrective actions. In light of this, the small-signal stability study is of growing
interest to power system operators.
3.5.2 Relationship between behaviour in response to small signal stability and relevant study
[36]
It is paramount to analyse the impact of massive addition of PVs both on distribution and transmission
systems. The effects of the high penetration of PVs on distribution systems are well described in [69]–
[72], where it is suggested that the high PV penetration can affect the voltage profile depending on the
loading conditions and the amount of the PV penetration.
However, the effect of PVs on the transmission systems can no longer be neglected. The objective of
the small signal stability analysis is to examine stability of the system under various PV penetration
levels. In order to locate the critical modes of the system, an eigenvalue analysis is conducted to explore
poorly damped modes within the frequency range of 0.1–2 Hz. According to [73] a damping ratio below
3-5% is critical. The objective of such analysis is to investigate whether the increase in PV penetration
affects the critical modes of the system. The following steps are proposed to undertake this type of
investigation:
1) Identify the most critical, i.e., poorly damped, modes of the system by performing eigenvalue
analysis on the base case with no PV generation.
2) Perform eigenvalue analysis for the cases after introducing various levels of residential rooftop
and utility scale PVs.
3) Compare the results of the eigenvalue analysis under different PV penetration levels to investigate
the impact of high PV penetration on small signal stability of the system under study.
4) Perform eigenvalue sensitivity analysis with respect to the displaced generators’ inertia to validate
the results achieved from the eigenvalue analysis.
5) To confirm the aforementioned critical eigenvalue, analyse the transient stability performance of
the system applying less severe disturbance and examine whether the identified critical modes
can be excited and substantiate the results obtained by eigenvalue analysis.
6) Check if the concentration of a large IBG within a critical area within a power system is affecting
the steady-state stability limits of that particular network area. Therefore, usually when performing
the steady-state stability limits the grid operators are requesting to include the involved IBGs
within the grid model developed for studies analysing the power exchanges between critical
areas.

80
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

3.5.3 Response of IBG to small disturbance analysis and impact on system performance
Unlike synchronous generators, IBGs do not interact with the power network through an internal power
angle. Therefore, there are no state variables such as angle, associated with the IBG model in the right
eigenvector (mode shape).
IBGs that are asynchronously connected to the grid will not cause electromechanical modes of
oscillation in their own right. However, introduction of large amounts of IBGs does have the potential to
indirectly change the electromechanical damping performance of the system by:
 Significantly altering the dispatch of synchronous generation in order to accommodate the IBG. If
the small signal stability is adversely affected by e.g. PV integration, critical synchronous generators
may need to be kept on-line or other countermeasures are required to maintain sufficient damping
for low frequency oscillations
 Significantly altering power flows in the transmission network
 Interacting with synchronous machines to change the damping torques induced on their shafts.
This factor depends on the dynamic performance characteristics of the inverter and on other
relatively fast-acting controls (e.g. STATCOMs which may be installed for voltage control
purposes.)
 The presence of the IBG may change the mode shape of the inter-area mode for the synchronous
generators that are not displaced by the IBG.
 The replacement of the synchronous generators by the IBGs might cause reduction in the system
inertia especially with a high penetration of IBGs.
 The small signal stability as well as transient stability needs to be evaluated from two metrics;
damping torque coefficient and synchronizing torque coefficient ([11],[74]). The displacement of the
synchronous generators with a low penetration of IBGs could lead to positive effect in terms of
synchronizing torque coefficient perspective, because the net load of the system becomes lighter
and therefore, the largest angle difference between the generators will be decreased. The case
study introduced in [68] shows that the eigenvalue frequency increases and damping ratio also
increases when the PV penetration rate changes from 10% to 20%.
 Both the first generation WECCIBG models and the IEC 61400-27 wind turbine generator models
[77] have not been developed explicitly with eigenvalue calculation (for small signal stability) in
mind. These IBGs include highly non-linear components and simplifications in the development of
the models. Thus, linearisation for eigenvalue analysis is not trivial nor necessarily appropriate
based on these simplified models. Some publications [75] refer to special cases where there is a
problem in linearizing the system matrix for the eigenvalue analysis when the first generation
WECC generic models is used. To avoid the problem, an extra term in the models may be
introduced, but there is no guarantee that similar problems may not occur in the future. Further
information to [75] can be found in [76]
 The authors in [68] proposed another approach for integration of variable energy resources in terms
of small signal stability. The approach suggested that for every 3-MW addition of renewable
generation to the system, there would be a 2-MW reduction in conventional generator commitment
and 1-MW reduction in their dispatch. While the choice of the cited “1/3–2/3 rule” could be quite
arbitrary, which can lead to potential small signal stability problems (note that the overall system
inertia is decreased). Consequently, with displacing/rescheduling of conventional units as a result
of the addition of PV generation, it is advantageous to determine if a particular generator’s inertia
has a significant impact on a particular inertial oscillation mode. This could be determined by
performing sensitivity analysis with respect to generator inertia and performing de-
commitment/rescheduling using the sensitivity to inertia as a constraint [68].
3.5.4 Recommended functionalities of IBGs for small disturbance analysis
The necessary functionalities of IBGs shown in Chapter 2 are examined in terms of the small signal
stability as shown in Table 3.9. Protections and limiters are ignored in this table due to the way this
analysis is performed, i.e. using small increments around the operating point. The rest of the controls
and capabilities need to be implemented to the IBG models as shown in Table 3.9. The main point here
is that the same system and control data for transient stability analysis and small signal stability analysis
are generally used. Unless a functionality is required for transient stability, it should be also required for
small signal stability.

81
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.9 Necessary IBG’s functionalities for small signal stability analysis
Category Functionalities Necessity Remark
Yes, if DC link
Control DC source Control model is
included.
Control Current control No
Control PLL Yes
Control MPPT No
Protection ALL functionalities related to protection No
Capability P(f) control (over/under frequency) No
Capability Voltage control by means of reactive power Yes
Capability Voltage control by means of active power [P(V)] Yes
Capability Synthetic inertia Yes
Capability ROCOF immunity No
Capability FRT (LV/HV) No
It may be non-required
functionality especially
when the voltage level is
not much decreased.
Even this functionality is
Capability Active behaviour during voltage fast variations Yes
implemented in the
model, it is highly likely
that this functionality is
not even activated for
tiny voltage dip.
Capability Power oscillation damping Yes
Note: Control for grid support functions are examined in the category of the capability.

3.6 UNINTENTIONAL ISLANDING OPERATION

3.6.1 Description of phenomena and mechanism of island formation
Islanding is when a portion of a power network is disconnected from the main supply, where the local
loads are entirely supplied by the local embedded generation and where the voltage and frequency
levels are maintained within permissible limits around nominal values.
Islanding itself is not necessarily undesirable, if intentional, but unintentional islanding can have
undesirable impacts on customer and utility equipment integrity. Some countries strictly prohibit the
unintentional islanding in terms of safety and it is regulated to prevent the unintentional islanding in a
few seconds after islanding.
To understand how an unintentional island may form, consider the schematic representation shown in
Figure 3.33, representing a local IBG and load which are normally connected to the main power system
through a circuit interrupted by the switch.

Figure 3.33 Schematic representation of unintentional islanding.

In order for this system to enter a sustained unintentional island when the switch is opened, the
fundamental-frequency active and reactive power (Pgrid, Qgrid) must be nearly zero at the moment
when the switch is opened. This means that the IBG output and the local load demand must match
closely in terms of both real and reactive power. If this is not the case, either the voltage or the frequency

82
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

will quickly drift outside of normal operating range when the switch opens, and the Loss of Mains (LOM)
condition can be easily detected. If such a balance does exist, then the island may “self-excite,” in the
sense that the IBG output current flowing into the load gives rise to a voltage that appears sufficiently
similar to the grid voltage that the IBG cannot detect the difference. The loading balance conditions that
could result in unintentional islanding is referred to as a non-detection zone (NDZ).
Theoretically, any subsection of the local electric power system that contains both IBG and loads, and
can be fully isolated from the utility voltage source by automatic protection/control or operator action,
can be considered a potential island. If a particular feeder contains downstream reclosers, sectionalizing
switches, or other circuit interrupters, the network section that is isolated by these devices would be a
“potential island”. If an IBG system is within the customer premises, the customer premises themselves
could be a potential island. This mechanism for the formation of an "intentional" island is in fact quite
complicated, using appropriate load reductions and / or adjustments on the generation in the cases of
connection as shown in Appendix 3-A.
If unintentional islanding conditions continue for a significant period of time (sustained islanding),
personnel safety could become a concern. Even if the unintentional islanding period is short (temporary
islanding), the potential degraded power quality could still be a concern.
3.6.2 Relationship of behaviour in response to islanding to relevant study
3.6.2.1 Role of power balancing
In order for an island to be sustained, both the active and reactive power demand of the load and power
system components must be satisfied. Since most loads and power system components consume or
absorb reactive power, there must be a source of vars in the potential island in order for islanding to be
sustained. The most obvious var source is capacitance, which may be deliberately added for power
factor correction or may arise as a parasitic form from underground cabling. Many of today’s residential
PV inverters are designed to operate at unity power factor, but, increasingly, larger inverters are being
equipped with the ability to operate at a fixed power factor according to a schedule or command. In this
case, the inverters may produce or absorb reactive power. If the demand of reactive power is larger
than the supply of reactive power in the island, the system voltage in the isolated island decreases in
case of the unintentional islanding condition due to insufficient var supply in the island. That leads to
decrease in active power load consumption in the isolated island due to the load voltage characteristics.
Such decrease in active power load in the island leads to prompt frequency rise beyond the mandated
limit and over frequency protection will operate. Thus, the reactive power balance in the island can
indirectly affect the active power balance in the island. Therefore, the risk of a sustained islanding is
very negligible if the demand of reactive power is larger than the supply of reactive power in the isolated
island.
3.6.2.2 Presence of rotating generators
If a potential island includes both rotating and inverter-based DGs, the case should be analysed
carefully. The increase of the inertia associated with the rotating generation leads to a significant
slowdown of the voltage and frequency variations, increasing the risk of possible counter-phase
reclosing events (i.e. the condition that the voltage angle difference between the two separated systems
becomes 180 degrees). This also applies in the case where there is a rotating load connected to the
islanded network, because its inertia is to be added to the inertia of the system. In such particular cases
dedicated power system studies are requested by grid operators in order to highlight the special
conditions for which the island operation will be avoided.
3.6.2.3 Automatic reclosers of network protections
The main purpose of the procedure for automatic selection of faulted line sections in MV feeder is the
fast fault detection and the isolation of the faulted section, and the automatic supply of the healthy
sections. As an example, in Italy MV distribution feeders are equipped with circuit breakers from the
establishment of the system, operated from maximum current relays (50/51) and directional earth-fault
relays located in HV/MV Primary Substation. These protection relays perform automatic three pole
reclosing cycles both involving the whole feeder, and sections of it, in case of MV network automation
(switch disconnectors along the feeders may open and close in coordination or not with the automatic
reclosing cycle of the CB, depending on the way the network is operated (isolated/compensated) and
from the nature of the fault (overcurrent/earth fault) in order to reduce interruption number and duration.
This type (three-pole) of automatic reclosure procedure may introduce further risks associated with the
possibility of out-of-synchronism reconnection of two separate grids, also in case of temporary islanded
operations (some seconds or hundreds of ms).

83
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The voltage phase angle within an island is not synchronized with the voltage phase angle of the
interconnected grid. This is not a problem in itself but if an undetected islanded network is reconnected
to the main supply system, this will lead to high transient currents. The jump of the voltage vector directly
affects connected electrical machines, which may then damage mechanical drives or prime movers such
as gas or steam turbines or combustion engines. Usually, no central control for frequency and voltage
is available within a local island and no synchro-check is installed at breakers at HV/MV substations, so
a shock-free resynchronization is not possible.
Anti-islanding detection studies are performed through field tests and/or computer simulation studies. It
is often the case that a field test is first performed, the mathematical simulation model is created second
and then verified through the field test. The examination of the anti-islanding detection is then performed
using the validated mathematical simulation model through the time-domain simulation. In Japan for
example, the certification test for the anti-islanding detection performance is performed in an authorised
laboratory, not in the time-domain simulation. The type of the certification test is provided by the
certification test standard. Conversely, the replacement of the overvoltage ground relay with the anti-
islanding protection especially in LV network may be accepted in exceptional circumstances by the
utility. Once the application is submitted by the IBG owner, the corresponding utility initiates the
examination study and the time-domain simulation is sometimes utilised under the various power
balance at the PCC in order to ensure that the anti-islanding can be resolved in a specified time duration
following islanding no matter how much the power balance at the PCC occurs.
Both the RMS model and EMT model are used for anti-islanding detection studies. In general, the test
system model is not as large as the model for the bulk power system stability studies. Therefore, the
constant voltage source may be assumed and the EMT model can be used for the anti-islanding
detection study. Some anti-islanding detection methods use the harmonic current injection. Because the
RMS model cannot represent harmonics, the use of EMT model is the only option for this type of study.
Alternatively, an induction motor model and an intricate generator control model sometimes need to be
used. The interference between one anti-islanding protection and another anti-islanding protection also
needs to be examined in the study. If the number of anti-islanding protection systems becomes too
large, it takes an extremely long time to complete the study. In addition, the number of the simulation
cases can be almost infinite when the anti-islanding active detection method is examined and when the
reactive power injection is not synchronized between one IBG and another IBG 25. It should be noted
that the anti-islanding detection performance could change depending on the reactive power injection
from IBGs. The RMS model is sometimes used for the anti-islanding detection study especially in the
case of the high penetration of IBGs.
Using RMS model for anti-islanding detection studies implies developing a time simulation dynamic
model of the network including the following key elements:
 a load flow simulation file including the details of the analysed area in terms of load and generation
profile;
 a dynamic simulation file including the details of all generating units and induction motor loads (if
any) within analysed area (IBGs and synchronous generators if both types are present);
 an incident simulation file for representing the loss of network element/elements which may lead to
the formation of an islanded area.
Usually the aforementioned files are specific for each of the following three different scenarios:
 un-intentional islanding when the analysed area has a power deficit of about 5%-10% from the total
load;
 unintentional islanding when the analysed area has a power excess of about 5%-10% from the total
load;
 unintentional islanding when within analysed area the generation profile is matching the load profile.
The results of the dynamic simulation are analysed using the following profiles: frequency, active power
(generated and consumed), voltage and reactive power (generated and consumed). The interpretation
of the results is focused on:

25
For example, in the case of reactive power variation method, the reactive power injection forms a sinusoidal wave, the
frequency component of which is about 5 Hz. The reactive power variation starts when the IBG is launched. Therefore, if there
are more than 2 IBGs, the phase difference of the reactive power variation injection can be in the range of 0 degree and 360
degrees. When the phase differences are assumed to be 360 types, 360 cases need to be examined. It is most likely that the
necessary number of cases could be nearly infinity if the number of IBGs becomes 10.

84
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 verifying if the frequency or/and voltage protection thresholds associated to the IBGs within
analysed area are reached or not and
 computing the ROCOF [Hz/s] – usually this should be identified for both situations: P(f) control
capability ON and OFF associated to the analysed IBGs.

3.6.3 Response of IBG to islanding and impact on system performance

In order to prevent islanding, various types of anti-islanding protection have been introduced. Those
protections are implemented in the inverter as an additional function and the IBGs execute an internal
algorithm producing a small perturbation on the network and checking the response of the grid. If the
response exceeds a certain limit, the inverter is automatically disconnected.
Anti-islanding (loss of main) passive protections/detection methods may include:
 ROCOF tripping: monitoring the power frequency variation rate and disconnecting the inverter when
it reaches a certain limit [Hz/s];
 Vector jump;
 Switch to narrow frequency band (on external command/signal or following the action of protection
listed in Clause 2.2.3, and/or ROCOF, vector jump, etc.);
 Transfer trip.
The following protection functions and their combinations may also assure anti-islanding (loss of main)
functionalities.
 Under/over voltage protection:
 Over-voltage 10 min mean protection;
 Positive sequence under-voltage protection;
 Negative sequence over-voltage protection;
 Zero sequence over-voltage protection;
 Under/over frequency protection.
TBN: only some of the detection methods are thoroughly described by protection relays relevant
standards (e.g. ROCOF and vector jump relays).
Besides what described above, further active means to detect an island may be present in the IBGs and
accepted from system operators and utilities.
3.6.4 Recommended functionalities of IBGS for unintentional islanding
The necessary functionalities of IBGs shown in Chapter 2 are examined in terms of the unintentional
islanding operation as shown in Table 3.10.
Because most of the functionalities are necessary for the study related to the unintentional islanding
operation, the only unnecessary functionalities for unintentional islanding are the following:
1) MPPT: Anti-islanding is most likely to be detected in less than 2 or 3 seconds after forming islanding.
The MPPT reference varies depending on the change in solar radiation. In general, the change in the
radiation is not assumed for examining anti-islanding detection.
2) Voltage control by means of active power: Anti-islanding is most likely to be detected in less than
2 or 3 seconds after forming islanding. The voltage control by means of active power is generally
provoked a few seconds after the voltage limit is violated, say over 1.09 p.u. Therefore, before such
control action, the islanding detection needs to be completed. In general, the voltage control by means
of active power is not assumed for examining anti-islanding detection.
3) Power oscillation damping: The functionality of the power oscillation damping is more likely to be
associated with HVDC mainly for improving damping of the power swing oscillation. The power swing
oscillation is generally manifested between one large generator group against another generator group.
When islanding is established, at least one generator group will be in the islanded system and the power
swing oscillation would no longer exists in the islanded network.

85
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.10 Necessary functionalities for unintentional islanding operation

Category Functionalities Necessity Remark

Control DC source control Yes
Control Current control Yes
Control PLL Yes
Control MPPT No See (1) above
Reduction of maximum inverter
Protection current when the DC voltage Yes
overcome a certain limit
Limitation of inverter current’s
Protection Yes
variation rate after a fault
Protection Current limit Yes
Protection DC overvoltage protection Yes
Protection Overvoltage/under voltage protection Yes
Over frequency/under frequency
Protection Yes
protection
Protection Protection for detecting balanced fault Yes
Protection for detecting unbalanced
Protection Yes
short-circuit fault
Protection for detecting single-line-to-
Protection Yes
ground fault
ROCOF tripping: monitoring the power
frequency variation rate and
Protection Yes
disconnecting the inverter when it
reaches a certain limit [Hz/s]
Protection Vector jump Yes
Protection Transfer trip Yes
Not all the utilities require to use this
Protection Anti-islanding active detection method Yes
method for IBGs.
Capability P(f) control (over/under frequency) Yes Unnecessary for LV network islanding
Voltage control by means of reactive
Capability Yes
power
Voltage control by means of active
Capability No See (2) above
power [P(V)]
Capability Synthetic inertia Yes
Capability ROCOF immunity Yes
Capability FRT (LV/HV) Yes
Zero current injection and Reactive current
control calculated by mean of power factor
Active behaviour during voltage fast
Capability Yes input. Maximum reactive current injection
variations
and Reactive current level depending on
voltage depth.
Capability Power oscillation damping No See (3) above
Note: Control for grid support functions are examined in the category of the capability.

3.7 OTHER PHENOMENA AND STUDIES

3.7.1 High frequency and low frequency control interaction [31], [76]
The possibility of the interaction between two or more devices is very broad. IBGs may interact with
each other, or they may interact with other power electronic devices such as HVDC ties, FACTS devices
(SVCs or STATCOMs). IBGs may also interact with the non-power electronic based devices such as
series capacitors, switched shunt devices and conventional generators.
The interactions could cause control instability phenomena. The fast, high gain controllers of IBGs
interaction with the power system, with other nearby plants such as HVDC converters, SVCs or
STATCOMs, or with the other nearby Wind Power Plants (WPPs) could be the possible reasons of the

86
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

occurrence of the control instability phenomena. In addition, IBGs which are connected to the weak AC
grid are more likely to encounter this control instability phenomena as open loop gain interaction with
other controllers could be higher than when they are connected and operated in strong AC systems.
Device controller interactions are expected and will be similar to the HVDC and SVC controls
interactions explored and reported by CIGRE WG 14.28. The interaction between the wind
generators/plant which are connected to the weak AC grid and the other power electronic devices are
also discussed in CIGRE B4.62 (See TB 671 [76]). Resonances in the grid can contribute to instabilities
due to interaction between PV inverters and the grid. An impedance-based representation can be used
to study resonances and analyse the influence of grid and inverter control parameters on instabilities
[78], [79]. The Nyquist and Bode criteria are extensively utilized to analyse closed-loop system stability
from the impedance-based representation in the frequency domain [80]-[83].
Sub-Synchronous Resonance (SSR) occurs due to the addition of series compensation onto the system
and Sub-Synchronous Torsional Interaction (SSTI) due to the addition of HVDC. The Sub-Synchronous
phenomena can be classified into three categories as illustrated in Figure 3.34. The potential effect of
both SSR and SSTI on the network is the interaction with generator shafts, and in very severe cases
they can cause generator shaft fatigue and failure. The power electronic control system can interact with
sub-synchronous modes of the network and cause Sub-Synchronous Control Interaction (SSCI). SSCI
can be more pronounced in a network with low short circuit ratios. SSCI can be more likely when these
devices are electrically close to each other or on the same bus bar. It can result in over-voltages, current
distortion, and potential damage to control systems themselves. Other types of Sub-Synchronous
Interactions exist between control systems and the transmission network, and also between control
systems at particular complementary control frequencies; these will become increasingly relevant as
regional levels of IBGs increase.

Figure 3.34 Classification of Sub-synchronous interaction [31].

Power electronic control systems used in Static Var Compensators (SVCs), FACTS devices and wind
turbine control systems, particularly DFIGs radially connected to a series compensated transmission
circuit, can interact with sub-synchronous modes of the network and cause SSCI. This control interaction
can be more pronounced in a network with low short circuit ratios (SCR). It can result in over-voltages,
current distortion, and potential damage to control systems themselves.
The control system interaction is more likely when these devices are electrically close to each other or
on the same busbar. These control interactions differ by regions, and therefore the operational co-
ordination of these devices is important for their effective use and to resolve certain network constraints
(thermal limit, voltage stability or rotor angle stability). With the increasing number of IBGs, FACTS, and
HVDC converters connected electrically and very closely together, having control systems that share
similar input values (i.e. all use bus bar voltage as an input signal to respond to changes), there is a risk
of undesirable control interactions could occur if such behaviour are not studied collectively.
The necessary functionalities for the control interaction related studies are shown in Table 3.11.

87
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.11 Necessary functionalities for controller interaction studies (simplified table)

Category Functionalities Necessity Remark

Control All Yes
Protection All No no protection due to small signal analysis
Capability All Yes

3.7.2 Maximum Penetration of IBGs

The increasing penetration of IBG has changed the dynamic behaviour of power systems. In the future,
even more IBG units are planned to be integrated, eventually displacing conventional synchronous
generators. Therefore, the investigation of the maximum penetration rate of IBGs as well as the
necessary amount of conventional synchronous generators than need to stay connected on system
stability could become one of the critical studies in power system control, operation and planning. In
general, IBGs affect dynamic system stability by changing the pre-disturbance operating conditions as
well as by exhibiting different dynamic response during and right after disturbances. The intermittent
nature of most of the IBG units such as WTG and PV units could introduce higher uncertainty in
determining pre-fault operating conditions raising also the need for evaluation of worst case operating
conditions. The probabilistic stability assessment has proven to be one of the promising ways to do this
type of study due to the appropriate accounting of uncertain parameters affecting their behaviour, e.g.,
system loading, fault location, fault duration, etc. The probabilistic stability assessment is typically
performed using Monte Carlo (MC) dynamic simulations. The maximum penetration rate of IBGs
considering the displacement of the synchronous generators (either disconnection or unloading of
generators) is investigated in [84]-[86]. According to the transient stability study, as shown in Figure 3.35,
a turning point in terms of power swing oscillation behaviour can be observed when the penetration rate
is around 50%, which is likely to be relevant to the maximum penetration rate of IBGs, i.e., the minimum
necessary amount of the synchronous generators.
1
no IBG
0.9 20% IBG
30% IBG 0.15
0.8 40% IBG
50% IBG
60% IBG
0.7
Cumulative Probability

0.6 0.1

0.5

0.4
0.05
-80 -60 -40 -20 0 20
0.3

0.2

0.1

0
-100 -80 -60 -40 -20 0 20 40 60 80
Transient Stability Index
Figure 3.35 Cumulative distribution function of transient stability index
for different spare capacity [84].

According to a frequency instability study, shown in Figure 3.36, a turning point in terms of the dynamic
behaviour/aspect of the frequency drop can be also observed when the penetration rate is around 50%.
In both cases the maximum penetration rate is around 50%. Although the results illustrated in the figure

88
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

could be different depending on the system condition and system configuration 26, those findings can be
a good reference for the system engineers in charge of the similar studies in their own network. Further
details and the relevant results are introduced in [84]-[86] (See also Appendix 3-K).
Figure 3.35 shows the cumulative probability with respect to the stability index. If the overall cumulative
distribution line is lower when the TSI is negative, the fraction of the unstable area will be reduced, which
means the study condition makes the system more stable. As shown in Figure 3.35, as the penetration
rate of IBGs increases, the system becomes more stable as long as the penetration rate of IBGs is
below 50%. Conversely, once the penetration is over 50%, as the penetration rate of IBGs increases,
the system becomes more unstable. Therefore, 50% can be the turning point of the dynamic
behaviour/aspect from the transient stability point of view.
Figure 3.36 shows the frequency nadir with response to the instant penetration of IBG in case of 40%
loading. Generally, the frequency nadir will not be widely distributed unless the total rated capacity of
the connected synchronous generators is equal to or less than the connected IBGs. However, when the
penetration rate of IBGs exceeds 50%, the frequency nadir becomes widely scattered mainly due to the
tripping of additional synchronous generators. Therefore, it is likely that the 50% penetration of IBGs is
the turning point of the dynamic behaviour from the point of view of frequency instability.
Figure 3.37 shows 50 Hz grid frequency on bus 60. An active power disturbance occurs at 1 sec, and
grid frequency drops to 49.3 Hz. After 8 seconds, one of the generators trips and decreases the
frequency nadir to 49.1 Hz.

Figure 3.36 Frequency nadir with response to the instant penetration of RES (40% loading) [86].

Figure 3.37 Grid frequency on bus 60 [86].

3.8 SELECTION OF TYPE OF MODELS [4]

Another aim of this chapter is to collect the features and capabilities of IBGs that are to be considered
for EMT based models and RMS based models used in power system analysis (e.g. stability, protection
coordination, faults, network planning and network operation).

26
The system in Ireland have actually operated safely even when the penetration rate of IBGs is over 50%.

89
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3.12 Selection of type of models for typical power system dynamic studies
Type of Studies RMS EMT
Frequency stability Recommended
Short-circuit provision from IBG Recommended
Low Voltage Ride-Through Recommended Recommended
Transient stability Recommended Recommended
Long-term voltage stability Recommended
Unintentional islanding detection Recommended Recommended
Protection coordination Recommended Recommended

The choice of a simulation model (EMT or RMS) is, however, strictly dependent on the specific
phenomena to be investigated. It is up to the power system engineer to decide which of the inverter
features and capabilities are relevant for the phenomenon under investigation. Since all models have
limitations, the selection of the model type is crucial and should be based on the objectives of the study
to be performed with the model. Therefore, the suitable model type could change depending on the type
of power system dynamic study and the system conditions to be studied.
3.8.1 Necessity of IBG models
The high penetration level of IBG has resulted in the displacement of conventional synchronous
generators. Therefore, the impact of IBG on the dynamic performance of the system increases. The
dynamic characteristic of IBG is different compared to synchronous generators, and with proper control
system design and functionalities of modern IBG technologies, they can provide many of the same or
even better services (e.g. voltage control, frequency response etc.). None the less, they do need to be
modelled differently and properly. Therefore, the development of the proper computer simulation models
for IBG with such additional functionalities is vital for power systems analyses.
Moreover, DSOs have a representation of their networks and have details about connected consumers
and producers to some extent. However, the limited data is generally not suitable for dynamic
simulations for either the distribution or the transmission system or, at least, has not been used for that
purpose in the past. DSOs can benefit time-domain simulations to assess protection system behaviour,
distribution network automated operation, unintentional islanding of part of distribution systems including
IBG, voltage issues, etc. For these types of power system dynamic studies detailed IBG models are
needed.
Therefore, the necessity of IBG models should be clarified for each type of power system dynamic study.
A few examples are given in the following (See also Table 3.12).
3.8.2 Frequency stability
Frequency stability studies are focused on the frequency response of the grid to a large disturbance,
such as the loss of the largest generating unit (or facility) on the system and assess if the resulting
frequency response of the system is stable and avoids under frequency load shedding. Such studies
are typically performed using RMS (positive- sequence) simulation models.
3.8.3 Short-circuit provision from IBG
Short-circuit studies are performed, typically by protection engineers, to identify the setting for protection
relays as well as to assess the capability of existing (or new) circuit breakers to be able to withstand and
interrupt the short-circuit levels that will be seen on the network during fault conditions. Such studies are
typically performed in EMT tools and other software tools specifically designed for short-circuit studies.
Thus, appropriate models of IBG are needed in such tools. An approach on how to calculate the infeed
of the fault current from IBGs is provided in [33].
3.8.4 Low Voltage Ride-Through
To properly assess the actual low voltage ride-through (LVRT) capabilities of IBG, analyses have to be
performed using detail vendor specific EMT type models and simulations. Furthermore, such simulations
may be verified by factory tests of the equipment. It usually then suffices to translate the observed
performance to simplified RMS models that will mimic the low voltage ride-through capabilities of the
equipment for large-scale power system dynamic studies. Thus, both appropriate RMS and EMT models
of IBG are needed.

90
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

3.8.5 Transient stability

For transient stability studies at bulk power system level, validated RMS (or positive-sequence) models
are needed for IBG. However, in some cases EMT models may also be needed when investigating e.g.
the connection of a large IBG power plant to a very weak part of the power system.
3.8.6 Long-term voltage stability
Long-term voltage stability is often analysed using continuous power flow analysis, such as P-V or Q-V
analyses. For such studies, the dynamics of IBG are often neglected, and a quasi-steady-state model
that respects the real and reactive power limits of the IBG is sufficient. If mid-term time-domain dynamic
simulations are performed, then a suitable RMS (positive-sequence) IBG model is needed.
3.8.7 Unintentional islanding
For this type of study, either an RMS or EMT model for the IBG is needed. Some anti-islanding protection
systems utilize the harmonics of the voltage for the islanding detection and therefore, the EMT model
for the IBG is the only option to analyse unintentional islanding. On the other hand, if the anti-islanding
protection system does not consider voltage harmonics and if the power systems model includes many
IBG models, the RMS model for the IBG may be adequate to analyse unintentional islanding.
3.8.8 Protection coordination
Many protection systems need to be coordinated. Depending on the type of protection systems to be
studied, either an RMS or EMT model for the IBG may be needed.

3.9 CONCLUSION
All numerical models, i.e. mathematical simulation models have their limitations. The identification of the
necessary functionality for a specific dynamic behaviour is the key for providing a practical model. The
modelling craftsman is devotedly involved in how to select the necessary functionalities of the IBGs.
Chapter 3 described several major phenomena for the power system dynamic studies with the IBGs.
The representative three phenomena are the dynamic behaviour in response to 1) frequency deviation,
2) large voltage deviation, 3) small and long-term voltage deviation. In order to obtain the appropriate
simulation results for each type of dynamic stability study, the necessary functionalities of the IBGs
which should be implemented in the model are defined and listed in terms of the power system
phenomena instead of the power system dynamic studies.
The functionalities are classified into three categories: 1) control, 2) protection, 3) capability. The detailed
clarification for the necessity of the use of the functionalities is also provided for each functionality. It
should be emphasized that the protection model needs to be more engaged in many power system
dynamic studies mainly because the operation of the protection could directly cause the disconnection
of the IBGs.
Other than the aforementioned phenomena, the unintentional islanding is also discussed in detail. The
anti-islanding protections (i.e. loss of main supply) are widely implemented in the IBGs which are
connected to LV network. The nuisance operation of the anti-islanding protection without the system
separation will be a growing concern in the LV network. Conversely, the prompt disconnection of the
IBGs using the anti-islanding protection is absolutely critical in terms of safety. The difficulty of those
type of studies is also discussed in Chapter 3. Especially when a large number of aggregated IBGs is
connected to the LV network, it is most likely that the number of simulation cases is tremendous.
Furthermore, the control interaction is briefly discussed and introduced referring to TB 671 [76]. Because
of the capability of the rapid control in the IBG for voltage and current and because of the increasing
amount of such rapidly acting controllers including HVDC and STATCOM, interference between the
controllers is starting to be noticed in many parts of the world. The control interaction studies using
detailed EMT models are likely to become popular.
The anticipated change in dynamic behaviour following faults with the high integration of the IBGs is
individually discussed in response to those phenomena. With the high integration of the IBGs, these
phenomena could be close to or occur at the same time.
The real-life examples are introduced for system engineers to realize how much the IBGs could impact
the power system dynamic behaviour following faults. Other studies which are not regularly performed
by the system operators are also highlighted, such as the study for the maximum penetration of the
IBGs. As stated in Chapter 2, IBGs cannot accomplish all the capability that synchronous generators

91
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

have. Therefore, the required amount of percentage of the synchronous generators in the system should
be more the focus from the power system operator's point of view. The new methodology using the
probabilistic approach is also described in Chapter 3.

92
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

4. EMT MODELS FOR INVERTER BASED GENERATION

4.1 INTRODUCTION
There are two commonly used power system simulation models:
 Electromagnetic transient (EMT) models and
 Phasor models or RMS-type models

The electromagnetic transients tools/program has been the most well-known and widely used simulation
tool as a circuit theory-based approach assuming a transverse electromagnetic mode (TEM) since its
original development in the Bonneville Power Administration of the US Department of Energy from 1966
to 1984 [87], [88], [91]. In EMT tools, power system components are adequately modelled to simulate
high-frequency transients in power systems. This makes EMT programs valuable when studying the
effects of power-electronic devices on system behaviour. In order to cover the necessary bandwidth,
these programs use small integration time steps of the order of 50 µs or less, making EMT simulation
much slower than RMS simulation. The common practice in dealing with large systems in EMT programs
is to divide the system into a study zone where the transient phenomena occur and an external system
encompassing the rest of the system, in order to reduce computational burden [92].
EMT and real-time time domain analysis programs solve the differential equations of the electrical
network as compared to transient stability analysis which uses RMS positive sequence phasor equations
to represent the electrical network. This distinction means that EMT analysis is capable of representing:
 The non-linear response of electrical devices (e.g. transformer saturation or surge arresters)
 Frequency dependent effects (generally over a wide range of frequencies, including DC, sub-
synchronous frequencies, harmonics, interharmonics, resonances and higher frequency transients)
 Unbalanced networks (suitable for single line to ground faults, negative and zero sequence
components of transients, etc.)
 Detailed power electronic devices (including switching transients and associated harmonics, as well
as the detailed controls and protection systems)

For IBGs, EMT analysis can be used to model the detailed switching devices (IGBTs, GTOs, thyristors,
diodes etc.) in a converter, as well as the control systems. EMT models are subsequently used to
investigate transient phenomena of power systems such as:
 Switching transients and over-voltages.
 Short-term analysis of the disturbances.
 Transient over-currents.
 Internal and external IBG protection performance.
 Power quality degradation and harmonics.
 Dynamic controls of electric generators and drives.
 Operation of FACTS, HVDC, SVC, STATCOMs, etc.
 Dynamic operation of wind plants, PV inverters, etc.
 Sub-synchronous phenomena (SSR, SSTI, SSCI) and torque amplification.
 Control interactions between complex devices.

Generally, EMT analysis is not required in the initial or feasibility stages of a project and is often only
introduced later in the overall design and study process. EMT models may also be required under
special system conditions where RMS transient stability programs may not be sufficiently accurate or
even suitable:
 Weak system conditions.
 Phenomena to be studied at other frequencies than fundamental frequency, e.g. SSR or
harmonics.
 Systems with multiple power electronic inverters or devices in close proximity.
 Applications where detailed control algorithms or the use of the actual controller code, compiled and
linked into the EMT program are required.

93
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

4.1.1 EMT algorithms and variations

The EMT algorithm which uses trapezoidal integration and nodal analysis was originally developed by
Dr. Herman Dommel in his 1969 paper [87] – this was the original EMT tools/programs. An independent
development effort of the EMT tool/program that covers DC network was undertaken by Dennis
Woodford [89], following the basic Dommel algorithm [87]-[90]. Other common programs using the EMT
algorithm or derivatives have also been developed. Commercial programs for translation from an RMS
program to EMT programs are available, as well as parallel processing add-ons and hybrid simulation
(running EMT and transient stability programs interfaced together in the same simulation). Other
common real time digital simulators have also been developed. Some ad-hoc tools can be also used for
hardware in loop real-time simulations.
The key concept of the aforementioned original EMT tools/programs is the use of a linear finite difference
equation to solve a differential equation – each passive electrical element of the power system (i.e.
inductors, capacitors, etc.) is represented individually as a resistor and parallel current injection (which
depends only on quantities from the previous time step (i.e. known as a history term).
The formulation for a simple inductor is shown in Figure 4.1, and for a capacitor in Figure 4.2. The
example of the aforementioned key concept is also shown in Figure 4.3.

Figure 4.1 EMT formulation of inductor differential equation as a difference equation via Trapezoidal
integration.

Figure 4.2. EMT formulation of a capacitor differential equation as a difference equation via
Trapezoidal integration.
The above EMT formulation allows each individual circuit element in a complex system of non-linear
differential equations to be represented by a resistor and current injection. The problem is reduced to a
linear matrix solution for each time step (i.e. solving N linear equations with N unknowns). Other circuit
elements (like transformers, transmission lines, machines, etc.) can also be derived in similar fashion.
Standard sparse matrix solution techniques can be used to solve such equations with minimal memory
and maximum speed [93].
The use of trapezoidal integration is deliberate and key in this formulation. An integration is the area
under the waveform with respect to time, and trapezoidal integration essentially approximates the
waveform (over 1 time step) as a trapezoid – the area therefore is the time step dt times the average of
the quantities at time T and T-dt, which is in the case of voltage, waveform [v(t) + v(t-dt)]/2. Higher order
approximations are possible, but can lead to numerical instabilities whereas trapezoidal integration is
guaranteed to be stable for linear circuits.
Note that the application of trapezoidal integration is a first order linear approximation applied between
successive simulation time steps. The smaller the time step, the more accurate the approximation will
be. This places an inherent error in any EMT simulation, as well as an effective limitation on the
frequency bandwidth over which the solution is valid. A good "rule of thumb" is that there should be at
least 20 time steps per cycle at the highest frequency of interest – for typical transient over-voltage

94
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

simulations with a frequency range of interest from DC to approximately 2 kHz this requires a time step
of 25 µs. The same simulation method can be used for transients over a wider range of frequencies (e.g.
for lightning transients) if a small enough time step is used.

Thevenin to Norton Transformation

Figure 4.3. Linear equivalent circuit - conversion of differential equations

to finite difference equations.
Some simulation programs use the concept of subsystems, which identifies sub-networks whose
network is coupled to other sub-networks only by traveling wave transmission line models. Bergeron or
frequency dependent models have no impedance element connecting the sending end to the receiving
end of the line or cable and the responses on either end to a change on the other end is delayed by the
traveling time of the electromagnetic (EM) wave. For an overhead transmission lines (in open air) the
EM wave travels with almost the speed of light. Therefore, a 30-km long line exhibits a natural delay of
approximately 50 µs. This technique of using Bergeron line models to couple sub-networks allows those
sub-networks to be solved on parallel processors without introducing artificial delays during the
exchange of state variables, provided the natural delay in the coupling lines are at least one simulation
time step long. As such it is often used in digital real-time (RT) simulators (DRTSs) which require parallel
processing for simulation speed and large networks are broken up onto numerous CPUs with
communication delays mapped to traveling wave line model delays. This technique can also be used to
interconnect sub-networks solved at smaller time steps e.g. a network of power electronic switches
represented as a transition between a large resistance in the off state and a small resistance in the on
state.
Many modern non-RT EMT programs also make use of interpolation, which allows switching instants to
occur at precise points between the regular time step grids. This has accuracy benefits such as precise
turn-on and turn-off instants for power electronic devices as well as stability benefits such as the
prevention of on/off/on/off and interactions due to trying to switch two or more devices in the same time
step. It can also be demonstrated that trapezoidal integration can result in time-step to time-step high
frequency numerical problems when a circuit with only inductors to ground or with capacitors in a loop
are represented. This is also solved with interpolation (i.e. half steps) or with the critical damping
adjustment procedure (CDA) [94].
Some non-RT programs are also able to change the simulation time step, either dynamically during the
run (perhaps to use a smaller time step around a switching instant) or allowing a reduction in the time
step near a transient. A reduction in the time step brings the time for a switching closer to the exact
point. However numerous devices may still need to switch in one time step which can lead to on/off/on/off
instabilities, or more importantly, the switching of one device may initiate the need to switch another
device, immediately. This illustrates the fundamental problem associated with modelling power
electronic switches in EMT programs: while in reality, parasitic capacitances, inductances, and a finite

95
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

switching time (during which the semiconductor layers lose conductivity) all play a role in the actual
transition between the “on” and “off” state of, say, an IGBT, these transitions are typically not properly
modelled in EMT type simulations. The transitions are approximated by “ideal switching” which in turn
can lead to numerical problems. For example, if the current through an inductor is forced from any value
i(t) to zero in one time step dt, it will result in a voltage across that inductor proportional to L*i/dt. The
smaller the time step the larger the artificial voltage spike. Interpolation and variable time step
techniques can minimize these numerical problems and so can proper applications of artificial damping
and snubber circuits (often used with fixed time step solvers) within the representation of the power
electronic switches in the model.
EMT algorithms are inherently solved in the time domain (like RMS transient stability programs) – i.e.
the solution produces time series of voltages, currents, control signals which can also be graphed as
such. Some programs also include a native frequency domain solution – this is useful for filter design,
or for computation/plotting of the frequency response of the system. Techniques for deriving the
equivalent impedance vs. frequency of a non-linear device can also be applied (known as perturbation
analysis) and can supplement the linear frequency domain solution of a network.
It is also worth noting there are variations in the main circuit nodal admittance (NA) matrix solutions –
the classic EMT algorithm uses a sparse matrix solver where G is the N x N conductance matrix, V is
the vector of node voltages to be solved, and I is the vector of known current sources. Ideal voltage
sources and zero impedance branches can be added to this formulation – this avoids ill-conditioned
matrix problems from the use of extremely small switch resistances (i.e. large conductance). A modified
version of these techniques is called Modified Nodal Admittance (MNA), which directly solves both nodal
voltages and current flows in the ideal branches [95].
Most modern programs are driven by a graphical user interface (GUI) – this allows the user to construct
both the electrical circuit and the controllers using building blocks that can be connected together. These
programs usually also allow interfacing to custom user-written code, often supplied by the manufacturer
of the device. In order to hide the confidential source code, the end user will often receive only binary
code (i.e. .lib, .obj or .dll files).
4.1.2 Real-time simulation
Digital real-time (RT) digital simulators (DRTSs) may require special processing in order to maintain
real-time speed. For example, interpolation of the electrical network may not be possible due to a non-
determinable number of switching events that may be required over a given small time period.
One solution used in DRTSs is to take advantage of the Dommel algorithm conductance for a switching
device, so it does not change when the state of the switching device changes. When a switch is in the
closed state, it is represented in the Dommel algorithm as an inductor (L) branch. Conversely, when it
is in the open state, it is represented as a series resistor-capacitor (RC) branch. The R in the RC branch
can be selected as a user-specifiable damping factor, so that if the closed branch was connected in
series with a branch with the open state, there would be a series RLC branch. This method avoids
switching of the conductance matrix altogether, with natural savings in processing time [96].
When a device model switches from an open state to a closed state, the energy in the capacitor C is
lost and should eventually be replaced. Assume that i is the rated switched current and v is the rated
switched voltage. The expenditure of energy in this case is equal to Cv2. When a device model switches
from a closed state to an open state, energy in the inductor L is lost, which is equal to Li2/2. In order to
minimize the losses, the suggestion on constraint is applied for selecting R, L and C: Cv2 should be
equal to Li2/2. Examination of L and C indicates that the ratio of the open state impedance to the closed
state impedance at any frequency (f) is above 1/(2 π f ΔT f)2 [97]. Consequently, the smaller time-step
leads to greater ratio of the impedances. Hence, it is desirable to keep the time step as small as possible
or ensure the switching frequency of the converter is not too high. For time steps around 2.5 µs the
converter switching frequency should not exceed 3 kHz. The smaller the simulation time step the larger
the allowable switching frequency.
Modern RT simulators also can directly implement “Discrete Switch Models” [98], [99], due to
improvements in sparsity/subsystem/partitioning matrix algorithms, as well as faster processors. If the
limit on the number of switch/conductance states is avoided by on-line computation of changes in
switches, this may alleviate the restrictions on topology and the number of switching elements in a given
case.
4.1.3 Advanced EMT modelling
Recent advances in EMT modelling include [91], [100], [101]:

96
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 Large system modelling and EMT dynamic performance studies.

 Parallel processing of EMT cases.
 Hybrid simulation with EMT and RMS programs.
 Databases to translate/initialize EMT models from RMS load flow/models.
Large systems can be modelled in today’s EMT programs. Such models have numerous SVCs, HVDC
links, wind plants, PV inverters, complex load models, large AC system models, comprehensive setup
for faults/contingencies, etc. This is really only warranted in complex networks with numerous complex
devices, or in weak system conditions, for most studies large system analysis is limited to RMS/transient
stability analysis. The lure of three phase analysis, proper representation of unbalanced/single phase
faults, use of exact controls/protection models, customizable models of power electronic devices, etc. is
attractive however the line between RMS tools and EMT tools is certainly blurring.
When large system studies are required, the CPU time for the analysis is a significant factor EMT tools
use small time steps (often 50 s, or less for some device models), three phases (instead of positive
sequence circuits) and include non-linear devices – all of which make models slower compared to RMS
transient stability analysis.
Parallel processing is now possible in some programs – systems can be broken up into subsystems,
using similar methods as developed for RT simulators. The power system can be separated at any
transmission or cable line location (specifically lines/cables whose travel time is larger than the
simulation time step) or at any transformer (by using a transformer scaling component). Each subsystem
can be operated at a different time step using interpolation of traveling waves in the line models – this
is a significant factor, as otherwise a single demanding device say a VSC inverter which requires a 5 s
time step to operate would dominate the time step for the entire simulation.
Hybrid simulation techniques have also been developed and can be used to run EMT and RMS transient
stability programs in the same combined simulation. The RMS program results (i.e. magnitude and
phase angle of fundamental voltage/currents at the interface) are used to update network equivalents
in the EMT programs. EMT results are processed via FFT/DFT (Fast Fourier Transform/Discrete Fourier
Transform) algorithms to extract the positive sequence behaviour, which is communicated to the RMS
program where it updates a dynamic load/generator model. Operation of the two programs can be made
transparent so a single start command initiates both programs.
Hybrid simulation can be viewed as the imbedding of complex/exact EMT models into RMS transient
stability programs. Alternatively, the hybrid interface can be seen to provide a dynamic network
equivalent to EMT programs. These techniques allow fast development and implementation of accurate
EMT models (which would otherwise be complex/time consuming to develop custom models in RMS
transient stability programs – say a detailed PV inverter model) into traditional and widely used RMS
algorithms.
The development of hybrid interfaces and parallel processing will allow for more detailed modelling of
power electronic inverters on a large scale – this will be important for the future as the power systems
become more complex.
4.1.4 Confidentiality aspects of EMT models
EMT models are often developed by the manufacturer using real code (i.e. the source code which are
used for the actual hardware) and are usually considered to be highly confidential. Users who require
access to these models are often required to sign strict non-disclosure agreements (NDAs) – this is
often troublesome in many interconnection agreements, where the developer must provide models to
the system operator, consultants, or other utilities.
One possible method to avoid some NDA issues if any part is deemed confidential is by “black-boxing”
– this takes either a complete control circuit model or electrical circuit model or in some cases both (i.e.
a complete model) and encloses the model into a single object or component thus removing the ability
to dig deeper into the model and see its sub-contents.
Some programs can automate the process of black-boxing or provide user-access passwords before
sub-pages can be opened. In other cases, the process to blackbox models can be time consuming and
expenses, particularly for the electrical system.
However, there are some clear disadvantages:

97
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 4.1 Type of phenomena and studies typically performed using EMT-type models
Sub- Type of Phenomena Relevant Key Words Type of Studies
Chapter
3.4 Behaviour in response to ●Device protection against ●Short-term voltage stability
large voltage deviations damage ●Transient stability
●FRT capability ●Provision of short-circuit current
●Grid support ●Low/High voltage ride through
3.7 Other phenomena and Low- and high-frequency Controller interactions, switching
studies interaction of controller transients, harmonic studies

 the lack of detailed knowledge of implemented functionalities,

 the risk of hidden failures,
 the impossibility to alter the model and decrease of computational performance as by external
interfacing no computation speed enhancements might be used.
4.2 POWER SYSTEM PHENOMENA REQUIRING EMT SIMULATION
Generally, in the case of small time constants of the investigated phenomena and weak systems, EMT
models and EMT simulation might be necessary. In Table 4.1 typical simulation tasks using EMT and
Real Time (RT) digital simulation programs are listed.

4.3 SELECTION OF ELECTRICAL INTERFACING MODELS AND CONTROL MODELS

Users of EMT programs for the study of power electronics, PV, wind, HVDC and FACTS, etc. should be
aware of the level of detail in their models, to ensure they meet the requirements for a given study.
There is an inherent difference in how EMT models are constructed and solved, versus how RMS
models or phasor models are constructed. RMS models are often constructed to interface to iterative
state variable solvers – this requires identification and storage of state variables, and iteratively solving
the controls to converge with the electrical solution for a given time step. This allows larger time steps
to be used in the RMS transient stability tools, or to avoid the need for very small time steps which would
slow down the RMS simulation. This requires a great deal of manipulation and approximation in the
controller code, and development of a custom RMS model is generally a time-consuming process.
The level of detail required - in both the control models and the electrical circuit - depends on the study
being performed. In general, the end users should:
 Understand the phenomena being studied.
 Understand where approximations are made in a model (i.e. what is included in the used model
and what is not included).
 Determine if approximations or omissions may affect their study.
For example, if a model does not include all protection/control systems, then it may be suitable for a
preliminary feasibility study, but would not be suitable for the final design/performance studies.
4.3.1 Inverter models
Methods of modelling a power electronic inverter or converter within the electrical EMT algorithm are
listed below according to a decreasing level of detail and increasing speed of simulation. As the level of
detail in the model increases, additional controls may be required (i.e. firing pulses), and the
computational burden increases (slowing down the simulation).
4.3.1.1 Discrete switch models
Discrete switch Models essentially require each individual switch device (IGBT, diode, thyristor, etc.) to
be modelled separately and solved by the main circuit solution algorithm. It also requires an almost full
representation of the inverter’s control system.
4.3.1.2 Averaged switch models
In the averaged switch models the source impedance is also changed, and the resulting source is
essentially a network equivalent. Averaging is done in the time domain (switching time) and its

98
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

components. One application for an Averaged Switch Model is for a modular multilevel converter (MMC)
or similar type of converter.
4.3.1.3 Simple source models
In this simplified model, the inverter output current reference signal (from an outer PI controller for
example) is converted to a three phase ABC reference system using the PLL angle which in turn is
injected into the electrical grid via current sources.
4.3.2 Controller models
EMT models are solved with fixed or sometimes variable time step techniques. The controller receives
the voltages and currents from the last time step, then the controls are called (possibly in the exact order
in which the real controls are called, and possibly with the same controller sampling time as the real
controls) to generate output signals such as firing pulses. Then the output signals are connected back
into the electrical simulation to turn on/off IGBTs, etc. The EMT method of modelling often uses much
smaller time steps than the real controller sampling time which means that less manipulation and fewer
approximations are required. However, the oversampling of the control system can sometimes lead to
inaccurate results if controller interrupts are not modelled appropriately. Programming of detailed power
electronic models is often much easier accomplished in the EMT world. The level of detail in the controls
is also related to the level of detail in the electrical portion of the models – for example a Simple Source
Interfaced converter model may inject only high level currents into the system (similar to RMS models)
and hence will not require inner PI loops, nor IGBT firing pulse logic, capacitor balancing logic, etc.
Controls for power electronic inverters can be implemented in many ways in EMT programs as described
in the following clauses:
4.3.2.1 Generic control models
Generic (EMT) models are probably quite similar to the models used in RMS programs. Standard
controller building blocks are used (say for an integrator, or a first order real pole, lead lag, etc.) without
attention to how each model is implemented (integration method used, Z transform methods, etc.) or
the time step used in the controls. The controls are usually sampled at the same time step as is used in
the electrical solution. Often these models are included in the programs as a library model (say for a
PLL function, or for VSC firing logic), or may be available as a collection of standard building blocks.
4.3.2.2 Detailed models with standard library building block functions
Detailed models can also be constructed using standard library building block functions. However, they
are constructed to a much greater level of detail. Since they are not an exact representation of the real
code, many parts of the code are omitted, and the details of how a given function is implemented are
ignored due to the inherent use of standard functions. These control models are often solved with the
same time step as the main electrical circuit.
4.3.2.3 Custom detailed models
Custom detailed models can be a near-exact representation of the real code. Most real controllers are
developed with a GUI (graphical user interface) – sometimes proprietary, but which can also be part of
a simulation package. The tool which is used to develop the real controls, inherently has controller
building blocks, which can be duplicated in a given simulation program on an individual building block
basis. Once an equivalent library of building blocks is complete, the overall controller functions can be
duplicated, either by manually manipulating control blocks or possibly automated, if such features are
provided. The overall level of detail in “Custom Detailed Models” varies, depending on the manufacturer
and developer of the model – factors to consider are:
 How much of the controller is represented and which portions are omitted?
 Are protection functions included?
 Is the model maintained and kept up-to-date with site changes or project changes? Are all gains
and settings the same as the “as-built” in service converter?
 Is the order of execution of the individual building blocks guaranteed to be the same as in the real
code?
 Is the time step used in the controls the same as the real hardware (which is likely different than
the time step used in the EMT electrical network solver)?
 Have the model been validated and compared to the site traces/real code?

99
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

4.3.2.4 Exact models

Exact models are constructed by taking the actual source code from the real controller, and
compiling/linking/interfacing it directly into the EMT simulation. Since the real code is used, the concerns
regarding the level of detail and approximations used above are avoided. Some of the special aspects
which must be considered when developing or using “Exact Models” are:
 Is the sample time of the real code reproduced? A sample/hold mechanism should be implemented
to ensure the real code is called only at the desired sampling instants.
 Is the model capable of being called more than once in the same simulation (i.e. is it re-entrant)?
This requires the identification and storage of all controller state variables, such that the state
variables from one call to the model are kept separate from other subsequent calls in the same
time step.
 Is the model capable of being initialized? This is particularly important if the model has long time
constants to avoid long run-up times and compatible with a program’s feature to stop and re-start
a simulation at any point in time (i.e. snapshots or multiple runs)?
 Possible code language and linking issues – when models from many suppliers are required to be
executed in the same simulation, this can create linking problems due to incompatibilities between
compiler languages or versions, or with overlapping subroutine names in models from each supplier.
This is largely resolved if the binary models are released in source code format (if allowed under
non-disclosure agreement (NDA) requirements – see above) or if the binaries are released in
standard .DLL format (which alleviates most compile/linking issues).
 Have all aspects of the controller been included? For example, protection functions imbedded in
the controller code are likely included in the “Exact Model” but are all such processor code and
external protection functions included?
Validation of “Exact Models” is inherently provided for or may not be required as the complete model
generally works “out of the box” without much tweaking or customization required.
An essential disadvantage of detailed (or exact) control models is confidentiality and NDA issues. Since
the models may include actual code, either in source code format or in binary form, the model capability
could potentially be studied and reverse-engineered/probed to determine its capabilities. Licensing of
such models is possible ensuring only validated customers can run the model to avoid such difficulties,
but often NDAs are required which prevent general distribution of the models.
The biggest advantage of “Exact Models” is the inherent accuracy and confidence level in the model. It
is the “real code” from the actual hardware, simply compiled, validated and linked into a simulation
program on a PC. This technique also provides a black-boxed model, somewhat resolving NDA
concerns, but also means the end user cannot see the block-by-block control functions and
implementation.
There is an effort underway to create a standard suitable for interfacing to both EMT and RMS models
– this effort is referenced as IEC-61400-27-1 [77] and is primarily aimed at models of wind turbine
controllers. Programs have been developed to directly import the DLL-files and to create the interface
to an EMT program (with support for state variable storage, control of the sampling time, re-entrance)
including the creation of the graphical appearance and inputs/outputs.
4.3.3 Selection guidelines
The level of detail required, in both the control models and the electrical circuit, depends on the study
being performed. The following table illustrates some of these requirements.

100
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 4.2 Control model for type of study

Study Control Model Required Electrical Model Required
Feasibility Study All Control Models are suitable All Models are suitable
Transient Over voltages Standard or Custom Detailed, or Exact Interface (satisfactory),
Averaged Switch (better) or
Discrete Switch (the best)
Harmonics All Control Models are suitable Averaged Switch
provided they represent the converter (satisfactory) or Discrete
impedance at the harmonic Switch (the best)
frequencies of interest
Interconnection Study Standard or Custom Detailed Averaged Switch
(satisfactory), or Exact (the best) (satisfactory) or Discrete
Switch (the best)
Weak System Studies Standard or Custom Detailed Averaged Switch
(satisfactory), or Exact (the best) (satisfactory) or Discrete
Switch (the best)
SSTI/SSCI Standard or Custom Detailed (maybe), Averaged Switch
or Exact (the best) (satisfactory) or Discrete
Switch (the best)
Control Design and Dynamic Standard or Custom Detailed (maybe), Averaged Switch
Performance/Rating Studies by or Exact (the best) (satisfactory) or Discrete
the Manufacturer Switch (the best)
Interaction Study Standard or Custom Detailed (maybe), Discrete Switch
or Exact (the best)

4.4 DETAILED DESCRIPTION OF INVERTER MODELS FOR EMT STUDIES

4.4.1 Discrete switch models
“Discrete Switch Models” essentially require each individual switch device (IGBT, diode, thyristor, etc.)
to be modelled separately and solved by the main circuit solution algorithm. The structure of the discrete
switch model is shown in Figure 4.4.

Half BridgeHalfMMC Converter

Bride MMC Converter

Gate pulses
D

Vd, Vq

dq Vref, abc D

PWM
abc

q D

PLL
...

Vabc
D

Figure 4.4. Structure of discrete switch model.

101
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Advantages:
 Most accurate method possible which is suitable for interaction studies and internal power
electronics design/insulation coordination studies (sometimes special non-linear switching models
can be used by suppliers).
 Numerically stable.
 Full controller interface is required and used, individual IGBT firing pulses are required from the
controls.
 Compatible with interpolation algorithm.
Disadvantages:
 Relatively small time steps must be used (often a 10 µs or smaller time is required) with
corresponding increase in execution time.
 May not be possible to use them with RT digital simulators in large subsystems due to high number
of switching elements.
4.4.2 Averaged switch models
Averaged switch models are a special type of interfaced model – the difference from a “Simple Source
Interface Model” is that the source impedance is also changed, and that the resulting source is
essentially a network equivalent. The main application for Averaged Switch Models is for multi-level
converters (MMC) – these are difficult to model as “Discrete Switched Models” due to the extremely high
component count. MMC converters have numerous series levels which is the key to why they can
produce such clean AC waveforms, as each level would require IGBT and diodes to be implemented
directly.
Averaging is done in time over the switching cycle and for components. Figure 4.5 shows the stages for
collapsing/creation of a network equivalent in each time step, while the accompanying text below
describes each step.

Half Bridge MMC Converter Averaged Model Source Interface

Req

Eeq

Figure 4.5. Stages of forward collapsing and back substitution for derivation of an averaged switch
model.

102
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The basic concept of an “Averaged Switch Model” can be illustrated by the following steps (with
reference to the figure above):
Forward Collapsing Process:
 For a given time step, the controls will determine which IGBTs are on and off – local integrations
for the snubber circuit and internal capacitance are also computed (using trapezoidal integration)
to determine the equivalent circuit for each stage as shown in the 2nd stage circuit.
 The R and I equivalents can then be collapsed (through Norton to Thevenin transformations) to
arrive at a single R and E source for each stage (as shown in the 3rd stage circuit).
 The N stages can now be summed, to determine a single equivalent Req and Eeq, representing
the entire MMC unit for this phase and arm. These interface to the main program and EMT circuit
simulation, which in the next time step will compute the new electrical solution.
Backward Process:
 The device was represented as a single “Eeq behind Req” Thevenin source (or equivalently as a
current injection in parallel with a source resistance Norton current source) for any given time step.
Both Eeq (or I) and Req are computed dynamically and will change at each time step.
 From the previous time step solution, the voltages at each end and the current in the equivalent
interfaced source (stage 4 circuit) can be measured (it is computed along with the voltages by the
main EMT electrical solution). The current is the same through all series branches.
 From the left of the diagram the 3rd stage of the circuit reduction and from the previous time step,
the individual values of E and R are remembered – given the new voltages and current from the
current time step, and using the old source values from the previous step, the voltage across each
stage can be computed (i.e. we now know both V and I across each stage from the previous time
step). This is a back-winding process.
 The equivalent circuit for each stage is shown in the 2nd stage – if the voltage across the branch
and the current through the branch from the last time step are known, and the previous current
injections and resistance values are remembered (i.e. the states of switches) for individual diodes,
IGBTs and snubber RC circuits for this stage; then the voltage and current across each capacitor
can be computed (snubber and internal capacitance) from the last time step.
 This completes the Backwinding Process, and the Forward Collapsing Process above starts.
Thus, at each time step, the voltages and currents of each stage are computed, the differential equation
of the capacitor is solved, and the IGBT and diode switch states are determined. A process of forward
collapsing and back-winding is used to present the entire N level power electronics arrangement as a
single resistor and source (which may change each time step). The computation of the equivalent E eq
and Req is a mathematically solid process based on a reduction/equivalencing of numerous series switch
elements. The example comparison of currents from averaged switch model and exact discrete switch
model for a transient event is illustrated in Figure 4.6.
Advantages:
 Extremely efficient as little power electronic switching is required as compared to a discrete switch
model which may have hundreds of switching devices.
 Numerically stable it is not an interface method, but rather a network collapsing method.
 Full controller interface is required and used – individual IGBT firing pulses are required from
the controls.
 Large time steps can be used much larger than for a “Discrete Switched Model” with
corresponding savings in execution time.
Disadvantages:
 The switching of IGBTs and diodes is not interpolated the circuit is called on the regular time
step grid, and the main program cannot interpolate the switching of the IGBTs or diodes. This
normally can result in spikes and on/off/on/off switching, but the MMC topology seems favourable.
 Harmonics are produced by the converter model, but due to the limitation of switching on the
regular time step grid, the harmonic output will not be as precise as compared to an interpolated
“Discrete Switch Model”.
 May not be suitable for detailed models, interaction studies or internal power electronics design
studies by the manufacturer
Real-time (RT) digital simulators also make use of averaged models (particularly for MMC VSC models)
due to the tremendous savings in processing required by the main circuit solution, and the ability to use
large time steps. A hybrid model could be further developed, which is based on the average voltage

103
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

source model with the two diodes. However, the individual submodule voltages and firing pulses of the
IGBTs are calculated for the represented voltage source [102].

0 2 4 6 8 10ms
Figure 4.6. Comparison of currents (a,b,c) from averaged switch model and exact discrete switch
model for a transient event [103].

4.4.3 Simple source and interface models

The simplest method to interface a power electronic inverter to the electrical EMT circuit is to take the
output current reference signals (from an outer PI controller of a cascaded PI control structure for
example), convert them to the three phase ABC reference system using the PLL angle and then inject
these signals into the electrical grid as current injections. This is shown below in Figure 4.7.

Figure 4.7. Simple current source model interfaced to the EMT electrical circuit.
In addition to the circuit shown, another interface has to be used on the DC side of the converter – the
DC current which is injected into a separate DC equivalent circuit (with the VSC capacitance) is obtained
by dividing the measured AC power by the DC voltage.
Advantages:
 Extremely efficient as no IGBTs or power electronic switching is required.
 Does not generate harmonics.
Disadvantages:
 Can be quite unstable in a non-iterative solution (i.e. EMT algorithms) due to the time step delay in
the input voltages (from the last time step) and the use of a current injection in the current time step.
 Does not model the inner PI controls and firing controls.
 Not suitable for detailed models or interaction studies.
 Voltage source device modelled as a current source.
There are adaptations of source type models with some minor improvements. For example, the inner PI
loop can be added to the above, thus generating ABC voltage sources instead of current sources, which

104
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

in turn can be interfaced to the electrical circuit as an inductive voltage source (using the inductance in
the power electronics circuitry). This can be extended ever further, to include the IGBT firing circuit
(comparing the ABC voltage references to a PWM triangular waveform for example), this however would
introduce some harmonics to the circuit.
In general, however, source or interface models (with current or voltage sources) should be limited to
use with relatively strong systems (See Clause 4.6.1), and used only for generic feasibility studies.
Caution must also be used with these models, as they can be numerically unstable.
4.5 INVERTER CONTROLS AND INTERNAL PROTECTION
The controls required in an IBG can be classified within three functional levels:
 Low level controls: inner current PI control loop, creation of firing pulses (PWM), PLL,
 High level controls: Id/Iq current reference, see Sub-Chapter 5.5 (RMS models),
 Plant level controls: frequency and voltage control, or P&Q control at plant level, see Sub-Chapter
5.6 (RMS models),
Some controls are necessary in EMT programs (just like in the real controls), but are not needed in RMS
programs due to the method of interfacing to the main program. For example, in MMC multi-level VSC
converters, IGBT firing controls are required for Detailed or Exact Models and must therefore include
capacitor balancing functions (effects which can be ignored in Simple Source EMT models or in RMS
models).
This chapter contains a description of the common control and protection functions which are often
required in detailed EMT models. These controls are often required in other EMT models, in addition to
the standard main high-level control functions which are common with RMS models discussed in
Chapter 5.
4.5.1 Inner loop PI controller
Many VSC inverters use a cascaded PI Vector control philosophy (Figure 4.8).

vd ,vq Vref,a bc
iqref Ki /s +
+ dq
+ +
Kp + abc
iq

id Xt

idref
Ki /s +
+ +
Kp + -
id
0.1
iq q
dq q
abc
PLL
ia bc

va bc

Figure 4.8. Typical cascaded PI control structure for VSC inverters.

The key concept of this control is the use of a PLL (see Clause 4.5.2) and DQ transformation to extract
the real and imaginary components of measured voltages and currents. This allows a controller to
separately control quantities which depend mainly on reactive power (such as the AC voltage

105
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

magnitude) or real power (such as a converter DC voltage or power flow). The two levels of PI control
use a slower outer PI loop to generate current references and fast inner PI loops to provide fast
regulation.
RMS programs often ignore the inner PI controllers - instead injecting the Id and Iq reference signals into
the electrical circuit which basically assumes the inner PI control loops are perfect controllers and always
control the actual currents to the precise ordered values. This approximation is often made due to the
small time step in the inner PI controllers, and the relatively high proportional gain. Unfortunately, this
simplification can also lead to inaccurate results in some cases – consider the recently found (but very
extensive) SSCI phenomena (Sub-Synchronous Control Interactions) when Type 3 (DFIG – Doubly Fed
Induction Generator) wind plant is placed near series compensated transmission line [14]. All
(uncorrected) DFIG turbines will show extreme instabilities when connected radially into series
compensated lines, due to the inner loop PI controls (of the rotor side of the turbine VSC converter)
interacting with the electrical sub-synchronous resonance due to the series capacitor.
This phenomenon cannot be observed in RMS programs because:
 They cannot represent electrical phenomena other than fundamental i.e. an electrical differential
equation representation, as used in EMT programs, is needed.
 The inner PI controller is usually ignored as it would require a very small time step in RMS programs.
4.5.2 Phase locked loop
PLL (Phase Locked Loops) are critical in any EMT model which interfaces to the electrical grid via firing
pulses (i.e. “Standard Detailed” models or better, but not “Simple Source” models). A typical PLL is
shown in Figure 4.9.

Figure 4.9. Typical phase locked loop (PLL)

PLLs are a key component of EMT models which generate firing pulses or operate in the DQ domain.
The output signal Theta is essentially the angle of the positive sequence fundamental frequency
component of the voltage. Many RMS programs provide the phase angle measurement directly and do
not have to deal with harmonics, unbalanced voltages, distortion, etc., so this function is not included or
is modelled as a simple time delay.
4.5.3 Pulse width modulation
Most 2 or 3 level (non-MMC) VSC converters use a pulse width modulation (PWM) or Ripple Control
(also called Delta PWM) firing methods.
2 level converters essentially generate a square wave signal on the secondary of the connecting
transformer – the IGBT firing basically connects the DC side voltage with either a positive polarity which
increases the instantaneous current in this phase or with a negative polarity which decreases the

106
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

instantaneous current in this phase. The positive to negative transitions are performed rapidly (usually
at a switching frequency above 2 kHz) with a relatively clean low-order harmonic profile i.e. the
frequency components are at fundamental frequency plus high frequency components near the
switching frequency, which are filtered via high pass filters (See Figure 4.10).
3 Level converters use a slightly more complex power electronic device arrangement, but result in
cleaner output with lower harmonics/filtering requirements and lower losses. Third harmonic can also
be added to the voltage firing reference, which increases efficiency and reduces losses.
Advances in voltage source converters for HVDC systems use MMC (multi-level converters) – these are
useful when the DC side voltage is relatively high (i.e. overhead line or cable transmission projects). In
this case the produced AC quantities can be nearly sinusoidal, with minimal high frequency noise [104]
(See Figure 4.11).
4.5.4 Internal protection modelling
EMT models can include detailed models of protection systems, including how these systems respond
to unbalanced faults (such as single line to ground faults), harmonics, distortion, etc. It is relatively
common for an RMS model to show good performance in a system study, yet for a turbine to trip in real
time (say due to IGBT over currents or other protection systems).
This is a significant factor in many RMS studies and Ride-Through criteria. Thus, a simplified model may
show a device riding through a given disturbance, but once detailed protection systems (and realistic/on-
site settings) are added, the machine may trip to protect itself from the damage.

D D D

D D D

Figure 4.10. Example of PWM firing for 2 level VSC converters.

107
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 4.11. Example of multi-level converter output (400 MW with 200 power modules per
converter arm).
It is recommended to include the detailed internal protection systems in final studies where controller
gains/settings are finalized, including RT simulator tests. Earlier/generic studies do not necessarily need
this – the hope is that protection operations can be relatively easily fixed by algorithm adjustment,
settings, delays, etc.
Some examples of inverter internal protections used in EMT models, in addition to the higher-level
protection listed in Clause 2.3.3 are:
 Reduction of maximum inverter current when the DC voltage exceeds a certain limit;
 Limitation of inverter current’s variation rate after a fault;
 Limitation of total reactive current;
 Differential AC overcurrent protection;
 DC overcurrent protection;
 Over/under voltage protection;
 Over/under frequency protection.
4.5.5 Measurements techniques
Other aspects to consider include voltage measurement time constants (RMS programs can generate
a pure step in a measured RMS voltage, whereas EMT programs inherently must include a delay effect
due to a true RMS or FFT/DFT measurement delay). This additional delay can generate instabilities in
the overall controller response, but it is masked in simple models or RMS models.
The input signals to converter controls often include:
 Instantaneous voltages and current.
 RMS quantities.
 Frequency measurements.
 Status of nearby breakers, disconnects, etc.
 Input from operators (reference signals).
There are numerous methods of measurements possible – for example: a simple RMS signal can be
derived with an analogue rectifier/smoothing method, true RMS, DFT/FFT methods, etc. The accuracy
of the method may have influence on the real control performance. Similarly, the type of transducer used
in the real system (CT, CVT, PTs, etc.) can also have an impact (due to saturation or non-linearity,
frequency attenuation, etc.).
Common measurements in EMT programs may show effects not seen in RMS programs – for example:
DC offsets, or fundamental frequency oscillations during periods of DC offsets, second harmonic

108
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

oscillations during periods of large negative sequence unbalances, single line to ground faults, etc. The
behaviour of EMT models during unbalanced faults can also be interesting (and realistic) – often a DC
offset or second harmonic can appear in a measurement and be amplified by a controller response due
to the measurement algorithm for the RMS signal. This is a common problem dealt with in real controls,
but often is not necessary in simpler models. In inverters used near synchronous machines, a second
harmonic oscillation of the rotor side quantities will also appear in detailed EMT models, due to negative
sequence content on the machine stator side.
For these reasons, it is important to understand the methods and types of measurements used in the
real controls when performing EMT modelling. If “Exact Models” are available from the manufacturer,
these issues will be largely resolved as the actual code is used for measurements.
4.6 SYSTEM MODELS FOR EMT STUDIES
The previous sub-chpaters discuss specific inverter modelling aspects for simulation studies. In addition
to the converter, filters, controls and the protection, a successful simulation also will require modelling
of the surrounding AC system.
4.6.1 System strength - short circuit MVA ratio
When connecting a power electronic converter to an AC system, the “system strength” is a key indicator
that is the size of the converter relative to the system 𝑆𝐶𝑀𝑉𝐴 (short circuit MVA), this is called the system
SCR (short circuit ratio). A commonly used adaptation of the SCR is the ESCR (Effective Short Circuit
Ratio) which subtracts the MVARs of nearby shunt capacitors or AC filters from the 𝑆𝐶𝑀𝑉𝐴 (as the
capacitors have a weakening effect on the fundamental frequency impedance, yet do not impact the
short circuit MVA).
These are defined as:
𝑆𝐶𝑀𝑉𝐴 𝑆𝐶𝑀𝑉𝐴 −𝑄𝐶
𝑆𝐶𝑅 = and 𝐸𝑆𝐶𝑅 =
𝑃𝑛 𝑃𝑛
Equation 4.1
where:
𝑆𝐶𝑀𝑉𝐴 is the short-circuit power of the AC grid at the PCC (MVA)
𝑃𝑛 is nominal power of the IBG (MW)
𝑄𝑐 includes the effect of AC side equipment associated with the IBG (filters, shunt capacitors,
synchronous condensers, etc.) (Mvar)

As a general rule of thumb, projects in weak systems (i.e. SCR less than 5) will face significantly more
challenges in the controls and protection of the device (see Sub-Chapter 4.1). The exact safe ESCR
level depends on the manufacturer and project – even higher SCR conditions can still pose the same
problems. More information is provided in TB671 [76].
For the projects which use “off the shelf” control designs (i.e. designs which do not undergo significant
tuning, special control design or custom modelling efforts) caution should be used, plan on significant
study efforts, delays, working with the manufacturer to design/tune controls, etc. in weak power systems.
4.6.2 Size of system model required and system equivalents
EMT simulations are often used to study power electronic devices and the interactions with the nearby
electrical system. Some factors to consider when determining how large the system model is required
are:
 Try to reproduce the frequency response of the electrical system in addition to the fundamental
frequency response. The use of system equivalents relatively far away from the study area (or
FDNE – frequency dependent network equivalents) should be considered for some studies.
 Are there other nearby complex devices which should also be modelled (i.e. other nearby FACTS
devices, power electronic devices, large industrial loads, series capacitors, etc.).
 The level of detailed required in your device. If detailed models are available from the manufacturer,
these should be preferred.
A common method to determine how much of the AC system should be modelled, is to first develop a
very large system model (this can be automated using translation tools) and to compare it to
successively smaller system models (which run faster). The goal is to achieve a reasonably accurate
frequency response (and sometimes the inclusion of nearby machines to get accuracy of the electro-

109
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

magnetic response) with the smallest model possible to keep the simulation speed reasonable. The
counterpart process is to develop a small system model that included all the models under study and to
compare it with its successively larger system until no significant resolution is achieved.
4.6.3 Further modelling aspects for EMT studies
Other factors in AC system modelling can be important, including [91]:
 Frequency dependent line and cable models to get the wanted accuracy for high frequency
responses and higher damping at higher frequencies.
 Transformers and saturation for transient overvoltage, grounding, and inrush phenomena.
 Load modelling to accurately reproduce induction motor loads (which dominate the ratings of
reactive power support needed in response to faults) and to model possible interactions between
nearby power electronic devices with complex/significant or industrial loads.
 Generator modelling and electro-mechanical models for other rotating machines – some system
studies (such are: SSR, SSCI, SSTI, dynamic performance studies, etc.) require modelling of
nearby generators.
 Power electronic converters generate harmonics and usually require filters. Load flow and RMS
tools are not concerned with harmonics or power quality, and only represent the fundamental
frequency MVARs of filters (i.e. an RLC filter is often modelled as a simple fixed shunt capacitor
bank). A comprehensive set of guidelines for harmonics is outside of the scope of this JWG,
however the end user should be aware of potential harmonic conditions and their effort on the
power system, and should expect models to include resistor, inductor and capacitor filter models
with appropriate tuned resonance conditions and damping.

4.7 CONCLUSIONS
In Chapter 4, guidelines for the application of EMT models for IBG are given. Advantages of EMT
simulation and typical cases where EMT modelling is inevitable are listed. Different levels of detail in the
models is discussed in the descriptive part on the electrical interface and control models. However, only
components exclusively used in EMT models are covered. These are especially the inverter models at
the switching level and low-level control functions like the inner loop current controller. The high-level
controls are included in Chapter 5 (RMS model) since they are the identical for RMS and EMT simulation.
The level of detail required in both the control models and the electrical circuit, depends on the study
being performed. In general, the end users should:
 Understand the phenomena being studied.
 Understand where approximations are made in a model (i.e. what is included in the used model and
what is not included).
 Determine if approximations or omissions may affect their study.
 Discuss the modelling of the external system.

110
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

5. RMS MODELS FOR INVERTER BASED GENERATION

5.1 INTRODUCTION
Power system dynamic simulation tools using RMS-type models were mainly developed to study
electromechanical oscillations of power systems consisting of large generators and motors. Phasor
simulation methods are used when the fundamental frequency behaviour is of interest, i.e. it is not
necessary to solve all the differential equations of the network resulting from its R, L, and C elements.
Instead, a much simpler set of differential-algebraic equations involving the voltage and current phasors
is solved. Time-varying complex numbers are used to represent sinusoidal voltages and currents at
system frequency, expressing their value either in Cartesian (real and imaginary) or polar coordinates
(amplitude and phase).
Although there is diverse range of opinions, but RMS-type simulation tools are more likely to consider
phenomena mainly in a bandwidth of at most 10 Hz (i.e., periods higher than 0.1 s are of interest)27. The
numerical integration time-step is therefore chosen in the range of 1–10 milliseconds, while software
employing variable step-size techniques can use larger time-steps during the periods of low dynamic
activity. For frequencies outside the operational range of a nominal network frequency (e.g. several Hz
smaller/larger than the rated frequency) or the analysis of harmonics is required, the use of Electro-
Magnetic Transient (EMT) simulation software is recommended, as they contain more detailed and
higher-order equipment models [105]. Even so, RMS-type simulations are much faster to execute than
EMT and are thus the dominant simulation method used in stability studies.
Simplified converter models are included in RMS programs (also known as averaged models). RMS-
type models provide the behaviour of the converter ignoring fast switching transients and any control
with very small time constants compared to the simulation time steps and the phenomena considered.
That is, in stability analysis there is usually no need to model the inverter in detail when its transients
are likely to decay within the time step of the time-domain simulation. Only the fundamental frequency
outputs of the converter are modelled, mainly reflected in the electrical control model. Due to this
averaged representation, the modelling of different technologies of inverter-based generators can be
unified using generic sub-models for the basic components.
The performance of these models should be valid, within a reasonable tolerance, for a wide range of
operating conditions of quantities such as voltage, active and reactive power at their terminal
connections. While it is desirable for the models to be valid for the entire operating range of the converter
(normal and abnormal), this is sometimes not possible due to the modelling limitations of RMS models.
RMS-type models are not recommended for use in power system dynamic studies in a weak grid. The
definition of a weak system is often defined using the Short-Circuit Ratio (SCR) and it is sometimes said
that RMS-type models might not be reliable in the case of the SCR of below 3. However, such criteria
can be different between the conventional technologies such as Line Commutated Converter (LCC) and
the recent technologies such as Voltage Source Converter (VSC). If conventional technologies are used,
the models should perform accurately for systems with a SCR of three and higher at the Point Of
Interconnection (POI). For the parts of the system with low SCR levels, detailed vendor specific models
may be needed. For very weak connections, a vendor-specific EMT-type model may be necessary even
for transient stability-type studies. In the case of LCC technologies, the following classification is
proposed when dealing with power electronic interfaced technologies connected to weak points in the
network:
 Connection is considered to be weak if SCR at the POI is less than 5 or if SCR at the MV
collection grid is less than 4 (whichever of these criteria is met).
 Connection is considered to be very weak if SCR at the POI is less than 3 or if the SCR at the
MV collection grid is less than 2 (whichever of these criteria is met).
The following approaches are recommended when undertaking power system studies for planning and
connection assessment:
 For weak connection applications:
 Comparison of RMS-type dynamic models against the detailed EMT-type models and field
measurements (at the time of commissioning) is advantageous.

27
This does not mean that the fault current cannot be modelled. The periods of few milliseconds may
be of interest.

111
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 Detailed RMS-type models are recommended as opposed to generic models.

 Changes to the plant control system and/or installation of supplementary equipment, e.g.
synchronous condensers, is less likely.
 For very weak connection applications:
 Verification of RMS-type models against the detailed EMT-type models is necessary before
carrying out any detailed connection assessment studies.
 In the event that an acceptable correlation does not exist between the RMS-type and EMT-
type models, either RMS-type models need to be revised, or EMT-type models need to be
used for the connection assessment studies.
 Changes to the plant control system and/or installation of supplementary equipment may
be necessary.
With more recent converter technologies, such as VSC, the requirements for connection to a weak
system may vary. The criteria are very much dependent on how the VSC is designed and the type of
control used.
Users need to be careful on the selection of the type of models between RMS and EMT. RMS models
usually do not include the inner current control loops of the converter, the detailed PLL model and
various other details of the converter. Thus, RMS models are not suited for any type of studies strongly
relying on the aforementioned elements (see Sub-Chapter 4.2), and in those cases, EMT models should
be preferred.
Early applications of RMS-type simulators can be found in publications from the 60s. However, most of
the methods used in today’s commercial software originate in the 80s, encouraged by the proliferation
of personal computers and the decrease in their price. Some of the most well-known, commercially
available, software used today are shown in [106]

5.2 POWER SYSTEM PHENOMENA REQUIRING RMS SIMULATION

RMS models are mostly used to investigate electromechanical phenomena and stability issues. Table
5.1 provides an overview of these along with the corresponding sub-chpaters in Chapter 3 where they
are better detailed. While most of these studies could also be performed with EMT models, the RMS
ones can provide an accurate response with much faster execution times.

Table 5.1 Type of phenomena and studies typically performed using RMS-type models
Sub-
Type of Phenomena Relevant Key Words Type of Studies
Chapter
3.3 Behaviour in response to frequency device protection frequency regulation and
deviations transient stability
system support
plant level control
synthetic inertia
3.4 Behaviour in response to large voltage Device protection against damage short-term voltage stability
excursions FRT capability
transient stability
Grid support
provision of short-circuit
current
Low/High voltage ride
through
3.5 Behaviour in response to smaller but V/Q control long-term voltage stability
longer voltage deviations
Reactive power capability below
maximum capacity diagram
plant level control
3.6 Modelling simplifications for small- small-disturbance angle
disturbance stability analysis stability

112
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

5.3 MODELLING OVERVIEW

Table 5.2 summarizes the different components available in the inverters and whether they are
represented explicitly in the RMS models depending on the type of study. These components are
revisited in the later sub-chapters along with example implementations.
It is important to select or develop IBG models which are suitable for bulk system impact studies and
are able to provide accurate and reliable results in time-domain simulations. Prior to the work reported
in this chapter, exhaustive work has been done in [3], [77] on modelling of wind turbine generators. If
the primary source (DC source control) of the IBG is neglected (i.e. DC source is assumed to be “stiff”),
the model of PhotoVoltaic (PV) units becomes very similar to that of the Type-4 wind turbine generator.
Thus, [77] may also be referenced for the modelling of individual elements in PV units.

Table 5.2 Overview of components and selection in RMS-type models

Small and
Large
Frequency long-term Unintentional
Category Functionalities voltage
Deviation voltage islanding
excursion
deviation
Control DC source control No Yes* No Yes
Control Current control No Yes* No Yes
Control Phase Locked Loop (PLL) No Yes No Yes
Control Maximum Power Point Tracking (MPPT) Yes** No Yes No
Reduction of maximum inverter current
Protection when the DC voltage overcome a certain No Yes* No Yes
limit
Limitation of inverter current’s variation
Protection No Yes No Yes
rate after a fault
Protection Current limit Yes Yes Yes Yes
Protection DC Overvoltage Protection No Yes* No Yes
Protection Overvoltage/Undervoltage Protection No Yes Yes Yes
Overfrequency/Underfrequency
Protection Yes Yes No Yes
Protection
Protection Protection for Detecting Balanced Fault No N/A**** No Yes
Protection for Detecting Unbalanced
Protection No N/A**** No Yes
Short-Circuit Fault
Protection for Detecting Single-line-to-
Protection No N/A**** No Yes
ground Fault
ROCOF tripping: monitoring the power
frequency variation rate and
Protection Yes Yes No Yes
disconnecting the inverter when it
reaches a certain limit [Hz/s]
Protection Vector jump Yes Yes No Yes
Protection Transfer trip No No No Yes
Protection Anti-islanding active detection method Yes Yes Yes Yes
Capability P(f) control (over/under frequency) Yes No No Yes*****
Voltage control by means of reactive
Capability Yes Yes Yes Yes
power
Voltage control by means of active power
Capability No Yes Yes No
[P(V)]
Capability Synthetic Inertia Yes Yes No Yes
Capability ROCOF immunity Yes No No Yes
Capability FRT (LV/HV) No Yes No Yes
Reactive power control in response to fast
Capability No Yes No Yes
and large voltage variations
Capability Power oscillation damping Yes *** Yes* No No
* If the DC link is included in the model
** For small isolated grid
*** If the damping controller is designed to damp the common mode of frequency evolution
**** External protection relay model may be used such as zero sequence overvoltage protection model for single

line-to ground fault

***** Change to No if the isolated network is LV network only

113
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The IBG models used for analysing the dynamic behaviour in bulk system studies must be appropriately
detailed to capture the phenomena of interest for the type of analysis considered. At the same time, they
should not increase the computation burden so as to be suitable for large-scale studies. Usually, these
studies are performed using positive-sequence, time-domain, simulations and the analysis is mainly
focused on how the IBGs react to large disturbances in the power system. For that reason, the very fast
dynamics of IBGs can be usually neglected, as the transient processes in IBGs are much faster than
those of synchronous machines. Furthermore, the presence of a DC source and a fast DC control (“stiff
DC”) allows the decoupling of the generation unit from its interface to the grid. These observations led
to the widespread use of RMS models to examine the influence of IBGs in response to large
disturbances in bulk power system studies.

Figure 5.1 Overview of IBG model.

In the remainder of this chapter, the main components of RMS-type IBG models will be presented and
analysed, relating to the functionalities listed in Table 5.2 and the modelling assumptions discussed
above. The model description is split into three levels as shown in Figure 5.1. Firstly, the individual
inverter interface components are detailed, assuming a current source model, along with the basic
control and protection functionalities. Details concerning the interface to the RMS simulation software,
the averaged inverter model, and the synchronization and power tracking blocks are considered under
the “stiff DC source” assumption. Secondly, the local (generator-level) active and reactive power control
blocks are presented, in connection to the frequency and voltage control requirements discussed in
Chapter 3 and the IBG capabilities. Thirdly, the plant-level active and reactive power control blocks are
presented. The chapter concludes with some insights about the future challenges about RMS IBG
modelling.
5.4 INVERTER MODEL, CONTROL AND PROTECTION
5.4.1 Interface models
The common feature of generators considered in this TB is the power electronic interface that connects
them to the grid. From a modelling point of view, a generic inverter interface model, including the DC
side dynamics, can be used for all considered sources. This provides a modular modelling and testing
approach and decreases replications and the amount of work needed for each model. Such modular
models have been presented in [105], [107]–Erreur ! Source du renvoi introuvable..

Figure 5.2 Inverter generator model.

Figure 5.2 shows the current source model used in this TB. The model receives the active and reactive
current command from the higher-level controls, implements the necessary inverter-level protections
and controls, and outputs the active and reactive current that is injected into the grid.
5.4.2 Active and reactive current transformation
Under the RMS approximation all current and voltage phasors refer to rotating orthogonal axes, denoted
by x and y respectively in Figure 5.3. Those axes rotate at the angular speed ωref imposed by the
simulation software. The outputs, active (IP) and reactive (IQ) currents are depicted in the vector diagram

114
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

of Figure 5.3. It is the function of the simulation program to perform the transformation from the d-q to
the x-y of reference, as shown in the figure. This is done by a simple axes transformation. For instance:
𝐼𝑥 sin(𝜃) cos(𝜃) 𝐼𝑄
[𝐼 ] = [ ][ ]
𝑦 −cos(𝜃) sin(𝜃) 𝐼𝑃
Equation 5.1

Figure 5.3 Reference frame for phasors and active/reactive currents.

Three phase inverters normally calculate active and reactive power based on the d-q coordinate. The
symmetrical coordinate is transformed into the - coordinate first, then the - coordinate is further
transformed into the d-q-0 coordinate. Some manufacturers calculate the power using only positive
sequence quantities. On the other hand, single phase inverters operate independently for each phase,
which means, the IP and IQ for three-phase, i.e. IP-a, IP-b, IP-c, IQ-a, IQ-b, and IQ-c are determined separately.
In the case of unbalanced fault, the injected three active currents and three reactive currents could
behave differently and the composition of the three vectors of IP-a / IQ-a, IP-b / IQ-b, and IP-c / IQ-c are
most unlikely to be represented using the symmetrical coordination without zero-sequence current
component. Most of the commercial software provides only positive-sequence current injection function,
while some commercial software provides both positive- and negative sequence current injection
functions. If the positive- and negative sequence current injection functions are provided, IP-a / IQ-a, IP-
b / IQ-b, and IP-c / IQ-c can be independently expressed using the two functions [115] and the composition
of the three vectors are represented superimposing IP-a / IQ-a, IP-b / IQ-b, and IP-c / IQ-c (See Appendix 5-
A).
5.4.3 Inverter
This inverter model defines the interface between the IBG and the grid. It consists of an algebraic,
controlled current source that computes the required current injected into the network in response to the
real active and reactive current commands from the electrical control described in Sub-Chapter 5.5 (see
Figure 5.2). This controlled current source also incorporates various inverter constraints, limitations,
protections, and the fast-acting inverter controls that mitigate over-voltages by reducing reactive current
output. Some of these are detailed below.
5.4.3.1 Low voltage power logic
An active current management logic can be employed during low voltage events to reduce the system
stress during the fault and immediately after. One such example control block is shown in Figure 5.4
(includes the low voltage power logic as well as some peripheral control elements).
In this control block, the Low Voltage Power Logic (LVPL) reduces system stress during and immediately
following sustained faults by limiting the real current command with both a cap and a ramp rate limit.
Under normal operating conditions, the filtered terminal voltage is above the user-specified breakpoint
(brkpt) value and the upper limit is deactivated. When the voltage falls below the breakpoint (e.g., during
a fault), a limit is calculated and applied to the output current. When the voltage is below a user-specified
zero-crossing point (zerox), the limit becomes zero. The user-specified ramp rate limit (rrpwr) restricts
the active current command increase rate during the post-fault power recovery period, until the voltage
recovers up to the breakpoint and such ramp rate limit is released.

115
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

iqrmax

-1 iq High voltage reactive

iqcmd
1+sTg current management

iqrmin Vt Interface to
network model
LVPL & rrpwr

1 ip
ipcmd
1+sTg ×
LVPL
lvpl1

Low voltage
1
active current
1+0.02s
management
Zerox Brkpt V
Low voltage power logic

Figure 5.4 Example of converter with LVPL block diagram with peripheral blocks [114].

5.4.3.2 Current limiter

An example of the current limiter model for PVs is given in Figure 5.5. This control block receives the
active (Ipin) and reactive (Iqin) current inputs as well as a flag (PQ flag) denoting which one has priority. If
the user selects reactive power priority, then the reactive current output (Iq_cmd) is allowed to vary
from -Imax to Imax, while the active current output (Ip_cmd) is limited by the value of the reactive current. So,
when active power priority is chosen, the active current output is limited between 0 and Imax, while the
reactive by the value of the active.

PQ _ flag

Current limit logic

 I qmax  I max

Q priority :  I qmin  - I max
I pin
 I p _ cmd
( PQ _ flag  0)  I p max  I max 2
- I q_2 cmd
I
 p min  0
 I p max  I max
I 0
P priority :  p min

( PQ _ flag  1)  I qmax  I max - I p _ cmd
2 2

 I q_ cmd
 I qmin  - I max - I p _ cmd
2 2

I qin I q_ cmd  lim( I qin , I qmin , I qmax )

I p _ cmd  lim( I pin , I p min , I p max )

Figure 5.5 Example of current limiter block diagram of PVs [113].

5.4.3.3 High Voltage Ride-through (HVRT)

In this component model, the terminal voltage is compared to an upper voltage threshold and if it is
exceeded, then some extra reactive current is absorbed by the inverter. The current change relates to
the voltage rise as follows:

116
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

𝛥𝐼𝑞𝐻𝑉𝑅𝑇 = 𝐾𝐻𝑉𝑅𝑇 𝛥𝑈
Equation 5.2
where:
 KHVRT is the reactive current rise factor,
 ΔU is difference between the terminal voltage and the threshold,
The HVRT event should end when a reset threshold voltage is crossed for a specific time. The reactive
current absorption should be maintained for a specific time after the end of the voltage swell (see Figure
5.6 (i)).

(i) HVRT control model [114]

Active current with up-ramp

 max( K q_ HV (U HV - U term ) + K Iq 0 _flag I q 0 + I q 0 _ HV , I qmin_HV ) (U term  U HV ) rate limit on voltage
 recovery
I q   I q_cmd (U LV  U term  U HV ) dIp_LV
 (U term  U LV )
I p _ cmd  min( K q_LV (U LV - U term ) + K Iq 0 _flag I q 0 + I q 0 _LV , I qmax_LV )
Ip

 I p _cmd (U LV  U term  U HV )
Ip  
 I
 p_FRT (U term  U LV U term  U HV )
0
 min( I p _cmd , I max _ FRT - I q )
 ( I p _flag =0) Iq
I q_ cmd 
I p_FRT   min( I p _cmd , I max _ FRT - I q )
2 2
( I p _flag =1)

 K I + K U + I ( I p _flag =2)

p1_ FRT p0 p2 _ FRT term p0 _ FRT

where;
Uterm: Inverter AC voltage
UHV: Threshold value of HVRT
ULV: Threshold value of LVRT
Ip_flag: Active current limit flag during FRT (e.g. HVRT or LVRT)
Ip_FRT: Active current during FRT
Imax_FRT: Maximum current during FRT
Kp1_FRT/ Kp2_FRT: Active current factor during FRT
Ip0: Pre-fault active current
Ip0_FRT: Initial active current during FRT
Ip_cmd: Active current control signal at local level
(ii) FRT model including both the HVRT and LVRT models [113]
Figure 5.6 Example of HVRT block diagram of PVs [113], [114].
Figure 5.6 (ii) shows the situation where both LVRT and HVRT are activated. When the terminal voltage
of the IBG (Uterm) is larger than the threshold value (UHV) the reactive current (Iq) is controlled via a
decrease in the limiter of Iq that results in Uterm becoming equal to or less than UHV. The coefficient Kq_HV

117
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

is chosen based on the sensitivity of the voltage to the reactive power injection (𝑑𝑉/𝑑𝑄), which can
vary depending on the external network.
5.4.3.4 Frequency Ride-Through (ROCOF Immunity)
The functionality and impact of this block is described in Clause 3.2.5. It is mainly aimed at preventing
the ROCOF relay or the anti-islanding protection from operating when the ROCOF is smaller than a
threshold value, e.g., 2 Hz/s in Japan. Thus, the frequency ride-through and the anti-islanding
protection need to be coordinated. Figure 5.7 shows an example of the control block diagram for the
frequency ride-through capability function. The model output consists of a blocking signal to the
ROCOF relay or the anti-islanding protection. Alternatively, we could ensure that the setting value of
the ROCOF relay is no less than the threshold value of the ROCOF immunity level. Yet, not all anti-
islanding relays utilize exactly the same ROCOF function. Thus, the aforementioned coordination
becomes important.

ROCOF relay or anti-islanding relay

ROCOF immunity
&
Setting value of ROCOF immunity (e.g. 2Hz/s) +
1
＋ 1
-
Frequency calculated from the 1
terminal voltage of the inverter 1+Ts － DT

DT

Dead time

Figure 5.7 Example of frequency ride-through block diagram.

5.4.3.5 Low Voltage Ride-Through (LVRT)

The importance and functionality of this protection is described in Sub-Chapter 3.3. LVRT describes the
requirement that generating plants must continue to operate through short periods of low grid voltage
and not disconnect from the grid. An example of such a requirement is shown in Figure 5.8. This curve
shows that if the IBG terminal voltage remains above the curve after a disturbance, then the unit should
not disconnect. If the voltage crosses this curve, then the IBG is allowed to disconnect. In LVRT
capability studies, the IBGs are generally assumed to be disconnected from the grid according to their
given LVRT characteristics even if the real IBG might remain connected. From a modelling perspective,
this is an additional undervoltage relay model with more complex time-voltage characteristics. A typical
difficulty of the modelling is how to deal with the second fault during or after the first fault.

Figure 5.8 Example of LVRT characteristics [17].

An example implementation in pseudo-code is presented in Figure 5.9. In this example, the delay of the
tripping relay is not modelled.

118
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

-
- 1 S + + + 1
Vt
0 R
timer X 0
+ 0 - + 0
V3 t1 V2-V1 V1
t2-t1
- 1
0 X
+ 0
+ 1
V2
0 X 1
Trip
- 0
- 1 t2
0 + -
+ 0 t3
V1
1
0
0

Figure 5.9 Example pseudo code of LVRT characteristics [116].

5.4.4 Phase Locked Loop (PLL) [117]

As explained in Clause 4.5.2, the objective of a PLL block is to determine the phase angle of the terminal
voltage. In RMS models, this angle is used to adjust the current injected into the grid with the proper
phase angle with respect to that voltage, so that the active current 𝐼𝑃 and reactive current 𝐼𝑄 have the
desired values.
A phasor diagram relevant to PLL dynamic simulations (RMS phasor) is given in Figure 5.10a. The
orthogonal x and y axes are reference axes, rotating with speed 𝜔𝑟𝑒𝑓 , on which all time-varying phasors
are projected when deriving the network equations. Thus, 𝑣𝑥𝑚 and 𝑣𝑦𝑚 are the projections of the voltage
phasor 𝑉̅𝑚 , while 𝜃𝑟 is the “real” phase angle of the same phasor with respect to the x axis. The
orthogonal d and q axes are rapidly adjusted by the PLL in such a way that, in steady state, the d axis
coincides with the voltage phasor. They rotate with speed 𝜔𝑝𝑙𝑙 . Hence, in steady state:
 the q and 𝜃𝑟 angles coincide
 the projection 𝑣𝑞 of the phasor 𝑉̅𝑚 on the q axis is zero
 the projection 𝑣𝑑 of the phasor 𝑉̅𝑚 on the d axis is equal to the voltage magnitude.

When the d-axis coincides with the voltage phasor, the projection 𝑖𝑑 of the current phasor on the d-axis
(see Figure 5.10a) is the active current, i.e.
𝑖𝑑 = 𝑖𝑃
Equation 5.3
while the projection 𝑖𝑞 on the q-axis is the reactive current changed sign, i.e.
𝑖𝑞 = −𝑖𝑄
Equation 5.4
The minus sign comes from the fact that the q-axis has been chosen arbitrarily ahead of the d-axis.
A generic model of PLL is given in Figure 5.10 (b). The terminal voltage 𝑉𝑡 is measured with a time
constant 𝑇𝑚 , to obtain 𝑉𝑚 . The same time constant is applied to the measured components 𝑣𝑥𝑚 and 𝑣𝑦𝑚 .
𝑖𝑑 and 𝑖𝑞 are determined by other controls of the inverter. 𝑖𝑥 and 𝑖𝑦 are the components of the current
injected into the network, in the (𝑥, 𝑦) reference axes.
The inner loop in Figure 5.10 (b) involves a PI controller which forces vq=0 in steady state. The output
of the PI controller is the speed 𝜔𝑝𝑙𝑙 of the rotating dq frame. By subtracting from 𝜔𝑝𝑙𝑙 the reference
speed 𝜔𝑟𝑒𝑓 , the rate of change of angle of the angle θ is found. The “estimated” phase angle θ of the
voltage is found by integrating the aforementioned rate of change. Since the other controls of the inverter
only know the phase angle q , they operate by assuming that Eqs. (5-3) and (5-4) hold true at any time.
Hence, after a disturbance, there is a discrepancy between the desired and the effective active and
reactive currents during the transient period.
Based on the value of q , the (𝑖𝑥 , 𝑖𝑦 ) components of the current are related to the (𝑖𝑑 , 𝑖𝑞 ) as shown in
Figure 5.3. Typical values for 𝑇𝑚 are 10-20 ms and for the gain 𝑘𝑝𝑙𝑙 between 30 and 60.

119
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

(a) Phasor diagram of PLL behaviour

(b) Generic PLL block diagram

Figure 5.10. Generic PLL structure and operation [117].

iqrmax

-1 iq High voltage reactive

iqcmd
1+sTg current management

iqrmin Interface to
Vt
network model
LVPL & rrpwr

1 ip
ipcmd
1+sTg ×
gain
1
LVPL
0
V
lvpnt0 lvpnt1
Low voltage active
current management

Figure 5.11 Example of simplified PLL behaviour [114].

When the measured voltage drops below some threshold 𝑉𝑝𝑙𝑙 , due to a fault in the AC grid, the PLL is
usually “frozen”. This is taken into account by the lower switch block in Figure 5.10 (b), which sets to
zero the input of the integrator, thus keeping q to a constant value.
In some cases, the exact behaviour of the PLL is not required and only its effect on the IBG output
during low voltage conditions is included. One such example is shown in Figure 5.11. The “low voltage
active current management” block roughly approximates the PLL behaviour during low voltage
conditions by decreasing the active power output.

120
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

5.4.5 Maximum Power Point Tracking (MPPT)

The functionality of this control and the impact on the system behaviour is discussed in Clause 3.2.5.
MPPT techniques are used when the power from the primary source to the IBG varies with time (e.g.,
the solar radiation in PVs, wind in WTs, etc.) and these are then utilized to always extract the maximum
power output. However, in most dynamic stability studies (which span up to several seconds) the
variability of the primary source is not considered due to the small time-constants involved. So, the
MPPT reference is treated as a constant value and is not explicitly modelled unless the DC voltage
needs to be modelled (see Clause 5.4.6).
5.4.6 DC source control
As explained in Sub-Chapter 2.4, in many studies the stiff DC source assumption is made and the DC
source control is not explicitly considered. However, there are some cases that it needs to be
considered. One such example is described below.
In some countries, fast reduction of active power is required when frequency exceeds a certain upper
limit (e.g., 50.2 Hz was specified in the past in Germany). The reduction follows a specific power-
frequency droop characteristic. When the grid-side converter reduces active power quickly, the DC
voltage can rise. Thus, it might be required to use the IBG chopper to dissipate this energy. Alternatively,
the primary source of the IBG (e.g., the machine side converter in wind turbines) has to synchronously
reduce the infeed to the DC circuit as well, in which case further control actions might be required on
the primary source (e.g., the WT rotor might accelerate and the pitch control used to adapt the set point).
If a fixed DC voltage is assumed, this behaviour is not visible. Even so, this behaviour is specific to
certain IBG types. Thus, in some cases the DC source control could be safely ignored, and in others
not. For example, in the case of significant voltage drop, the DC protections are likely to be operated
which result in the disconnection of the IBGs. In this case, the DC source control model needs to be
used (see also Clause 3.3.4).
In the case of PVs, the DC source model is equivalent to the PV array model, while the DC control model
is equivalent to the DC capacitor model. The PV array model is illustrated in the Appendix 5-B and only
the DC capacitor model is briefly introduced.
To track the maximum power point of the PV array, the PV should control the DC bus voltage of the DC
capacitor, shown in Figure 5.12.
DC Capacitor DC/AC converter

I array I dc Ut U term
I ac
U dc xT P + jQ
C

(a) Outline of PV inverter

I array
P 1 I dc - 1 U dc
÷
 sC

(b) DC capacitor model

Figure 5.12 Example of DC source model [118].
P is the output power of PV inverter and η is the efficiency of PV inverter. The DC power is converted
to AC power through DC capacitor and DC/AC converter. The DC capacitor could be modelled as
following equation:

121
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

1
I dc  I array - U dc
sC
Equation 5.5
where:
 𝐼𝑑𝑐 denotes DC current of DC/AC converter;
 𝐼𝑎𝑟𝑟𝑎𝑦 denotes the output current of PV array;
 𝑈𝑑𝑐 denotes the DC voltage of PV array;
 𝐶 denotes DC capacitor.
5.4.7 Measurements
Electric quantities like voltage, current, active power, reactive power and frequency are measured and
usually reduced to a DC quantity, i.e. steady-state based quantity. While the filtering and processing
may be complex, it can usually be represented, for RMS-type modelling purposes, by a single first order
lag element with an equivalent time constant. For many systems, this time constant is small (in the range
of 0.01 to 0.02 s) and provision should be made to set it to zero [119]. Single-phase voltage and current
sensing, in general, requires a longer time constant in the sensing circuitry to eliminate ripple. Reliable
frequency measurement, especially under non-stable conditions, requires filtering and signal processing.
The resulting delay should be taken into consideration.
5.4.8 Interface protection systems
In many studies, it is necessary to model the inner and outer inverter protections (see Clause 2.3.3).
Once a protection operates, the inverter stops for a certain time before restarting (e.g., in the range of
150 seconds - 300 seconds for residential PV inverters in Japan). As shown in Figure 5.13, the inverter
protection operation results in the power electronic device blocking the inverter (i.e., the circuit breaker
of the PV does not open) and the short disconnection time of 10 seconds.

Release OVR, UVR,

On delay
OFR, UFR
timer

Operation of active 150-300

anti-islanding relay seconds

Operation of passive
anti-islanding relay

Standby
Release voltage rise On delay
mitigation func. timer No
Meet with grid interconnection
Release inner 10 requirement?
inverter protection seconds
Yes
Restart
Release malfunction
detection of inverter

Figure 5.13 Example of restarting sequences for PVs in LV network [120].

However, once one of the voltage, frequency or active anti-islanding protection operates, the circuit
breaker of the PV opens and the PV is disconnected from the grid for 150 - 300 seconds. This way, the
operation of the OVR/UVR, OFR/UFR and active anti-islanding protection is treated as the existence of
a system fault. Because it takes 150 - 300 seconds to search for the fault section using the reclosing
system, the PV is not allowed to restart until after the aforementioned protections operate. If the
simulation time is small (e.g., 30 seconds in short-term simulations), such protection operation can be
treated as a permanent disconnector. Figure 5.13 shows an example of restarting sequence for
residential PV inverters in Japan. Figure 5.14 also shows an example of grid protection system of PVs.
These are explained in more detail below.

122
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

High/Low Voltage detection Delay time

UHVP2 < Uterm > tHVP2

U term UHVP1 ≤ Uterm ≤ UHVP2 > tHVP1

ULVP2 ≤ Uterm ≤ULVP1 > tLVP1

Uterm < ULVP2 > tLVP2

trip_flag
&
High/Low frequency detection Delay time

fHfP2 < f > tHfP2

fHfP1 ≤ f ≤ fHfP2 > tHfP1

f
fLfP2 ≤ f ≤ fLfP1 > tLfP1

f < fLfP2 > tLfP2

Figure 5.14 Example of grid protection system control block diagram [113].

5.4.8.1 Voltage relay models

These protection relays model the over- and under-voltage tripping signals of the IBG. Thus, the terminal
voltage is monitored against the over- and under-voltage thresholds with associated time-delays. Figure
5.15 shows an example of those relay block diagrams for single-phase inverters in Japan.

A phase voltage with relative to

neutral point between A and C phase 1
1+Ts +

Setting value of OVR

e.g. 1.15 p.u.
-
On delay
timer

e.g. 1.0
C phase voltage with relative to neutral 1 seconds
point between A and C phase 1+Ts +

Setting value of OVR

E.g. 1.15 p.u.
-

A phase voltage with relative to 1

neutral point between A and C phase 1+Ts -

Setting value of UVR

e.g. 0.8 p.u.
+
On delay
timer

1.0
C phase voltage with relative to neutral 1 seconds
point between A and C phase 1+Ts -

Setting value of UVR

e.g. 0.8 p.u.
+

Figure 5.15 Example of over- and under- voltage relay models for single-phase PV inverter.
If unbalanced faults are not considered, the positive-sequence terminal voltage is compared with the
predefined thresholds and, if the latter are crossed, specific tripping signals are activated with the time

123
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

delay specified by the on-delay timer. The voltage at the terminal of the inverter is normally used with a
certain time delay, often set to 0.01 to 0.02 second as the response time of a transducer.
5.4.8.2 Frequency relay models
Similarly, to the previous protections, this relay model represents the over- and under-frequency tripping
signals. In this model, the measured frequency is monitored and compared against predefined
thresholds and if these are crossed, specific tripping signals are activated with a time delay specified by
the on-delay timer. Figure 5.16 shows the example over- and under-frequency relay block diagrams for
single-phase inverters in Japan. The frequency measurement at the terminal of the inverter is normally
obtained with a certain time delay. The time delay which is not actively revealed but included in Figure
5.16 (See pink heavy line) is often set to 0.01 to 0.3 second as the time constant of the first-order lag
element of a transducer.

Frequency calculated from the

terminal voltage of the inverter + On delay
timer

-
Setting value of OFR e.g. 1.0
e.g. 51.0/61.2 Hz seconds

Frequency calculated from the

terminal voltage of the inverter - On delay
timer

+
Setting value of UFR e.g. 1.0
e.g. 49.0/58.8 Hz seconds

Figure 5.16 Example of over- and under-frequency relay models for single-phase PV inverter.
A brief overview of recommended protection settings based on European regulation is presented in [17],
while an example of selecting values for the voltage and frequency protections thresholds is shown in
[23].
Similar protections are described in [121]–[123] in the case of fuel-cells. According to those references,
for secure operation, the fuel-cell units are assumed to be disconnected from the network if they are
operating far from their rated values. This is important to protect the power electronics converter of these
units. A time delay of five cycles is assumed for measurements and disconnection of any fuel cell after
reaching the critical limit.
5.4.8.3 Instantaneous overvoltage relay model
Contrary to the over-voltage relays, the instantaneous over-voltage relays are sometimes present in the
inverters as inner protections. Figure 5.17 shows an example of such a relay block for single-phase
inverters in Japan. If unbalanced faults are out of scope, the positive-sequence terminal voltage is
compared with the predefined thresholds and, if the latter are crossed, specific tripping signals are
activated with the time delay specified by the on-delay timer. The voltage at the terminal of the inverter
is normally used with a certain first-order lag. The time constant of the first-order lag element is often
set to 0.01 to 0.02 second as the response time of a transducer.

124
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

A phase voltage with relative to

neutral point between A and C phase 1
1+Ts +

Setting value of OVR

e.g. 1.2 p.u.
-
On delay
timer

e.g. 0.1
C phase voltage with relative to neutral 1 seconds
point between A and C phase 1+Ts +

Setting value of OVR

e.g. 1.2 p.u.
-

Figure 5.17 Example of instantaneous overvoltage relay models for single phase PV inverter.

5.4.8.4 Passive anti-islanding protection

In some studies, it is necessary to model the anti-islanding protection of the IBG. Figures 5.18 and 5.19
show some examples of passive anti-islanding protections based on
1) Voltage phase jump method, and
2) Rate of change of frequency (ROCOF) method.
Figure 5.18 depicts an example of a voltage phase jump type passive anti-islanding protection which is
currently used in Japan. The model involves a selector switch function so that one model can represent
the four different approaches currently available. The user selected value of VPJ1 (in degrees) defines
the frequency for the passive anti-islanding protection relay to operate. For example, a VPJ1 of 3
degrees is equivalent to 0.414 Hz where the fundamental period is 20 ms for a 50-Hz system.
Figure 5.19 shows an example of a rate of change of frequency type passive anti-islanding protection
which is currently used in Japan. Although the name of the anti-islanding protection is ROCOF, this
protection covers two aspects: ROCOF and OFR/UFR. If the specified values of the moving average
are large, the relay characteristics approach OFR/UFR. If the setting values of the moving average are
small, e.g. 0.1 s, the relay characteristics approach ROCOF.
VPJ1
|ΔT1|＞
360×Rated Frequency
e.g. 3 deg
1.0
1.0

+ ΔT1
1
Frequency x-1 Moving average Dead time
-
On delay timer
1+0.01s
e.g. 1s e.g. 0.12 second
Type 1-3

e.g. 1-32 cycle

Time delay
of transducer
0.0

Moving average VPJ1 Selector switch function

|ΔT1|≦ FIX=1.0 (Type 1-3) Dead time
e.g. 1-32 cycle 360×Rated Frequency
FIX=0.0 (Type 4)　　　　
e.g. 3 deg e.g. 0.0 second
Type 4

VPJ2
|ΔT2|＞ Dead time
360×Rated Frequency &
Time delay
1.0
0.0

+ ΔT2
Moving average Dead time
-
e.g. 1-32 cycle

0.0

Moving average VPJ2

|ΔT2|≦
e.g. 1-32 cycle 360×Rated Frequency

Figure 5.18 Example of voltage phase jump type passive anti-islanding protection block diagram
[124].

125
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Setting value of frequency deviation -

e.g. 0.2 Hz
On delay
＋ timer
1 Moving +
Frequency &
1+Ts average
e.g. 0.5
－
e.g. 0.01s e.g. 0.04s seconds

Moving
Dead time
average
e.g. 0.0s
E.g. 1.0s

1 -
|VAB|
1+T1S
Setting value of UVR +
e.g. 0.01s
e.g. 0.7272 p.u.
1 -
|VBC|
1+T1S & 1
Setting value of UVR +
e.g. 0.01s
e.g. 0.7272 p.u.
1 -
|VCA|
1+T1S
Setting value of UVR +
e.g. 0.01s
e.g. 0.7272 p.u.

Figure 5.19 Example of rate of change of frequency type passive anti-islanding protection block
diagram [124].

5.4.8.5 Active anti-islanding protection

Some IBGs are equipped with active anti-islanding protection. This type of protection is relevant to
various types of phenomena described in Chapter 3 and is detailed in Sub-Chapter 3.6. Figure 5.20
depicts an example of a reactive power variation type active anti-islanding protection. Because the
starting time of the injection from the IBGs is not always the same, it is easy to predict that the injection
signal can be cancelled if more than two IBGs with this anti-islanding scheme are connected to the same
feeder, which leads to failure to detect the islanding status due to insufficient reactive current injection
(see Sub-Chapter 3.6). In order to overcome the technical challenge, GPS signals have been used for
synchronizing the injection signal of the sinusoidal wave.

Sinusoidal wave

Rectangular wave

-
+
+
|x|
× + × Reactive current

|Df|＞Fact K 1
e.g. 0.0
1.0

1 + Df
Frequency
1+0.01s
Moving average Dead time
-
Dead time
e.g. 0.0 second
× On delay timer
E.g. 0.7 second
Dead time
e.g. 0.0 second
e.g. 1 cycle e.g. 10 seconds
Time delay
of transducer
0.0
Moving average

e.g. 1 cycle |Df|≦Fact

Figure 5.20 Example of reactive power variation type active anti-islanding protection block diagram
[124].
Figure 5.21 depicts another example of active anti-islanding protection based on frequency shift. It must
be noted that the sign of the injection signal would be fixed once the frequency deviation exceeds a
certain threshold value in the case of the rectangular wave, while the magnitude of the injection signal

126
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

would increase once the frequency deviation exceeds a certain threshold value in the case of the
sinusoidal wave. It can be easily seen that the injection signal can be cancelled if more than two IBGs
with this anti-islanding scheme are connected to the same feeder, which leads to failure to detect the
islanding status due to insufficient reactive current injection (see Sub-Chapter 3.6).

Sinusoidal wave

Rectangular wave |Df|＞Fset

Selection of
Injection Signal 1.0

1
Fix Sign of Injection Signal
1+T s
e.g. 0.1

0.0

|Df|≦Fset e.g. 0.02 Hz

-
+
+
|x|
× + × Reactive current

|Df|＞Fact K 1
e.g. 0.9
1.0

1 + Df
Frequency Moving average Dead time On delay timer Dead time
1+0.01s -
E.g. 0.5 second e.g. 0.2 second
e.g. 32 cycle e.g. 0 seconds
Time delay
of transducer
0.0
Moving average
|Df|≦Fact e.g. 0.2 Hz
e.g. 2 cycle

Figure 5.21 Example of frequency-shift type active anti-islanding protection block diagram [124].

5.5 FREQUENCY AND VOLTAGE CONTROL AT COMPONENT/LOCAL LEVEL

An overview of the electrical control is shown in Figure 5.22. This control monitors the IBG active and
reactive power outputs as well as the terminal voltage and frequency, and computes the active and
reactive current commands to achieve the active and reactive power and terminal voltage set by the
upper control levels. The current commands are then passed to the interface model (see Sub-Chapter
5.4). The individual control components are detailed in this sub-chapter.

Figure 5.22 Local frequency-and voltage controller [111].

5.5.1 Reactive power control

Reactive power control at the local level can be divided into two types: 1) static or slow control, or 2)
dynamic control. The first type reacts to slow variations of the operating point by adjusting the reactive
power output according to some control mode, for instance constant reactive power or constant power
factor. The second overrides the first during significant voltage dips/sags and controls the reference of
the reactive power output. The two controls are complementary, acting at different time-scales. This
clause describes the two reactive power controls individually.
5.5.1.1 Static control
The objective of this control is to determine the reactive current command (Iqcmd) required so as to
provide a specific reactive power or satisfy a voltage control law. This may entail processing the

127
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

measured reactive power generation (Qgen). Several approaches can be used to define the reactive
power control (Qcmd), these are described as follows:
 Reactive power imposed by external reference: Reactive power tracks an external signal, which can
be a constant value (Qcmd=Qref).
 Voltage-dependent reactive power: The reactive power changes as a function of some measured
voltage Qcmd=f(V). This can be at the terminal bus, at a user-specified remote bus (e.g., the POI), or
a fictitious (synthesized) point in the power system.
 Constant power factor: The reactive power is computed based on the active power output to meet
a given power factor (cosφ=constant).
 Active power-dependent power factor: The reactive power is computed based on the active power
output to meet a power factor varying with active power (cosφ=f(P)).
One example model with several corresponding block diagrams is given in Figure 5.23. The selection
block shown in Figure 5.23 (a), allows switching between several reactive power control modes
(including constant power factor, unity power factor, constant reactive power, etc.).

(a) selection of Q control mode

(b) reactive power control

(c) Q droop function model

(d) electric control

Figure 5.23 Example of reactive power control block diagram [114].
An example of a reactive power control block is shown in Figure 5.23 (b). It should be noted that the
parameter fN shown in Figure 5.23 (b) is used for WTGs. In the case of PV models, fN is set to 1. The
regulator consists of a PI controller. The time constant Tc reflects the delays associated with cycle time,
communication, and additional filtering in the controls while the voltage measurement lag is represented

128
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

by the time constant Tr. The PI control gains, Kpv and Kiv, are field adjustable to meet performance
objectives and may be adjusted in the model. Higher gains will give better response to grid voltage
disturbances. However, higher gains result in increased risk of instability – not unlike the way AVR gains
can destabilize conventional synchronous machines. As the external system weakens, the effective
closed-loop response gets faster. Thus, selection of higher gains for system performance must be
accompanied by analysis that assures stable operation under all credible operating conditions –
especially the minimum short circuit strength condition.
The Q Droop function (Figure 5.23 (c)), is a relatively slow-acting function that reduces the effective
voltage reference (Vrfq - Vqd) as reactive power changes (see also Figure 5.24). This improves
coordination between multiple integral controllers regulating the same point in the system. By default,
the Q Droop function is disabled. It may be enabled by setting the gain parameter, Kqd, to a nonzero
value. There are three options for the reactive power input Qinput shown in Figure 5.23 (c): reactive power
generated by the PV plant, reactive power flow in a user-specified branch, or a synthesized reactive
power. The electric control which is responsible for the reactive current is also shown in Figure 5.23 (d).
Another reactive power control block diagram is shown in Figure 5.24. This control block diagram can
switch the power factor mode to the reactive power control mode using the flag PF_flag. A closed-loop
method or an open-loop method can be selected using the flag Q_flag. In the latter case, the inverter
AC voltage, Uterm is used for deriving Iq. It is noted that Q_flag=0, i.e. the open-loop method is mainly for
the long-term system studies. In other words, the main purpose of the open-loop control is to change
the PV reactive power output when the fast dynamics are negligible. Besides, an additional flag Iq_flag
is used to disable the reactive power controller and keep the reactive current constant. The input signals
Qord and PFref come from the plant level controller or are set equal to the power flow if the plant level
controller is not included.

1
U term
1 + sTm 1
÷ 1 + sTinverter I max
Reactive power control

Pmea
dQord_ max 0
PFref 1 - PF2
Q _ flag
ref

PFref
× 0 Qref
Kq +
1
PF _ flag 1 ramp limit 1
sTq
dQord_ min
Qord Delay
- I q _ flag
1 I qin
- I max
(Tq_ord) Iq 0 0
1 Qmea
Q
1 + sTm

Figure 5.24 Example of reactive power control block diagram [114].

5.5.1.2 Reactive power control in response to fast and large voltage variations
The objective of this model is to detect voltage sags and compute the reactive current injections required
during short-circuits in the network. During the voltage sag the inverter supplies the grid with reactive
current proportional to the amplitude of the voltage sag (based on the average three-phase voltage).
Usually, the reactive current injection during voltage sag is computed using the following formula:

DU
I qLVRT  I qini + K LVRT In
U ini
Equation 5.6
where:
Iqini is the reactive current injected before fault,
KLVRT is the reactive current droop factor (may take values between 0 and 10, depending on the
inverter type),
ΔU is the grid voltage drop/rise during the fault,
Uini is the voltage level before fault,
In is the inverter rated current.

129
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

5.5.2 Active power control

Similarly, to the control of reactive power considered in Clause 5.5.1, the objective of this control is to
determine the active current command (𝐼𝑝𝑖𝑛 ). Several approaches exist for defining this value, and two
examples are explained below and shown in Figure 5.25:

1 Pmea
P
1 + sTm
Pord Delay dPord_ max
I max
Active power control

(Tp_ord) ramp limit

dPord_ min - Reset if active
power recovers
Pref 1
Kp + after fault clear
Pm 1 dPm _ max min sTp
1 + sTmpp
ramp limit
- I max
0 I pin
P _ flag
1 1
U term
1 + sTm ÷ 1 + sTinverter
1

Figure 5.25 Example of active power control block diagram [113].

5.5.2.1 Active power imposed by external reference (P_flg=0):

In this mode of operation, active power tracks an external signal (Pord) with some delay. This signal is
limited by the maximum available power (Pm) and compared to the measured active power of the IBG
(Pmea). Finally, a PI controller drives the current setpoint. The active power is likely to be controlled for
the additional requirement/capability, such as the LFSM-O and FSM. The ramp rate limits shown in
Figure 5.25 are employed to represent such capabilities.
5.5.2.2 Voltage-dependent active power (P_flg=1):
In this mode of operation, the active power current changes as a function of some measured voltage
(Uterm) and the reference power (Pref). This can be the terminal bus, a user-specified remote bus (e.g.,
the POI), or a fictitious (synthesized) point in the power system. The main purpose of this type of control
is to change the PV output when the fast dynamics are negligible. This type of model is used mainly
when studying long-term dynamics.
Alternatively, voltage control can be performed by decreasing the active power with a fixed ramping rate.
In such a case, Pord in Figure 5.25 may be used for the reduction of the active power of the PVs instead
of the voltage-dependent active power control. Residential PVs which are connected to LV network may
be equipped with voltage control by means of active power. The representative flowchart of the voltage
rise mitigation function with characteristics of PV inverter is illustrated in Figure 5.26. Multiple strategies
for the voltage control by means of active power are implemented in the flowchart and some strategies
control not only the active power but also the reactive power. It is also noted that Figure 5.26 illustrates
the procedure of how to determine Pord in Figure 5.25.

130
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Feature 2 　　　Active power output (Pout)

　　　Representative Pout

Pout decreases while

P
PF is over 85%

Voltage
detection 0

Q*
PF = 85%
Calculation of Vmean
* Negative value denotes consumption of reactive power

No Feature 3
Vmean > V threshold?
No
Yes Vave > V thr MPPT
Decremented P = No
0?
PF > 85%?
No
Yes
Yes No
No
Increment of Q Q = 0?
(consumption of Q)
Yes Decrement of Q
Decrement of P (Supply of Q)

Increment of P

Feature 1
The initial Pref is not equal to Pini,
Rated capacity but rated capacity × Limit of PF
×Limit of PF
Initial output (Pini) Pout is constant while
Pref is over Pini
P

0 Active power output (Pout)

PF = 85% Active power reference (Pref)
Q*

Representative Pout

* Negative value denotes consumption of reactive power

Figure 5.26 Representative flowchart of voltage rise mitigation functions [125].

5.6 FREQUENCY AND VOLTAGE CONTROL AT PLANT LEVEL

In this sub-chapter, the control at the plant level is described. The major difference between the control
at the local level (Sub-Chapter 5.5) and the plant level is the existence of aggregated control quantity.
Thus, the plant level control is generally implemented for utility-scale PV power plants which consists of
multiple PV inverters, while the local level control is generally implemented on a single PV inverter.
Because of the scale of the generation, the utility-scale IBG is expected to participate in both frequency
and voltage control via the active and reactive power control of the IBG. This plant level control is
separately modelled with the electrical control in the inverter. In addition, such plant level control
sometimes needs to be coordinated with other controllers inside power plants such as SVC and
STATCOM.
Figure 5.27 shows a typical topology of a PV power plant. It includes several medium-voltage radial
feeders and step-up transformers from inverters to the feeders. Some plants include capacitors or other
reactive support systems that work in conjunction with the inverters to meet reactive power capability
and control requirements at the point of interconnection. A plant controller provides the power factor
reference to the inverters and if present, plant-level reactive power support equipment. It processes
measurements at the point of interconnection and commands issued from a remote operations centre
or from the transmission system operator.

131
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 5.27 Representative PV power plant topology [111].

Much of the existing PV generation in large networks consists of small dispersed PV systems connected
directly to the distribution grid. These systems do not typically have a plant controller, and the inverter
manages the grid interface. Some PV systems as large as 20 MW are directly connected to distribution
substations using a dedicated medium-voltage feeder.
Although there are some exceptions, PV plants are generally considered unable to be dispatched
because the energy source (solar irradiance) is variable. On the other hand, reactive power is able to
be dispatched within the capability of the inverters and plant-level reactive compensation.
5.6.1 Reactive power control
Reactive power control at the plant level can also be divided into two types: 1) static or slow control or
2) dynamic control. The first reacts to slow variations of the operating point by adjusting the reactive
power output according to some control mode, e.g. constant reactive power or constant power factor.
The second overrides the first during significant voltage dips/sags and controls the reference of the
reactive power output. The two controls are complementary, acting at different time-scales. This clause
describes the two reactive power controls individually.
The difference of the reactive power control at the local and plant level is the existence of coordination
between multiple IBGs and the enhanced capability of the voltage control. In the case of plant level
control, the reactive power flowing near the POI is measured and the reactive power output coming from
each inverter is adjusted in a coordinated manner. This coordination extends not only to multiple
inverters inside the same PV power plant but also to external reactive power compensators such as
SVC and STATCOM to achieve better performance of the voltage control.
The plant-level control module may include any or all of the following reactive power control modes:
 Closed loop voltage regulation (V control) at a designated bus with optional line drop
compensation, droop response and dead band.
 Closed loop reactive power flow regulation (Q control) on a user-designated branch, with
optional dead band.
Figure 5.28 shows the reactive power control block of a centralized plant controller used by the Western
Interconnection in North America (Western Electricity Coordinating Council, WECC). Additional dynamic
reactive power support, e.g. through STATCOM and SVC, can also be provided in the PV power plants
and the wind farms.

132
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

VcompFlag Vref
Ibranch 1 1 - +
|Vreg – (Rc+jXc)· Ibranch|
Vreg 0 1 + sTf lt r
Qmax
+ dbd emax
+ 1
Kp + Ki 1 + s Tf t
Qbranch Kc Ref Flag Qext
0 s 1 + s Tf v
emin Freeze st at e if
- Qmin Vreg < Vf rz
1
1 + sTf lt r
+
Qref

Figure 5.28 Plant level Q control block diagram for PV plants and wind farms [111].
Another reactive power control block diagram at plant level is shown in Figure 5.29. This control block
diagram can switch between power factor control mode and the reactive power control mode using the
flag PFPOI_flag. The reactive power control at plant level can also provide constant voltage control at
POI bus via the voltage droop control using the flag QPOI_flag. The response speed of the plant level
control is relatively slow compared to local level control. Such plant level control generally operates
correctly when the voltage is not unreasonably low.

U POI_ ref
U POI 1
- Qmax QPOI _ flag
1 + sTm
0
0 1 Qord
k qv QVPOI _ flag Kq _ POI (1 + ) 1
sTq _ POI
1
1
QPOI - Qmin
1 + sTm

1 QPOI _ref
PPOI
1 + sTm
0
PFPOI _ flag
PFPOI_ ref 1 - PFPOI_
2
ref

PFPOI_ ref
× 1

Figure 5.29 Plant level Q control block diagram for PV plants [113].
Not all plant level controls are assumed to employ dynamic reactive power support at the plant level.
Especially when a slow communication environment such as an EMS is used for detecting the measured
quantities, it is almost infeasible to apply the dynamic reactive power support at the plant level. For
example, the German grid code requires the support of grid voltage by injecting the reactive current
during a fault by the generation units.
Required positive sequence
reactive current deviation △IQ

[p.u.]
Dead band

Transmission
Code 2007
0.5 k
SDL Wind 2009 Voltage drop
or increase △U*

-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.1 0.2 0.3 0.4 0.5 0.6 [p.u.]

 1 
-0.5 U*     U* - U*
 1 + Ts 
I
Gain : k  Q
U
-1 k  0 - 10 p. u.

Figure 5.30 Dynamic voltage control characteristics [126].

133
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 5.30 describes the required behaviour during voltage sag or rise [126].
5.6.2 Active power control
With the decommissioning of large conventional units and the proliferation of IBGs, the latter will be
called upon to provide much (or in the future, all) of the ancillary services once provided by synchronous
machines. Depending on the type of the primary source and the existence or not of storage, IBGs can
participate in some or all of the depicted services.
Non-dispatchable IBGs (such as PVs, WTs, etc.) can participate in inertial and primary control (see
Figure 5.31) by not operating at their MPPT setpoint and allowing some flexibility (reserves) to provide
these services, or with the support of adequate energy storage. If such a reserve is not available, the
IBG can only react to overfrequency transients. On the other hand, dispatchable units (such as fuel-
cells, microturbines, etc.) can participate in other services. At the moment, most grid codes do not
require IBGs to provide such support and have strict qualification requirements to participate in the
ancillary services market.

Figure 5.31 Example of frequency control categories [127].

5.6.2.1 Primary frequency control

A plant controller may provide several active power control services, such as:
 Constant active power based on an external signal (Plantpref). During the simulation, this can be
set at the generator output in the solved power flow case
 Governor droop response with different characteristics for over- and under- frequency
conditions, based on frequency deviation at a user-designated bus.
Representative primary control block diagrams are shown in Figures 5.32 and 5.33. It can be seen that
Figure 5.32 includes the dead band for the frequency control and Figure 5.33 includes the ramp rate
limit for the change in active power output.

Plant _pref
Pmax Freq_f lag
+ f emax 0

Pbranch 1 - Kpg + Kig 1 Pref

1
1 + sTp 0 + s 1 + sTlag
Ddn f emin
f dbd1,f dbd2 + Pmin
-
Freg
+ +
Freq_ref Dup
0

Figure 5.32 Plant level primary frequency control block diagram for PV plants and wind farms [109].

134
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

D Pfdn _ max

f dn K pfdn
PPOI _ flag
- f up D Pfup _ max
Pmax 0
0 Pord
1 - - 1
f K pfup K p_ POI (1 + ) 1
1 + sTm sTp_ POI
-
0 dPref_ max Pmin
PPOI_ ref Ramp rate limit

dPref_ min
1
PPOI 1 + sTm

Figure 5.33 Plant level active power control block diagram for PV plants [113].

5.6.2.2 Inertial response (or synthetic inertia)

The inertial response has already been in commercial use. However, the grid codes are different from
country to country and from TSO to TSO. Wind turbine generators (WTGs) equipped with the inertial
response feature have been in operation on the Hydro-Québec system since 2012. This regulated
inertial response profile is shown in Table 5.3 and Figure 5.34. As shown in this figure, once the
frequency drop exceeds the specified threshold value, the predetermined series of control action is
taken. There are many other approaches for emulating inertial response. For example, ROCOF
schemes has been studied. However, those approaches are still at a research stage. Limiting the
frequency nadir should be the first priority for TSOs to cope with the technical challenge coming from
the low inertia network and only this type of control action shown in Figure 5.34 is currently taken in
some TSOs such as Hydro-Quebec and Ontario’s Independent Electricity System Operator.
The inertial response model has not yet completely prepared and still at model validation stage.
Therefore, the detailed control block diagram is not revealed and only fundamental representation of the
inertial response control is currently provided as shown in Figure 5.35. The example model validation is
introduced in [9]. Although there is overall a good match between the measured response and the
simulated response, the actual active power output did not always track the predetermined inertial
response profile. In addition, such active power output is different depending on the WTG type and
active power output level [9].
From modelling point of view, the response time of measurement is critical especially for emulating
inertial response. As already discussed in Chapter 2, such response time cannot be reduced to zero and
the inertial response cannot be completely emulated using IBGs. Therefore, the fast frequency response
instead of the inertial response is recommended as the technical term [9] (see Sub-Chapters 2.2 and
Clause 3.2.5-7). In order to consider the total response time of the fast frequency response, the
following elements need to be clarified.
 Measure – Measure and identify frequency deviation and fast frequency decrease.
 Identify – Identify occurrence of severe event that requires FFR.
 Signal – Communicate action to be taken.
 Activate – Actuate the resource.
 Activate fully – Full response from resource.
It must be mentioned that the frequency cannot be easily measured especially during fast frequency
deviations. Due to measurement time delay, filtering and signal processing, the fast frequency response
will not be available in the first moment like from synchronous generators (see also Clause 5.4.7).
In RES without some means of energy storage, e.g. PV systems and micro CHPs, there is no easily
accessible fast frequency response. However, the fast frequency response can be obtained from such
RES by adding the short-term energy storage, combined with suitable control of the power electronics
converter.

135
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 5.3 Recommended performance guidelines for inertial response profile in Hydro-Quebec [48]
Parameters Proportional function Step function
(Closed loop) (Open-loop)
Deadband ≤0.3Hz ≤0.5Hz
Active power contribution ≥6%
Duration of active power contribution ≥10 s
Activation time ≤1 s
Transition time for maximum generation reduction ≥3.5 s
Maximum generation reduction during recovery ≤20%

Figure 5.34 Inertial response profile [48].

Figure 5.35 Functional representation of a closed-loop inertia-based FFR control [9].

5.6.2.3 Limited frequency sensitivity model – over-frequency (LFSM-O)

The objective of this control is to reduce the active power output of the inverter when frequency rises
above a determined value to mitigate further increase (similar to droop control but for over-frequency
only). This procedure was discussed in Clause 5.4.6 to influence the modelling or not of the DC source
control. Different grid codes and guidelines propose alternative implementations [17].
These behaviours are easily modelled by limiting the active power current. For example, this can be
given by the following formula and shown in Figure 5.36:
I pHFRT  I pini + K HFRT  f - f start 

Equation 5.7
where:
Ipini is the active current delivered at fstart frequency
KHFRT is the active current rise factor. For example, a decrease of 40% of Ipini per Hz

136
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 5.36 Example of LFSM-O block diagram.

5.7 FUTURE TECHNICAL CHALLENGES OF IBG MODELLING

As previously described, various numerical models, i.e. mathematical simulation models have been
developed and integrated into commercial tools. The developed models are expected to be well
validated using measurements from real IBGs. Even if the detailed developed models are successfully
provided, parameter identification is another important task (see Chapter 7). The model parameters
such as those of the inverter controllers and those of the auxiliary controllers can be quite different
depending on the manufacturer and even on the type of IBGs released by the same manufacturer. It is
almost impossible to derive universal parameters from the diversified control model parameters for all
kinds of IBGs. Thus, deriving aggregated control model parameters remains a critical technical
challenge (see also Chapter 6).
The operation of IBGs following faults is related to the operation of protections for IBGs. In order to
examine the impact of IBG disconnection following faults, detailed protection models are important to
ensure the reliability of time-domain simulations. However, details on the measurement of key electric
variables, such as frequency, are not explicitly revealed mainly because information on the underlying
filtering techniques is confidential. Detection algorithms can also be quite different depending on the
manufacturer and even on the type of IBGs produced by the same manufacturer. Therefore, a proper
characterization of the detection logics also remains a critical technical challenge.
5.8 REMAINING TASKS
This chapter covered the fundamental aspects of RMS-type IBG models. The focus was on generic
inverter models and mainly on PVs. Future steps could include analysing the differences between
various inverter technologies, the influence of the primary source, or the existence of battery storage at
the DC side, on the IBG model. A unification of the various generator types (fuel cells, type-4 WTG, PV,
etc.) and technologies under a single generic model would facilitate the development and maintenance
of dynamic power system models.
Moreover, the emphasis of this chapter and the example models provided in the Appendix 5-C assume
3ph balanced operation (or single-phase system). However, this is not always the case, especially in
low-voltage distribution systems. The dynamic performance of 3ph and single-phase IBGs under
unbalanced system operation is very important, especially in distribution grids with increased
penetration of IBGs. Future steps could include analysing the impact of unbalanced operation on IBGs
and the required modifications to the existing models to properly capture this behaviour. In addition, the
appropriate protections, controls, and functionalities should be revised under this consideration.
Finally, another promising technique for modelling IBGs is with dynamic phasors [129]. This method
combines the computational performance of RMS models while allowing for higher accuracy and better
representation of unbalanced operating conditions and harmonics. While these types of models have
been studied for several years, the lack of commercially available simulation software hinders their
applicability. Such models could bridge the compromise between EMT and RMS concerning accuracy
and speed of simulation.

137
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

5.9 CONCLUSIONS
Chapter 5 investigates the methods for modelling IBGs under the RMS modelling assumption and
provides a number of example block diagrams and characteristic functions to achieve this. In addition
to the individual IBG model elements provided and described, some all-in-one models have also been
provided in the Appendix 5-C. The model elements consist of control and protection components,
respectively. Most of the IBG model elements presented in this chapter are currently implemented in
commercial time-domain simulation tools around the world. However, it should be noted that only some
capabilities/controls may be implemented in real IBGs. In addition, the specific capabilities and controls
implemented in the IBGs vary from country to country. Therefore, the numerical models, i.e.
mathematical simulation models can play an important role to figure out the recommended specification
of the capabilities from TSO/DSO’s point of view.
In this chapter, each IBG model element was classified into two categories: 1) Component level, and 2)
Plant level. The plant level elements were extracted from the controls which are generally required for
the utility-scale IBGs. It should be emphasized that the representative parameters or the representative
ranges of parameters of the all-in-one models, as well as the individual elements, were revealed as
much as possible.
Finally, this chapter has also discussed the limitations of RMS models depending on the phenomena
analysed and the required sampling frequency. If the RMS models are not suited to a specific type of
study, the EMT model can be one of the options. Chapter 4 details the way to model IBGs from EMT
modelling perspectives. Finally, the future technical challenge of IBG modelling has been briefly
discussed focusing on the establishment of the aggregation approach for distributed IBGs with
diversified control parameters.

138
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

6. MODELLING OF AGGREGATED DISTRIBUTED

INVERTER-BASED GENERATION
6.1 BACKGROUND AND MOTIVATION
It is well known that there are many individual wind turbine generators (WTG) in wind plants and many
individual solar PV units in solar PV plants. Because different renewables could show different dynamic
behaviour following faults, the individual modelling is an ideal solution for representing such dynamic
behaviour precisely. However, as the number of renewables increases, it is almost infeasible to model
huge number of renewables individually, mainly due to the very high computation burden. Therefore,
the aggregation techniques presented in this chapter to lump together all individual WTGs or PVs could
be one of the solutions to achieve a reasonable balance between the accuracy of the individual
behaviour and the permissible computation burden. The level of modelling detail that will ultimately be
required depends not only on the type of studies performed but also on the expected maximum
instantaneous wind/PVs penetration to be modelled.
Apart from reducing computational demands, another strong reason for using aggregated models is that
they are adequate for most of the studies required by utilities, i.e., grid planning and performance of the
system at the point of interconnection (POI) where the events to be considered are in the timescale from
tens of milliseconds to seconds. Also, in large power system dynamic studies, it may not be reasonable
to model each individual unit and individual low voltage feeder.
Overall, modelling of aggregated Inverter-Based Generators (IBG) is an ongoing field of research and
its importance is likely to increase with IBG deployment all over the world. The modelling of aggregated
IBG for large-scale power system dynamic studies is required for:
1) Steady-state power flow studies
2) Power system dynamic studies
The variance of the terminal voltage of IBGs connected in different locations of a feeder or the variance
between the medium voltage (MV) network and the low voltage (LV) network may cause a very different
dynamic response of these IBGs. The current practice of the modelling of the aggregated IBG is
classified into three categories in the order of complexity (from less detailed to more detailed modelling)
[7], [130], [131]:
 Joint aggregation of IBGs in MV and LV networks without considering the variation of the dynamic
behaviour of P and Q during and following faults (e.g., simple netting of IBGs with load),
 Joint aggregation of IBGs in MV and LV network (with) considering the variation of the dynamic
behaviour of P and Q during and following faults (WECC approach using PVD1 model, see below)
 Separate aggregation of IBGs in MV and in LV networks (with) considering the variation of the
dynamic behaviour of P and Q during and following faults (German academic approach, see below).
While the modelling of the aggregated IBG for the steady-state power flow analysis, the power flow
studies have already been performed and applied in some countries such as the U.S. and some
European countries. The industrial practice for the aggregated IBG modelling for 2), the power system
dynamic studies are very limited [6]. Proposals come mainly from academic research and are
summarized in [6] with recent contributions in [131], [132], [133].
Practical applications for aggregation include the generic wind generation model defined by IEC TC
61400-27-1 Ed.1 that is used for wind parks of significant size in power system dynamic studies in some
European countries. However, no generally accepted approaches for the modelling of the aggregated
IBG, such as medium-scale, distributed wind parks, and aggregation of small-scale PVs in low voltage
networks are currently available in some countries (contributions in [131], [132] are still of academic
nature). Conversely, validated user-defined models including IBGs at distribution level are used in power
system dynamic studies in other countries. The modelling of the aggregated IBG for the power flow
studies and the power system dynamic studies have been developed by WECC. The aggregation
approach of the collector system of large PV plants and the aggregation approach of residential PVs in
LV networks have been separately established. Because WECC has used the composite load model
including motor loads, a simplified, yet reasonable, aggregated model of distributed IBGs can be added
into the composite load model. On the other hand, the current modelling of this aggregated IBG used in
WECC still lacks certain stability study modelling details. Ongoing research is addressing the technical
challenges related to aggregation of the IBG’s control input parameters for power system dynamic
studies.

139
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

This Chapter describes 1) how to aggregate MV and/or LV network which include a group of RES, 2)
how to represent the equivalent of MV and/or LV network which include a group of RES. Chapter 6 also
provides recommendations for the derivation of the aggregated model parameters through measured
data.
Most importantly, the performance assumptions for distributed IBG in MV and/or LV networks can have
a large influence on the stability results of the bulk transmission level as penetration levels of distributed
IBGs increase. For example, varying the low voltage ride-through settings of distributed PVs can impact
the voltage at the bulk transmission level (See Figure 6.1).

(a) transmission bus voltage

(b) total active power output from distributed PV

(c) average bus frequency

Figure 6.1 Response to a three-phase fault with varying control parameters for PV in LV networks
[134].

140
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

It is noted that CA Rule 21 in the legend denotes a tariff (or set of regulations) that describes the
interconnection, operation and metering requirements for distributed generators that will be connected
to a utility’s electric system.
The methodologies described in Sub-Chapter 6.2 were developed by the Western Electricity
Coordinating Council (WECC) to represent wind and solar PV power plants in large-scale power flow
and power system dynamic studies for the use in the North America Western Interconnection [7], [135]
and have recently been adopted by the North American Electric Reliability Corporation (NERC) [134].
These WECC methods are currently one of the most elaborate standardised industry guidelines to study
the aggregated impact of distributed IBGs in large-scale power systems. These methods are
continuously evolving; in their current state, they are believed to create accurate modelling results for
winds/PVs penetration levels of up to approximately 50% of instantaneous interconnection-wide load.
A refined aggregation method, representing aggregated IBGs in further detail compared to the WECC
methodology, was used in a case study for assessing the dynamic stability in an example power system
resembling the German situation in the year 2022 [130]. Modelling recommendations based on this
refined aggregation method are added in the following sub-chapters where appropriate.
6.2 POWER FLOW REPRESENTATION
The aim of power flow modelling of wind/PV generation is to accurately represent distributed IBGs in
grid planning studies in order to determine network expansion and reactive power/voltage control
schemes coming from the reactive power range of inverters (See Appendix 3-G). A power flow case can
also be used to initialize the state variables of dynamic models for large-scale power system dynamic
studies.
Solar PV systems can be designed for a wide range of applications from small residential systems to
utility-scale, (sub-) transmission-connected power plants as described in the following clauses
(1) Distribution-connected residential and commercial PV systems
Distribution-connected residential (small-scale) and commercial (medium-scale) PV systems typically
connect to the customer side of the meter at single phase (e.g. 120/230V) or three-phase (e.g.
208/400V). Typical residential solar PV systems have a nameplate rating of less than 10 kW per phase
and have a single inverter, while commercial systems can reach a capacity of several MW and typically
have multiple inverters. Plants as large as 14 MW may be connected directly to an existing medium-
voltage primary distribution feeder, or to the unit station through a dedicated feeder.
(2) Large, utility-scale wind and solar PV plants
Modern large, utility-scale inverters used in wind and solar PV plants have nameplate rating ranging
from 1 MW to 4 MW. They connect at a terminal voltage of about 600 V or lower. Utility-scale PV and
wind plants may have hundreds of these units and the plant capacity of more than 100 MW. Furthermore,
they would be connected to the (sub-) transmission network.
6.2.1 Equivalent representation of utility-scale wind and solar PV plants
Utility scale generators are usually electrically separated from the transmission system by two
transformer stages and equivalent line impedance. In large and medium size wind and solar PV plants,
individual generating units are tied to a medium voltage (typically 12.5 kV to 34.5kV in the North America
and 10 kV to 36 kV in Europe [135]) collector system through step-up transformers and connected to
the transmission system at a single location, referred to as the POI. Several inverters may be connected
to a single pad-mounted transformer. The collector system consists of one or several feeders connected
together at a collector system station. One or more station transformers at the collector system station
are used to step-up to transmission system voltage. Unless the collector system station is adjacent to
the POI, an interconnection transmission line will be needed.
Reactive power support at the POI, to the extent that it is demanded by interconnection requirements,
can be provided by the inverters, dedicated plant-level reactive power support equipment such as fixed
or mechanically-switched capacitors, or a combination of both. STATCOM devices may also be installed
to provide dynamic reactive power support. The amount and nature (static or dynamic) of reactive power
compensation is driven by interconnection requirements and collector system design considerations,
including voltage profile and losses. Figure 6.2 shows a typical topology for a large, utility-scale solar
PV plant.

141
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 6.2 Example of utility-scale PV power plant topology [7].

In power flow studies (and in all positive-sequence transient stability studies), wind and solar PV plants
can be modelled with a single-machine equivalent; an explicit representation of the individual wind and
solar PV units is not necessary as long as their static and dynamic settings are similar. This equivalent
representation is shown in Figure 6.3.

Figure 6.3 Single-machine equivalent representation of an example solar PV plant [7].

The interconnection transmission line, station transformer(s) and plant-level collector station reactive
power compensation should be represented explicitly, according to established practice. The rest of the
components are equivalent representations, as follows:
 The equivalent generator and associated power factor correction capacitors represents the total
output of all generating units in the plant. For plants that were commissioned in multiple stages,
separate equivalent generators may be used in order to consider different static and dynamic
settings per changing interconnection requirements.
 The equivalent generator step-up transformer (pad-mounted transformer) represents the aggregate
effect of all step-up transformers.
 The equivalent collector system branch represents the aggregate effect of the collector system.
With the proper model parameters, this model should,
 Approximate plant power flow characteristics at the interconnection point,
 Collector system real and reactive power losses and voltage profile at the terminals of the “average”
inverter, in the plant.
As with any other models, the single machine equivalent representation has some limitations. Due to
collector system effects, terminal voltage of individual inverters could vary, especially in very large plants
where the electrical distance between inverters may be significant. Inverters that are closest to the POI
may experience significantly different terminal voltage compared to those which are electrically farther
away. In actual operation, terminal voltage of some inverters may reach control or protection limits,
resulting in different terminal behaviour (including partial tripping for power system dynamic studies).
Similarly, the collection network may have complicated harmonic characteristics, which requires due
consideration in deriving the equivalent collection circuit and the reactive power assessment.

142
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

During the design stage, or in special cases, it may be reasonable to use more detailed representation
of the collector system to capture these details. However, this type of modelling detail will not usually be
required for large-scale power system dynamic studies. Exceptions are study cases where a large
number of plants were commissioned in multiple stages for which different static and dynamic
settings/capabilities were implemented per the changing interconnection requirements at the
commissioning dates. In power flow studies, the evolving requirements for steady state voltage control
should be considered in order to accurately model reactive power flows and related losses. Validation
of the aggregated models by comparison with results obtained with detailed models is recommended in
order to assess the model inaccuracies which may be significant in certain cases [136].
Some examples of typical data values for overhead and underground cables in the power system are
presented in the Appendix 6-A-1, in order to facilitate computational simulation of the line connections
in the electrical network. The referred data includes nominal voltage, geometrical line configuration,
positive-sequence line impedance and susceptance, line rating and X/R ratio.
The following considerations refer to each of the components of the PV or wind plant single-machine
equivalent representation.
(a) Interconnection Transmission Line
Standard data includes nominal voltage, positive-sequence line parameters (impedance and charging)
and line rating (See Appendix 6-A).
(b) Plant Station Transformer
Transmission-connected PV plants or wind farms require a station transformer that should be
represented explicitly. Some plants may have several station transformers. Standard data includes
transformer nominal voltage of each winding, impedance, tap ratios, regulated bus and set point, and
ratings. Positive-sequence impedance for station transformers is in the range of 6% to 13%, and X/R
ratio in the range of 20 to 50 in North America (accordingly ANSI / IEEE C37.010) and up to 120 in
Europe (accordingly the Eco design Directive from the European Commission - Regulation (EU) n°
548/2014 to achieve loss minimization) [137].
(c) Plant Level Reactive Power Compensation
PV or wind plants could have station fixed and/or switched shunt capacitors installed at collector system
as well as reactive power compensation at the Point of Interconnection. If present, the shunt capacitors
should be modelled as constant impedance devices in power flow studies, to capture voltage-squared
effects, with each switched capacitor modelled explicitly. Standard data includes nominal rating,
impedance, and controlled device, if applicable. Operation of the shunt devices is coordinated with the
plant-level reactive power controller. Plants may also have dynamic reactive power compensation
devices such as STATCOMs or SVC (static var compensators).
(d) Equivalent Collector System
Central-station plant collector systems consist of one or more medium voltage feeders which may be
relatively long. Factors to be considered in feeder design include cost, active power losses, and voltage
performance. A typical design goal is to keep average active power losses below 1-2%. At full output,
active power losses can be higher, as much as 2% to 5%. The collector system network is typically
underground. For that reason, the equivalent collector system X/R ratio tends to be low compared to
typical overhead circuits. The equivalent collector system impedance tends to be small compared to the
station transformer impedance, but it still cannot be ignored.
A simple method developed by NREL [138] can be used to derive equivalent impedance (Zeq) and
equivalent susceptance (Beq) of a collector system consisting of radial elements. This method creates
accurate results as long as static and dynamic settings of all inverters in a plant can be treated as similar;
in special cases where advancing requirements for steady state voltage control should be considered,
a more elaborate method to determine the equivalent collector impedance may be needed in order to
accurately model reactive power flows.
The computation of the equivalent collector impedance is as follows:
I

Z n 2
i i
Z eq  Req + jX eq  i 1
N2
Equation 6.1

143
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

I
Beq   Bi
i 1

Equation 6.2
where I is total number of branches in the collector system, Zi is the impedance (Ri + jXi) for i-th branch,
ni is the number of inverters below the node i, and N is the total number of inverters in the PV plant or
wind generators in the wind plant. Branch impedance data can be obtained from collector system design
(conductor schedule) for the project. As stated before, the equivalent impedance computed in this
manner approximates real and reactive power losses seen by the “average inverter” in the PV plant or
“average generator” in the wind plant. This calculation can be easily implemented in a spreadsheet.
Figure 6.4 shows a simple example with nine branches (I = 9), and 21 inverters (N = 21). The
corresponding calculations are shown in Table 6.1. In this example, the inverters are 7 clusters of 3
inverters. In general, larger power plants would have lower Zeq and higher Beq considering that more
parallel feeders would be required.

Note: 1MW, 1.1 MVA, +/-0.95 p.f. are rated values of PV inverter.
Figure 6.4 Sample utility-scale PV plant topology [137].

144
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 6.1 Computation of collector system’s equivalent parameters for sample system in Figure 6.4.

Table 6.2 Sample equivalent collector system parameters [139]

Plant size Collector voltage Feeder R (p.u.) X (p.u.) B (p.u.)
100 MW 34.5 kV All underground 0.017 0.014 0.030
100 MW 34.5 kV 33% overhead (Carrying 100% of WTG*) 0.018 0.079 0.030
110 MW 34.5 kV All underground 0.012 0.011 0.036
200 MW 34.5 kV Some overhead 0.007 0.025 0.055
200 MW 34.5 kV 33% overhead (Carrying 100% of WTG) 0.010 0.039 0.099
300 MW 34.5 kV Some overhead 0.005 0.020 0.085
300 MW 34.5 kV Some overhead 0.006 0.026 0.150
Note: per unit parameters are on a 100 MVA base and collector system kV base.
*: % of carrying WTG denotes what percentage of the WTGs (or PVs) are connected to the feeder.

In this table, the parameters are derived from Equations (6-1) and (6-2). For example, the branch from
node 1 to node 4 has 3 inverters connected to the nodes 1 through 4, as shown in Figure 6.4, therefore
n=3. The same refers to the branches between nodes 2 and 4, or 3 and 5. The branch between nodes
4 and 5 has 9 inverters downstream, thus for this branch n=9. The branch between node 5 and the
substation has 12 inverters downstream, because all the inverters connected to the nodes 1, 2, 3 and 4
count for this branch. “Partial R sum” and “Partial X sum” in the Table 6.1 should be replaced with “Rn2
total sum” and “Xn2 total sum”, respectively. The partial results of calculations are shown for collectors
1 and 2 in Table 6.1. The sum of all Rn2 is 9.4788 and of all Xn2 is 6.7666. Thus, equivalent resistance
and reactance are expressed as Req = 9.4788/212= 0.021494 p.u. and Xeq = 6.7666/212 = 0.015344 p.u.,
since the total number of inverters in Figure 6.3 is N = 21. The equivalent susceptance is the sum of the
susceptances of each branch, yielding Beq = 0.000005 p.u. Table 6.2 shows some examples of the
equivalent collector system parameters for several plants of the different nameplate capacity and
different collector system.
Because the overhead lines and underground cables have different parameters, the collector system
parameters are affected by such fraction of the element of the feeder. Percentage of wind generators
connected to the feeder can affect the flow on the feeder. In addition, WTGs may have different power
factor than, for example, residential load, which can also affect the flow on the feeder.
(e) Equivalent Plant Step-Up Transformer
A large PV or wind plant has several pad-mounted transformers, each connected to one or more PV
inverters or wind generators. Assuming that all step-up transformers are identical, and each connects
to the same number of inverters, the per-unit equivalent impedance (ZTeq) and the equivalent’s apparent
power rating (STeq) can be computed as follows:
M
STeq   STk
k 1

Equation 6.3

145
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

STeq
ZTeq  M
STk

k 1 ZTk

Equation 6.4
where M is the total number of step-up transformers in the collector system, STk is the apparent power
rating of k-th step-up transformer in MVA, and ZTk is the per-unit impedance of k-th step-up transformer
on its own MVA base.
In the particular case where all step-up transformers are identical, and as such STk=ST and ZTk=ZT, the
assumed equations, Equation 6.3 and Equation 6.4 can be simplified to the following equations Equation
6.5 and Equation 6.6.

STeq  M  ST
Equation 6.5

ZTeq  ZT
Equation 6.6
For the example system discussed above where all step-up transformers are identical, the equivalent
transformer impedance would be 6% on a 24.5 MVA base (7×3.5 MVA), with an X/R ratio of 10.
Step-up transformers associated with utility-scale PV plants are in the range of 500 kVA to 2 MVA, and
have impedance of approximately 6% on the transformer MVA base, with X/R ratio of about 8.
(f) Equivalent Generator Representation
For power flow simulations, the aggregated PV or wind generator should be represented as a standard
static generator, as opposed to a negative load, so that various IBG controls can be implemented for
dynamic simulations. Active power level and reactive power capability must be specified as described
in Appendix 3-G.
Representation of reactive power capability of the equivalent inverter depends on the reactive power
range of the inverters, and how that range is utilised in operations. For example, the equivalent generator
for the sample system shown in Figure 6.4 would have a nameplate rating of 21 MW (21 × 1 MW) and
23.1 MVA (21 × 1.1 MVA). If the inverters participate in steady-state voltage control, then the equivalent
generator should be modelled with a reactive range of +/- 0.95 power factor, which corresponds to
setting Qmin = -6.9 Mvar and Qmax = +6.9 Mvar, respectively. If the inverters operate at a fixed power
factor, then the equivalent generator should have Qmin = Qmax at the corresponding power factor level
(considering the correct sign in terms of leading or lagging operation). At an output level below rated,
the reactive power limits should be adjusted according to the inverter reactive power output that is
programmed into the controls and adjusted by a power factor correction that considers the effect of the
terminal voltage of the inverters.
6.2.2 Equivalent representation of distribution-connected PV
In large-scale power system dynamic studies for systems with low wind/PVs penetration, the distribution
system does not have to be modelled in detail and the load can be aggregated at the transmission buses
with embedded distributed generation netted as negative load. However, in large-scale power system
dynamic studies for which the aggregate distributed generation has the potential to affect grid reliability
and compliance with system planning and performance standards (e.g., [140]), distribution-connected
PV systems should be represented explicitly and not netted with load.
Modelling distributed generation explicitly, yet in an aggregated way, allows for proper load scaling and
gives planners the ability to account for existing and emerging (changing) interconnection requirements
and performance standards applicable to distributed generation. These may include requirements for
voltage and frequency tolerance/ride-through capability, static and dynamic reactive power support, anti-
islanding, etc.
A modular approach to represent IBGs in bulk system studies as illustrated in Figure 6.5 is
recommended [141] to ensure accurate representation of the resources for the specific bulk system

146
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

study type. The hierarchy of the clustering of IBGs for model aggregation should consider differentiation
of the IBGs per [141]:
 Resource type in order to derive meaningful dispatch scenarios rather than worst-case dispatches
for bulk system planning studies.
 Interconnection requirements performance in order to represent the fundamentally different steady-
state and dynamic behaviour among the legacy distributed generators.
 Technology-type, e.g., inverter-coupled versus directly-coupled synchronous generator, in order to
accurately represent the technology-specific dynamic behaviour.

Figure 6.5 Composite load model with distributed behind the meter generation [111].
Industry practice in North America ([140], [142]) recommends that any multiple smaller generation
facilities connected to an equivalent distribution feeder (See Figure 6.5) with an aggregated generation
capacity of 20 MVA or more should be modelled as aggregated units and not be netted with load. In
other words, the aggregated generation of no greater than 20 MVA may be represented as the negative
load. It should be noted that the threshold of 20 MVA is an arbitrary choice based on an engineering
judgment that plants of such size or larger may noticeably impact the performance of the transmission
system and can therefore not be ignored.
As outlined in the previous clause, large commercial-scale PV systems should be represented with a
discrete lumped model with an equivalent LTC transformer and a single equivalent series impedance
representing the impedance of the feeder, station transformer, and secondary network. A similar
approach should be used for high penetration residential PV systems. However, this approach may
reach its limitations for the following study cases:
 High winds/PVs penetration levels (e.g. above approximately 50%) of instantaneous
interconnection-wide load, i.e. kW or MW or GW loads)
 A significant amount of reverse power flows from distribution grid to transmission grid
 Substantial amounts of (distributed) generation connected at different voltage levels in a region.

In those special cases, a refined representation of the distribution system is recommended considering
the multiple equivalent impedances of HV-sub-transmission lines as well as MV-primary and LV-
secondary feeders separately. Wind/PVs connected to a certain voltage level can then be aggregated
into an equivalent generator.
One of the goals of the aggregation approach should be to capture the effect of reactive power support,
as well as the voltage tolerance characteristics of PV systems in steady-state simulations. In particular
voltage stability and the dynamic stability simulations, the goal is to represent dynamic performance
during and following abnormal conditions. Depending on the characteristics of the distribution systems
and their level of uniformity in the study case, different equivalent impedances may be used for urban,
sub-urban and rural feeders to accurately model the voltage at the POI for WPP or PV plants with the
distribution system.

147
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

In regions where interconnection requirements for distributed generation have been changed or are
expected to change substantially in the future, separate equivalent generators may be used for each
WT or PV technology generation in order to comply with their modelling parameters. This is important
when considering different static and dynamic settings/capabilities for voltage/frequency tolerance/ride-
through and reactive power standards. Existing wind/PV plants are typically not upgraded to meet the
latest interconnection requirements and may need to be considered separately (especially in power
system dynamic studies).
Finally, the proper representation of load, especially its voltage dependency, is very important [6].
Suitable load models should be used while considering that it is often difficult to obtain the required data
and to validate the load model.
Behind the meter distributed solar PV generation may be modelled as part of load. For dynamic stability
simulations, composite load model with distributed generation can be used. This model is used in WECC
for dynamic stability studies of the systems with high penetration of renewable resources, including
distributed solar PV (roof-top solar panels). It is represented in Figure 6.5.
This model includes substation transformer, distribution feeder, shunt capacitors, several types of
induction motor load (three-phase and single phase), electronic and static load and distributed
generation. Utility scale generation is modelled as aggregated generators on low voltage side of
transmission transformers, and retail scale generation (roof-top PV) is modelled as part of composite
load model. It is also aggregated by distribution feeders. The composite load model for dynamic stability
studies that includes distributed generators has the generation part substantially simplified. Currently,
only solar PV can be modelled as a part of the composite load.
6.2.2.1 Single equivalent distribution impedance approach with non-changing interconnection
requirements
The recommended power flow representation for medium penetration distribution-connected PV plants
with a single equivalent impedance and non-changing interconnection requirements is shown in Figure
6.6. Note that the load was moved to the low voltage bus as well. The term medium penetration level
denotes “a level of penetration of renewables where changes are required to either control strategies or
capabilities, or both, to manage the system, but where renewables are still seen as a relatively small
part of the overall energy portfolio on the electricity system.” (See CIGRE TB 527 [143])

Typical load flow model Medium-penetration PV on distribution system Recommended load flow model

Transmission Transmission Transmission

system system system

Unit station Unit station

transformer transformer

Load Distribution
system

Utility-scale PV Equivalent
Impedance

Pad/Pole
transformer
Residential
PV Load

Commertial

Figure 6.6 Recommended power flow representation for study of medium-penetration PV scenarios
with single equivalent distribution impedance [7].

Typical power flow data for the single equivalent distribution feeder is shown for North America in Table
6.3 below.

148
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 6.3 Suggested data for single distribution network equivalent [139]
R, p.u. X, p.u.
Station transformer impedance, p.u. on the transformers air-cooled MVA base 0 0.1
Equivalent feeder, service transformer and secondary impedance, p.u. on 100 MVA,
0.1 0.1
12.5 kV base*
* For substations with aggregate load larger than the 100 MVA system base, the equivalent impedance shall
be divided by the scaling factor in order to appropriately initialise the system variables in the power flow
calculation.

Transformer impedances are on the air-cooled transformer MVA base, which should be appropriate for
the amount of load served depending on the applicable regional distribution planning criteria. Distribution
transformer overrating and a redundant number of distribution transformers should also be considered.
The feeder impedance is on a 100 MVA base and the collector system kV base (12.5 kV to 34.5 kV for
North American grids and 10 kV to 36 kV for European grids). The feeder impedance data should be
adjusted, depending on load level, to obtain a reasonable voltage drop from the station transformer
secondary to the typical utilization point, or approximately 3%. This percentage value is recommended
by WECC based on industry experience [7].
Examples of modelling of such aggregated solar PV plants and voltage drop depending on the station
transformer and feeder models are shown in Table 6.4. Load was modelled with 0.95 power factor,
generation was modelled with 0.95 lead/lag power factor regulating voltage at 1.0 per unit at the
generator terminals. The voltage drop is shown between the lower voltage bus of the station transformer
and the Solar PV/Load bus. The voltage drop in Table 6.4 was calculated such as to have the voltage
at the generator terminals at 1.0 per unit.

Table 6.4 Example of modelling equivalent solar PV on distribution network

Load Solar PV Equivalent Feeder Station Transformer Voltage drop
MW MW R p.u. X p.u. MVA R p.u. X p.u. %
10 5 0.1 0.1 10 0 0.1 0.7
10 5 0.2 0.2 10 0 0.1 1.3
25 5 0.1 0.1 25 0 0.1 2.9
10 20 0.1 0.1 10 0 0.1 0.7
20 5 0.1 0.1 20 0 0.1 1.9
20 5 0.2 0.2 20 0 0.1 4.2
20 5 0.1 0.2 20 0 0.1 2.5
20 10 0.1 0.1 20 0 0.1 1.3

6.2.2.2 Multiple equivalent distribution impedance approach with changing interconnection

requirements
The recommended power flow representation for medium penetration distribution-connected PV plants
with multiple equivalent distribution impedances and changing interconnection requirements is shown
in Figure 6.7 based on [130], [131]. Both the distribution system load as well as the various distributed
generation technologies (load type) are split across the distribution voltage levels (as far as this
information is available). Discrete modelling of synchronous generation-based technologies and
inverter-based generation technologies is recommended. Wind and PV plants may be modelled in power
flow as a combined equivalent generator if the combined dispatch is handled with care (wind and solar
generation may have different production patterns that are determined by the meteorological conditions)
[10]. Discrete equivalent generators should be modelled for each group of distributed generators that
were commissioned in a period during which certain interconnection requirements were actively
enforced [130], [131].
As an example of power flow data for the multiple equivalent distribution feeders is shown for the
German situation in Table 6.5 below. Here transformer impedances are again calculated on the
transformer self-cooled MVA base, which should be appropriate for the amount of load served. This is
dependent on the applicable regional distribution planning criteria, distribution transformer overrating
and a redundant number of distribution transformers. The feeder impedance is on a 100 MVA and 0.4
kV and 20 kV base for European grids, as was the case in Table 6.3.

149
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 6.5 Suggested data for multiple distribution system equivalents for typical German active
distribution systems (Data provided here is simply an example, more detail discussion required for
specific systems) [6], [131], [137]

Urban Sub-urban Rural

Base for R, X,
p.u. values* p.u. p..u R, p.u. X, p.u. R, p.u. X, p.u.
Station transformer impedance Transformer
self-cooled
0.0 0.16 0.0 0.10 0.0 0.08
MVA base

MV-primary equivalent feeder 100 MVA,

0.00625 0.00625 0.03325 0.03325
20 kV
LV-secondary equivalent feeder* 100 MVA,
5.18750 0.81250 8.18750 0.87500
0.4 kV
Service transformer and secondary 100 MVA,
impedance* 0.4 kV
* For substations with aggregate load larger than the 100 MVA system base, the equivalent impedance shall be divided
by the scaling factor in order to appropriately initialise the system variables in the power flow calculation.
** It is recommended not to explicitly model these impedances in positive-sequence, bulk system stability studies in
order to prevent numerical issues in the power flow initialisation; they can be accounted for by utilizing the appropriate
voltage trip parameters in the standard WECC PVD1 model. However, for detailed distribution analysis (EMT-type)
especially for weak grids (a.k.a. low SCR) they may play a significant role.

eHV eHV

HV
HV HV
HV

Load
PF100
aRCI NEW
PF100

HV HV

Load

LVRT PF100 LVRT PF100 aRCI

MV

MV
Load NEW PF095 NEW LVRT NEW

CHP / Bio
PF100 PF100

Wind

LV Photovoltaic

Class specifies
the fault behviour
LV
Load NEW PFPOW NEW Class

Figure 6.7 Recommended power flow representation for study of medium-penetration PV scenarios
with multiple equivalent distribution impedance and changing interconnection requirements [130],
[131].

150
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The feeder impedance data should be adjusted depending on the load level for normal power flow cases
and distributed generation feeder penetration for reverse power flow cases so as to obtain a reasonable
voltage drop between the station transformer secondary to the typical utilization point, or approximately
2–3%.
6.2.3 Essential assumptions and control to be modelled
6.2.3.1 Active power output level
PV plant output varies as a function of solar input and, to a lesser extent, temperature. Wind power plant
output varies depending on the season and time of the day. Typically, PV plants achieve full output for
several hours of the day under clear sky conditions. The active power level assumed for the PV or wind
plant depends on the purpose of the study with the plants modelled at full output for the generation
interconnection studies and at partial or zero output for regional transmission planning studies
depending on the season and the time of the day.
For wind or PV plants that use an active power-dependent reactive power control mode, at least those
two active power levels that correspond to the extreme reactive power exchange cases, e.g., zero and
maximum reactive power exchange should be modelled. Otherwise, the full impact of wind and solar PV
plants on power flows and dynamic behaviour will not be adequately examined.
6.2.3.2 Reactive power capability
Interconnection requirements and performance standards addressing reactive power capability from
inverter-based generating systems are still evolving. Inverters used in utility-scale PV systems are
currently required and designed to provide a certain minimum reactive power support at full active power
to the grid (typically with a power factor of 0.95 leading and 0.95 lagging). The amount of reactive power
at partial output is generally even higher and depends on the inverter current limits and on grid voltage
conditions. The reactive power capability curves for inverters differ from those of synchronous machines
because they are normally limited by internal voltage and current constraints (See Appendix 5-C).
Inverters are typically designed for continuous operation between 90% and 110% of nominal terminal
voltage. Depending on the interconnection requirements and the inverter design, the full reactive power
capability can only be achieved within a smaller voltage band around nominal voltage. In special cases,
PV inverters are designed to provide reactive power support even if solar irradiation is zero; such
STATCOM functionality should then be considered in the study.
Distribution-connected PV systems have traditionally been operated at unity power factor over their
entire active power output range, i.e., with zero reactive power exchange to the grid. Hence, many
existing inverters used in residential and commercial areas were not designed to inject any reactive
current. However, new interconnection requirements for distribution-connected PV systems are
increasingly requiring reactive power capability from small-scale plants because this reduces distribution
grid upgrades through the mitigation of overvoltage in LV-secondary feeders when many solar PV
systems are connected to the same feeder. For studies with a mid-term and long-term planning horizon,
the reactive power capability and controls of distribution-connected PV systems should be modelled.
At the plant level of utility-scale PV plants, a portion of the inverters in the plant may be turned off at low
plant power output (via the valve device blocking), resulting in a reduction of reactive power capability.
Therefore, reactive power output for an equivalent aggregated generator modelled in power flow studies
should be modified accordingly.
To meet operating requirements at the point of interconnection, the settings of individual inverters can
be adjusted via a plant-level reactive power controller. The plant level controller also coordinates
operation of the switched capacitors, if present.
Several control modes are required (refer to [23] and [144] for further details):
 Closed-loop voltage control - Maintain voltage reference at the scheduled value over a certain range
of active power output while considering the reactive power capability of the PV plant.
 Open-loop voltage (droop) control – Increase or decrease reactive power output linearly as a
function of voltage. This type of control allows the PV plant to provide voltage support while avoiding
large reactive power swings that a small PV plant would see when connected to a relatively stiff grid.
 Fixed power factor control - Maintain a fixed power factor at the interconnection point close to a
specified value.
 Active power-dependent power factor control [144] – Set the power factor value at the
interconnection point according to an active power-dependent characteristic curve. Typically, unity

151
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

power factor is used for low to medium active power output. For medium to high active power output,
the power factor is often linearly increased to the plant’s maximum inductive capability.
 Reactive power control - Maintain reactive power flow at the interconnection point within some
specified limits.
6.3 SIMPLIFIED DYNAMIC MODEL
Even if the aggregated model is perfect for power flow studies, it will probably not be sufficient to
represent the dynamic behaviour of IBGs following faults. One critical point that needs to be considered
is the self-disconnection of IBGs. The self-disconnection of IBGs is most likely to occur when IBGs do
not meet the LVRT requirement of mainly LV networks. The total generation disconnected on secondary
side of one HV-MV (or MV-HV) transformer will depend on how much the voltage decreases. The self-
disconnection of IBGs is more likely to increase as the system voltage decreases. Sub-Chapter 6.3
introduces the model which can represent the dynamic behaviour following faults in an aggregated way
considering the self-disconnection of IBGs.
6.3.1 Model for distributed and small PV
Unlike utility-scale PV plants, i.e. the central station PV plants in Clause 6.3.1, distributed PV systems
are connected at the distribution level. Reliability and interconnection requirements vary from country to
country.
For example, in North America, the requirements outlined in IEEE Standard 1547-2003 [18] are reflected.
In contrast with bulk system utility-scale PV plants reliability requirements, distributed PV systems in
North America do not participate in steady state voltage regulation, and tighter bounds on operation for
off-nominal voltage and frequency conditions result in significantly different fault ride-through capability
[111]. IEEE Standard. 1547 has been amended in 2014 and now allows for both steady state voltage
regulation and extended frequency and voltage ride-through. However, binding and standardised
requirements are not expected to enter into force until sometime in the future years. Hence, it can be
assumed in near-term studies that the PV inverters connected in distribution systems will continue to
comply with IEEE Standard 1547-2003 [18], and will operate under constant power factor or constant
reactive power modes of operation. This allows for the (temporary) elimination of the closed-loop voltage
regulator dynamics, along with the elimination of the DC dynamics (for the same reasons described for
the Central Station model), making for a substantial simplification of the model with respect to that of
the Central Station. However, unlike a utility-scale PV plants, the terminal voltages seen by the individual
inverters within the composite load in the large-scale power system dynamic model are likely to vary
substantially. A different control model should be used to capture the effect of the diverse terminal
conditions on the aggregate generation [111].
In Europe, and particularly in Germany, distributed PV systems are required to provide steady state
voltage regulation since the year 2011 [144]. This should be adequately modelled and may increase the
model complexity. However, voltage and frequency ride-through is currently mandated only for wind/PVs
connected to medium voltage (primary) distribution feeders but not for those connected to low voltage
(secondary) distribution feeders. Ride-through requirements are likely to be introduced for LV-secondary
connected distributed generation in the near future [130] as penetration is reaching a level where the
aggregated impact of these small-scale, residential plants cannot be ignored in large-scale power
system dynamic studies any longer.
In some countries, such as Japan, voltage and frequency ride-through has been mandated for almost
all types of IBGs regardless of the voltage level. On the other hand, the anti-islanding protections have
been required for IBGs which are connected to LV network only [145]. Therefore, the IBGs in MV network
and LV network need to be separately aggregated for the modelling.
Generally, the minimum level of modelling detail should allow for basic reactive power control modes,
such as:
 Constant reactive power, initialized at the generator output in the solved power flow case,
 Volt/var control at the generator terminals, with user-defined Q versus V characteristic and optional
line drop compensation, if any.
The model should also allow for basic active power control modes, such as:
 Constant active power, initialized at the generator output in the solved power flow case,
 Over-frequency response, with user-defined dead band and droop.

152
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

6.3.1.1 WECC distributed PV system model (PVD1)

The model for distribution-connected and small solar PV plants used in WECC (which is called PVD1)
is much simpler than the models used for large plants previously described. This model also requires
the plants to be modelled explicitly as generators in power flow calculations. PVD1 model is a simple
current injector with basic functionality for reactive power control, active power control and protection.
The block diagram of this model is shown in Figure 6.8.

Figure 6.8 Block diagram of WECC small PV plant model [111].

Model parameters are expressed in per unit on the generators air-cooled MVA base. To scale the
dynamic model to the size of the aggregated PV generation, the generator MVA base parameter must
be adjusted. To represent, for example, unity power factor, Qmax and Qmin should be set to zero both in
the power flow case and in the dynamic model. If the reactive power control is disabled, the reactive
power range (Qmax and Qmin) needs to be set to zero.
If the frequency response is disabled, the droop gain parameter (Ddn) needs to be set to zero; thus, the
plant operates in constant active power mode. Similar to the large-scale PV plant model, the user can
specify whether active or reactive power takes precedence, by setting the Pqflag parameter. Tables 6.6
and 6.7 specify PVD1 input parameters and sample settings and its internal variables, respectively. In
Table 6.6 the two starred parameters represent the low voltage disconnection response of the PV. The
values provided here are standard parameters used by the WECC. If detailed representations of the
distribution network are available for simulation, these values can be refined to represent the
characteristic response of PV disconnection for individual feeders. This requires additional simulations
with a detailed model of the distribution network.
The model allows for (partial) tripping of the generation based on over- and under-voltage and frequency
settings monitored at the equivalent generator terminals. The user can set the voltage and frequency
dead bands, how much generation trips, and what fraction of the generation is restored as the
disturbance subsides. Partial tripping is relevant when the equivalent generator represents distributed
generation systems, each of which would experience a transmission-level disturbance very differently
depending on the electrical distance to the bus where they are aggregated. Partial tripping (caused by
either the disconnection of IBG or the valve device blocking) also allows for consideration of the evolving
interconnection requirements that might require tripping in the past and present but that might require a
ride-through in the mid-term future. In this model, trip thresholds are not time-dependent. If this
functionality is desired, a standard generator protection model such as overvoltage and undervoltage
protection models can be used. In Table 6.6, a standard set of parameters for the PVD1 model are

153
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

provided. Here Vt0 and Vt1, the low voltage trip parameters, are given as 0.88 p.u. and 0.90 p.u. To
compliment these, an example for a distribution feeder from California, USA is provided in brackets
alongside the values in Table 6.6.

Table 6.6 Input parameters and sample settings of PVD1 [111]

PVD1 Input Parameters and Sample Settings
Name Description Typical Values
Pqflag Priority to reactive current (0) or active current (1) -
Xc Line drop compensation reactance (p.u. on mbase) 0
Qmx Maximum reactive power command (p.u. on mbase) 0.328
Qmn Minimum reactive power command (p.u. on mbase) -0.328
V0 Lower limit of deadband for voltage droop response (p.u.) -
V1 Upper limit of deadband for voltage droop response (p.u.) -
Dqdv Voltage droop response characteristic -
fdbd Overfrequency deadband for governor response (p.u. deviation) -
Ddn Down regulation droop gain (p.u. on mbase) -
Imax Apparent current limit (p.u. on mbase) 1.0 to 1.3
Vt0 Voltage tripping response curve point 0 (p.u.) 0.88 (0.85*)
Vt1 Voltage tripping response curve point 1 (p.u.) 0.90 (0.94*)
Vt2 Voltage tripping response curve point 2 (p.u.) 1.1
Vt3 Voltage tripping response curve point 3 (p.u.) 1.2
vrrecov Voltage tripping is latching (0) or partially self-resetting (>0 and ≤1) 0
Ft0 Frequency tripping response curve point 0 (Hz) 59.5
Ft1 Frequency tripping response curve point 1 (Hz) 59.7
Ft2 Frequency tripping response curve point 2 (Hz) 60.3
Ft3 Frequency tripping response curve point 3 (Hz) 60.5
frrecov Frequency tripping is latching (0) or partially self-resetting (>0 and ≤1) 0
Tg Inverter current lag time constant (s) 0.02
*Values for California example feeder [146].

Table 6.7 Internal variables of PVD1 [111]

PVD1 Internal Variables
Name Description
Vt Terminal voltage (p.u., from network solution)
It Terminal current (p.u., from network solution)
Pref Initial active power (p.u. on mbase, from power flow solution)
Pext Supplemental active power signal (p.u. on mbase; zero unless written to by external model)
Pdrp Governor response (droop) power (p.u. on mbase)
Qref Initial reactive power (p.u. on mbase, from power flow solution)
Freq Terminal frequency deviation (p.u., from network solution)
Freq_ref Initial terminal frequency deviation (0)
Fvl Multiplier on current commands in high voltage condition
Fvh Multiplier on current commands in low voltage condition
Ffl Multiplier on current commands in high frequency condition
Ffh Multiplier on current commands in low frequency condition
Ipmax Dynamic active current limit (p.u. on mbase)
Iqmax Dynamic reactive current limit (p.u. on mbase)
Iqmin Dynamic reactive current limit (p.u. on mbase, = -Iqmax)
Ipcmd Active current command (p.u. on mbase)
Iqcmd Reactive current command (p.u. on mbase)
Ip Active terminal current (p.u. on mbase)
Iq Reactive terminal current (p.u. on mbase)
Pgen Electrical power (MW)
Qgen Reactive Power (Mvar)

154
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

6.4 CONCLUSION
Correct modelling the aggregated response of distributed IBG will influence the stability results at the
bulk transmission system level and become more critical as penetration levels of distributed PVs
increase. A modular approach to represent IBGs in bulk system studies as illustrated in Figure 6.7 is
recommended to ensure accurate representation of the resources for the specific bulk system study
type. The hierarchy of the clustering of IBGs for model aggregation should consider differentiation of
IBGs per:
 Resource type, to derive meaningful dispatch scenarios rather than worst-case dispatches.
 Interconnection requirements performance, to represent the different steady-state and dynamic
behaviour among the legacy distributed generators.
 Technology-type, e.g., inverter-coupled versus directly-coupled synchronous generator, to
accurately represent the technology-specific dynamic behaviour.
Dynamic models for distributed IBGs are available to model the evolving interconnection requirements
related performance requirements. WECC’s simplified distributed PV model (PVD1) currently seems to
be the most promising concept to reach a reasonable balance between modelling accuracy,
computational requirements, and handling of the system model, but some further improvement may be
needed. In addition, there is other remaining work. For example, the way how to aggregate tremendous
number of residential PVs in LV network from HV network perspectives is still under research (See also
Sub-Chapter 5.7). Also, there is the work to introduce modular approach (also known as bottom-up
approach) to the modelling of IBG, especially the units connected to distribution systems and to develop
more accurate aggregated models [147].
A refined representation of the distribution system is recommended by considering the multiple
equivalent impedances of HV-sub-transmission lines as well as MV-primary and LV-secondary feeders
separately if any of the following conditions apply:
 High WTs/PVs penetration levels (e.g. above approximately 50%) of instantaneous interconnection-
wide load, i.e. kW or MW or GW loads)
 A significant amount of reverse power flows from distribution to bulk system level
 Substantial amounts of (distributed) generation connected at different voltage levels in a region.
In those special cases, WT/PVs connected to a certain voltage level would then be aggregated into an
equivalent generator.

155
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

156
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

7. VALIDATION OF INVERTER-BASED GENERATOR

MODEL
7.1 INTRODUCTION
As a final step in the process of model development the model validation is to
 confirm the assumptions made in the model creation
 proof that the model performs in the way it is designed to
 confirm the conditions where the model should be used, and
 identify any limitations that the model may have.
Model validation is a very important part in the model development. The users of any model need to
understand the model’s characteristics, its limitations and have confidence in it. The developers need
to ensure its fidelity before finally issuing it for use. A model can be developed by the engineer who will
use it later in his/her simulation studies. In this case, the details of the model validation can be
understood and followed. In most cases though, a model is developed by third parties or exists in a
commercial software suite. In this case, validation by the user is even more important.
Usually the models used by the PV manufacturers are very different from those used by the power
companies. The manufacturers’ models are usually built at the individual cell level with detailed control
and protection logic circuits. This type of model is complicated and contains intellectual property from
manufacturer, and is usually for a specific project. Models of this type are used in EMT studies for
inverter design to assess the transient stresses on the devices or for designing control and protection
function at a microsecond scale. As the model has very small time constants the model is not suitable
for power system dynamic studies in a large power grid.
Because the power grid under study is in general very large, the models of IBGs used for studies in a
large power grid can be simplified. Individual inverters will be lumped together to model the entire power
plant. By doing so, the model has a large time constant and can be executed efficiently with the power
grid model in terms of numerical stability and computation time in simulation.
The focus of this simplified model is on the voltage and current at the connection terminals to the grid
and its P and Q regulation behaviour in response to the control commands or a change in the grid. It is
a RMS model, representing the inverter outputs and control behaviour to the positive sequence
components in the AC system, for studying balanced events from either normal variation or transient
faults. These models are designed for power system dynamic studies like system stability, voltage
regulations or frequency control as described in previous chapters.
There is no difference between the RMS model validation approach and the EMT model validation
approach. However, the type of disturbance or the type of phenomenon (which should be observed) can
be quite different depending on type of model (RMS or EMT). This chapter focuses only on the RMS
model validation.
IEC 61400-27-2 (though mainly for wind plants) presents some useful methods for the model validation.
The validation is a process of comparing the simulation results produced on the model with the data
obtained independently by other means. In practice, obtaining a set of good data for model validation
can be difficult and expensive especially in high power applications. However, the quality of the data is
critical in the model validation process.
7.2 SOURCE DATA FOR MODEL VALIDATION
7.2.1 Field measurements
Field measurements are considered to be best for model validation and the in-the-field type-testing of
IBGs is getting common in many countries and has already been mandatory in some countries,
particularly in Germany. Providing the data quality is guaranteed it is reasonable to use this
measurement data and the model validation is easily accepted on this basis. The field measurements
can be either obtained from specific tests during the system commissioning tests or via a set of recorded
data from system operations. In practice, however, data from field measurements may not be readily
available due to the following reasons
 field measurements and data from them may be expensive

157
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 insufficient data available; only limited tests may be allowed and conducting a test on a real system
requires permission from the authority and detailed coordination of many parties involved
 the commissioning tests will not be available until after the design work is completed
 the accuracy and performance of measurement instruments could not provide the quality of the data
required
 The power systems from which the field measurements were taken are different from the conditions
of the model development. It is important to check under what conditions the measurements were
obtained and an engineering judgement may be required on the validity of the field measurements.
7.2.2 Physical laboratory test systems
Generating the data from a physical test system in a laboratory can be a good alternative for model
validation. The test system is a representation of the AC grid but at a laboratory scale in which, the
engineer has freedom to set the system for its need. It can offer the flexibility of changing the system
strength, simulating the variations in voltage, frequency and fault events in the AC grid. With a test the
system engineer can set up some extreme cases for model validation.
When building a physical test system, it is important to take into consideration the inverter characteristics
and its operation with the primary energy source – solar generation. An analogue system is a good way
to simulate the electrical-mechanic interactions in the AC grid, but it has limitations and it is not easy to
adopt changes. Using a digital system, especially a real time digital simulator, is a popular choice
nowadays. There are several commercial products available for testing the power system including
converter systems. With the development of the IGBT, voltage source converter technology, there have
appeared hybrid simulators where the inverter is a real physical element. Below are two examples.
7.2.2.1 Test facility in China
China EPRI (also known as CEPRI) established a commercial test platform for grid compliance test of
photovoltaic inverters as shown in Figure 7.1. It mainly consists of controllable DC source, grid emulator,
fault generator, anti-islanding test facility and other necessary components such as transformers,
switches and data acquisition equipment. This platform can be also used for model validation studies of
PV inverters. This is the “test-container” according to IEC 61400-27 [148] for WTGs, which can be
applied to PV.
The controllable DC source can emulate the output of PV panels. Different I-V and solar radiation
characteristics of PV panels can be achieved by programming. The grid emulator can generate voltage
and frequency changes of AC system. Harmonics of different orders and amplitude can be injected
through the grid-emulator. The fault generator consists of reactors which can be flexible and combined
in serial and parallel connection. In addition, different grid fault and voltage dip can be emulated by this
fault generator through switch operation. The anti-islanding test facility consists of adjustable R-L-C load
which is used to create the condition where islanding may occur. Using this platform, a set of tests which
are necessary in model validation studies can be performed:
1. Small voltage disturbance test at AC side.
2. Large voltage disturbance (short circuit fault) test at AC side.
3. Frequency disturbance test at AC side.
4. Disturbance test at DC side.
5. Active power dispatch command control test.
6. Reactive power dispatch command control test.
The test system is configurable. It is easy to change its configuration to conduct different tests with
different conditions. It is noted that type-tests of IBGs have already been mandatory in some countries,
such as China and Germany.
Analogue test systems can have limits in representing a high voltage power system. For example, the
R/X ratio of a piece of copper wire at a low voltage is higher than that in the overhead lines of the high
voltage transmission systems. As a result, the wires in the test system can introduce more damping and
the system is less oscillating than the system it is trying to represent. This problem becomes more
severe with generators and transformers. With a generator, it was found that it is easy to reduce the
leakage inductance X but difficult to reduce the winding resistance R without significant enlarging the
size of the generator. A way to cope with it was to increase the leakage reactance X of the model
generator intentionally in order to make its R/X ratio equal to the real generator. In the case of
transformer, the losses of the model transformer are proportionally higher than the high-power
transformers due to higher copper loss. These discrepancies should not be ignored when building a test
system or when analysing the simulation results for model validation.

158
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

10kV Bus
Measuring External Grid
point
Controllable DC PV
Power Source Inverter
ZS
US
Grid disturbance
generator

(a) Single line diagram of test circuit

(b) Controllable DC Power Source (c) Grid disturbance generator

(d) Overview of test facilities

Figure 7.1 Model validation test systems for PV inverter in CEPRI.

159
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

7.2.2.2 Test facility in Japan

The traditional dynamic electric-mechanic laboratory facility can be retrofitted to a test system for
electronic inverter systems. The synchronous generators are useful devices in the test system for the
studies like rotor angle stability. The power swing oscillation is mainly provoked by the physical presence
of the synchronous generators connected through the transmission lines/cables to the rotating machines.
The power swings between the power grid and the dynamic response of the IBGs can be studied on
this test system.
Figure 7-2 shows an old test facility but still in use in Central Research Institute of Electric Power Industry
(CRIEPI), Japan. Figure 7-2 (a) shows a synchronous generator unit rated at 100 kVA. It has a massive
core rotor instead of a laminated core to represent the eddy current on the surface of the rotor. This
design gives a large leakage reactance X so as to make its ratio of X/R a representation of generators
up to 1000 MW class. This model generator can produce the dynamic behaviours close to those that
are 10,000 times greater during fault conditions. The length of the emulated transmission line can be as
much as 600km. This facility has been satisfactorily used in study of low frequency power swing
oscillations.

(a) Synchronous generator units (100 kVA) (b) 275 kV emulated transmission line (LL voltage: 3.3 kV)

(c) Single phase PV inverters (d) DC source emulator

Figure 7.2 Model validation test systems for PV inverter in CRIEPI.

160
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Connected with the inverters as shown in Figure 7.2 (c), this system has been used in the test of the
control system of the inverters.
The effectiveness of the PVs with and without low voltage FRT capability to the power oscillation
damping was tested on this platform. The generator N2 was connected to an infinite busbar AC system
through a 300 km emulated transmission lines. PV generation was consisted of 24 single phase roof-
top PV inverters and each was rated at 4-5 kW. The total generation from the PV panels was 76kW,
which was embedded within the load network as showed in detail. The total load was 76kW exactly.
Their connection point was close to the generator N2. A three phase fault with a duration of 80 ms was
applied i.e., the faulted line was tripped 80ms after the fault occurred. The automatic recloser for the
faulted line was disabled in the test case. Two tests have been carried out; (1) without PV and load, and
(2) with both PV and load, but the PV does not have FRT functions.
In test (1) the synchronous generator (N2) entered an out-of-step condition immediately after the fault
occurs.
The result of the test (2) was showed in Figure 7.3.

330 0 V
165 0 V 165 0 V 220 V a)
( 275 kV) PV
( 66 kV ) ( 66 kV ) ( 6. 6 kV )
60 kVA b)
PV
Tr . # 10
c)
PV
90 kVA× 2
Loads
Dis tribu tion
275 kV Emulated Transformer
300 km L ines

165 0 V 220 V
   0 PV a)
70kW Infinity ( 66 kV ) ( 6. 6 kV )
60 kVA b)
N2 Bus PV
Tr . # 11
70kW (Utility)
CCP：0kW PV c)
70 kW51kvar
Rating: Loads
PV+PL Dis tribu tion
100 kVA/9 0 kWPV:76kW
Transformer
PL：76 kW a)： 4kW × 4 sets: VAB
b)： 4kW × 4 sets: VBC
c)： 4kW × 4 sets: VCA

200 4
Summation of PV 1-24 Output
180 Rotor angle relative to infinite bus
3.5
Terminal voltage of PV
160
3
140

2.5
120

100 2

80
1.5

60
PV15-16 1
40 Restart

0.5
20 PV17-24 PV1-8
PV9-16 PV7-8 Restart PV9-16
Stoppage Stoppage Restart
Restart
0 0
0 PV1-8 20 PV1-6 40 60 80 100
Recovery Stoppage Times in second

Figure 7.3 Example interactions between synchronous generator and 24 single-phase PV inverters.

161
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

As the PV inverters did not have the FRT control function, the PV inverters were disconnected when the
terminal voltage of the PV dropped below the 70% of the nominal voltage. There were a number of
stoppages and restarts by different suites of PVs. It took over 90 seconds for the generator to result in
the out-of-step operation/condition.
These tests showed one fact. i.e. the PVs installed near a power source and do not have the low voltage
FRT function can actually improve transient stability in comparison with the PVs which have the low
voltage FRT function (see also Clause 3.3.3 with Figure 3.20). When the PVs are located close to the
synchronous generator (i.e. near power source) and without low voltage FRT capability, they can
produce an effect equivalently as dynamic braking. As shown in Figure 7.3, the disconnection of PVs
near the power source resulted an increase in the net load. This reduces the active power transfer over
the transmission lines and makes the system more stable.
As already mentioned in Chapter 3, the influence of the PVs on transient stability depends on the location
of the PVs i.e. near the power source or the power sink. If the location of the PVs is closer to the load
centre, i.e. the power sink, the opposite and adverse effect can be observed.
Because the FRT requirement will prevent the PVs from disconnecting from the grid, the equivalent
dynamic braking effect will be diminished, which means transient stability decreases as the FRT
requirement is applied to the PVs which are connected to the bus near the accelerating generators.
The dynamic braking effect coming from the disconnection of PVs depends not only on the location of
the PVs in the network but also on the penetration level of PVs. If the penetration rate is low, the
equivalent dynamic braking effect may be negligible and the lighter power flow over the transmission
lines contributes to the increase of the transient stability
7.2.3 Real-time digital simulator in North America
Nowadays, digital real time simulators (DRTS) are often used to help in the development of converter
systems. In order to test such a converter, the DRTS which represents the surrounding rest of system
(ROS, e.g. the DC and AC grids to which a DC/AC converter may be connected in real life) is connected
to appropriately sized amplifiers such that the simulated AC and DC quantities can be imposed upon
the converter as necessary. By feeding the electrical responses of the converter back to the DRTS a
power hardware in the loop (PHIL) simulation can be established. In such a PHIL simulation, all the
control functions of the converter are fully represented.
If a PHIL setup is unavailable, the physical control and protection unit can be connected to the DRTS to
form a controller hardware in the loop (CHIL) simulation. Such a setup can be used very effectively to
test the control and protection system within a realistic system environment. There can be many
measurement points where the parameters of the primary system, variables within the inverter and the
control signals and variation are accessible.
The Centre for Advanced Power Systems (CAPS) at Florida State University in Tallahassee, FL, USA,
can serve as an example for a state-of-the-art PHIL and CHIL facility. This laboratory was designed to
test electrical and mechanical hardware under test (HUT) with power ratings up to 5 MW. As depicted
in Figure 7.4, the facility offers many different flavours of power connections all fully interfaced with the
DRTS.
Controller hardware in the loop (CHIL) experiments are achieved through either dedicated wiring using
analogue and digital signals or standard communication network protocols such as distributed network
protocol (DNP3), and IEC 61850 (communication networks and systems in substations). In both cases—
PHIL and CHIL—the HUT is part of a closed-loop simulation which allows evaluation of dynamic
responses, while the remaining part is modelled and simulated using the real-time environments.
CAPS has utilized real-time simulation capabilities successfully in numerous projects, such as dynamic
testing of a prototype 5 MW high-temperature superconducting motor [149], a high-speed generator
[150], superconducting fault current limiters [151], PV converters, and numerous power converters for
the US Navy. CAPS also successfully conducted co-simulations of electro-mechanical and electro-
thermal systems [152], [153].
Most recently, integration of all the PHIL and CHIL resources begun to allow experimenting with up to
approximately 100 distributed control nodes simultaneously. This will provide researchers with the ability
to test and evaluate control algorithms for emerging next generation terrestrial and shipboard power
systems in a realistic system environment to meet technology readiness level 6 (TRL6) requirements at
reasonable system scale. Figure 7.5 is a systems level overview of the open-source distributed control
concept that is envisioned after full implementation of this DURIP award. The two major DRTS engines
at CAPS are made by two different manufacturers. They provide the ability to simulate large-scale

162
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

systems and provide emulation characteristics and commands to physical equipment. The physical
systems near the top of the diagram are connected via fibre optics to the DRTS. At each of the physical
devices is an array of distributed control nodes corresponding to physical controller devices. In addition
to the physical systems, virtual loads and sources are present in the simulation. In the same way as the
physical devices, the virtual devices are connected to physical controller devices. Each of the distributed
control platforms are connected via Ethernet to create a true networked distributed control layer. An
emulated network layer is also added, which via emulation software can more accurately represent the
complexity of a physical distributed network.
Dynos

VVS
0…4.16 kV 0…1.1 kV
0…0.4/0.8 kA 0…2.5 kA

Embedded Controls
1.5 kA

Real-time Controls
Network Simulator
OPNET
DRTS

Figure 7.4 The 5 MW PHIL facility at Florida state university-CAPS.

Low Voltage 10kVA

5 MW MMC-based Other
5 MW VVS Distributed Control
Dynamometer 5MW MVDC Hardware
Test Bed
Controller Controller Controller(s) Controller(s) Controller(s)

Network

Controller(s) Controller(s)
Simulated Simulated
Sources Loads

Real-Time Computational Network Emulation

Systems (RTDS, OPAL-RT) (Opnet, Omnet)

Figure 7.5 CAPS open-source distributed control system layout.

7.2.4 Other model responses

Comparison between the different models running on different simulation programs is also widely used
in model validation. For example, a simplified RMS model can be checked against a well- proven EMT
simulation. It should be pointed out that it is often not that easy to reach a full agreement between RMS
and EMT simulations. One should not expect so either, even when running the same event with the
same conditions. This can be due to many reasons. The EMT model is built with details of converter
components and its control system for simulating the fast transients and capturing the variations in very

163
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

small time scale. The RMS model is often built with some approximations for representing the
characteristics of the converter to the external system at the point of interest. The RMS is used to study
relatively slow variation processes by ignoring or simplifying the fast variations. The different algorithms
used in the RMS and EMT simulations can also introduce some discrepancies to a certain extent.
Also, the IBG models used in different software platforms are compared and the models are adjusted to
show identical performance in all the software used. In WECC, three different software platforms are
used for power flow and dynamic stability studies with RMS models. In development of the new models,
the models are developed simultaneously in these three simulation programs. The model performance
is compared and the models are validated using these three software platforms to ensure that the
models and simulations are identical.
7.3 RECOMMENDED SPECIFICATION OF MEASUREMENT DEVICES
The precision of the measurement transducers like CT and VT needs to be carefully considered. Their
precision can affect the readings of the parameters like active power flows. IEC 61400-21-1 [154]
provides some guidelines to follow. The bandwidth of the transducers should adequately cover the range
of frequencies of the measured parameters. The response speed of the transducer is critical for
obtaining the quality measurement data for validating the model dynamics. It is desirable that the
response speed of the transducer is equal to or less than 20ms for the RMS model validation.
The transient (or so called dynamic) fault recorder is an instrument designed to capture slow variation
and fast transient events. It is capable of logging several analogue and digital measurements
simultaneously. It is recommended to install a transient fault recorder at the grid connection point of the
PV generation site and log the voltage and current of the connection circuit as well as major parameters
from the inverter. These measurements can be useful for checking the plant performance as well as for
the model validation.
Based on the industry experience for model validation, a sampling rate of 10 kHz has been used for
data recording. This rate was found to be adequate to capture most of transient events in terms of P
and Q and V and I for RMS model validation.
7.4 VALIDATION PROCEDURES
The process of the model validation can be described by the following steps:
1. Preparation for model validation
 Set up the model for simulation studies in the computer program
 Ensure the input parameters (like P, Q command order, solar irradiation energy input (W/m 2)
and output parameters (like P, Q and V, I at the PCC) are accessible and can be measured
 Ensure the AC network model is accurate in terms of strength, loading level etc. for the type
of studies
 Run the simulation with different conditions and scenarios to ensure the model is robust and
the results are reliable
2. Selection of the type of tests/studies to be performed
 Define the events (or studies) that the model is intended for
 Define the tests to be carried out with the details of operation points, the variation range,
time interval and test duration and measurements etc. (as will hereinafter be described in
detail).
3. Performing the validation test
(this can be an independent study if the simulation method is used for model validation)
 Performing the tests and logging in the measurement data in the tests
4. Performing the computer simulation studies
 Run the computer simulation study for each case defined in Step 2.
5. Comparing the test results
 If the simulation results match the test measurement data within the defined tolerance, the
validation is completed and the model is accepted.
 Otherwise, change should be made to the model and then repeat the tests in step 4 till the
satisfaction results are obtained.

164
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 7.1 Pros and cons of play-back method and full grid simulation method [155]
Play-back method Full grid simulation method
Advantage Uncertainty coming from the grid Interaction between IBG and grid can
model can be removed. be validated.
Disadvantage Interaction between IBG and grid Uncertainty coming from the grid
cannot be fully validated. model may be included.

Model validation is an iterative process where changes to the model are often needed. Depending on
the range of variations and the events that model covers, some parameters of the model may have to
take different values for different events even the model may have to take a different structure to best
fit to the event. The Sub-Chapter 7.6 in the following presents an example where the measurement data
is grouped into three different catalogues depending on the variation speed used to verify different parts
of a model (it is a suite of models containing some logic switches to change the model structure as well
as parameters).
IEC 41600-27-2 [155] presents the following two methods and techniques for model validation. One is
called “Play-back method” and the other “Full grid simulation method”. The Play-back method is a test
with the system being connected to a very strong AC grid. As the AC grid is so stiff, there is no interaction
between the IBG and the grid. Any changes found will be from the inverter so it is similar to an open-
loop test without any feedback from the grid. The Full grid simulation method takes into account the
interactions between the IBG and the grid. It is like a closed-loop test. Table 7.1 presents the pros and
cons of these two methods.
The above steps of the model validation procedure can be illustrated in Figure 7.6.

Step 1 Preparation for Validation

Step 2 Selection of Type of Test/Study

Step 3 Performing Test (in Lab./Field)

Step 4 Computer Simulation Study

Comparison Between Measured and Simulated Response Further work on Model Parameters
and/or Model Structure
Step 5
No
Is the comparison satisfactory?
Yes
End of Validation

Figure 7.6 Flowchart of general model validation procedure.

A set of criteria is needed for comparison to judge the adequacy of model validation. Which parameter(s)
should be used for this comparison depends on the studies. For example, in the rotor angle stability
studies the maximum power transfer over a transmission line can be chosen to be the parameter to
judge the system stability. In the case of a synchronous generator, this line is one that is connected to
the generator and used to transmit the power flow out from the generator. In the case of PV plant, it is
the line that is connected to the inverter and used to transmit the power out of the inverter.
The detailed model validation procedure based on “industry practice” has been reported in [156].
7.5 STUDIES AND TESTS TO BE CARRIED OUT FOR MODEL VALIDATION
It is recommended that the following studies and tests in Table 7.2 should be considered for model
validation. The measurements required are P, Q, Vrms, and Irms at the output terminals of the inverter
or at the grid connection point for several inverters.

165
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 7.2 Type of studies and tests for model validation

Tests Main features

This test is to determine the dynamic response of the inverter system in
System dynamic response to response to a small change in control signal. The control signal can be either
control signal Vref or Qref, or active power P or frequency f. The test is for checking the
parameters like droop, deadband, control limit and range, the rise time,
settling time and overshoot (if there is) in response.
Set the inverter to the appropriate operation mode. In the tests for the
response to P (or f), a variable prime energy source at the inverter DC side
may be required, and the voltage and current at the inverter DC side should
be monitored.
Change the control signal in a small range, e.g. (-10%, +10%)
Measure the time-trend of voltage, current, active and reactive power of the
inverter.
This is a test of close-loop control function to the small variations.
System dynamic
The test is to determine the dynamic characteristics of the inverter system
response to small
in response to a change in the connected AC grid. The simulated variable
disturbances in grid
represents the disturbance in the AC voltage or frequency.
Set the inverter at an appropriate operational mode (P/V or P/Q); emulate
a step change in the grid voltage, or a frequency change according to the AC
system frequency characteristic; monitor the time-trend of the variables at
the output of the inverter.
This test should cover the sunny, cloudy and night conditions.
This is a test for validating the model that is designed for long-term voltage
Long-term voltage stability study by simulation in time-domain
stability
Analyse the AC grid long-term voltage stability at the inverter connection
point. This covers the time duration from a few to several tens of seconds,
and includes the effect of the load, tap changes, power plan control (AGC,
excitation limit) and system operator action etc.
Set the inverter at an appropriate operational mode (P/Q or P/V etc.);
Emulate a change in the AC voltage that can cause an occurrence of a long-
term voltage stability; monitor the time-trend of the variables at the output
of the inverter.
Repeat the test with different power input to the inverter at the DC side.
It is a test for determining the inverter capabilities in response to the large
Short-term voltage stability. voltage drop in the AC system and in the case of a short circuit.
This includes FRT and short-
Set the inverter to an appropriate operation mode;
circuit current contribution
Simulate the grid voltage dip for a specified duration. The depth of the dip
is in the range of 80% to 25% of the remaining voltage. Duration should be
defined according to the relevant standards. In general, the duration is
inversely proportional to the depth of the voltage dip, i.e. the lower the
voltage dip, the shorter the dip duration is.
Monitor the parameters at the output of the inverter to determine its
capability during the fault and in the recovery period.
Simulate the grid voltage dips down to very low level in the range of 15% to
5% of the remaining voltage for a duration such as 150ms. In this case, the
current from the inverter is the maximum current to the AC fault.

166
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Tests Main features

This test is for determining the inverter response to the changes in
Rotor angle stability frequency and voltage in a time scale from tens milliseconds to a few
seconds.
It works in the time phase when electro-mechanical interactions take place.
The different systems can have very different transient behaviours.
Analyse the AC grid transient rotor angle stability. This should include
generator and motor dynamics, generator inertia, SVC, MSC and reactor
control and DC system control.
Simulate some credible fault events that can cause the generator Out-of-
Step.
Start at an appropriate operation condition, test the inverter response to
the changes in the AC system when a single Out-of-Step and/or multiple
Out-of-Step happens.
This is the test for determining the operation range, protection and dynamic
Solar radiation variation characteristics of the inverter in response to the change of input energy as
prime energy source.
The strength of the PV solar irradiance depends on geographic location,
installation, PV panel and time and season variation. It can vary in a wide
range from a few thousand W/m2 to hundreds W/m2 and even zero.
Set the inverter operate in the energy delivery mode;
Simulate the change of the input power at the inverter DC side in a similar
pattern as the solar radiation variation
Monitor the outputs of the inverters as well as the inputs of the inverter
This test is for determining the inverter dynamic response to the large
Frequency instability discrepancy from the nominal of the system frequency
Analyse the grid frequency characteristics in the event of generator tripping
or a line outage causing a large loss of demand. As the result the energy
generation and demand become significant imbalance temporarily. The
frequency change also depends on the system inertia.
Simulation a tripping event that causes the AC frequency exceeding the
LFSM level
Monitor the inverter response to such an event
This is the test to determine the inverter control capability in response to
Unintentional islanding islanding.
detection
Depending on the way the inverter is connected, the inverter can operate in
the f/V mode (the inverter feeds the island solely) or in the P/V mode (the
inverter is in parallel with an AC line feeding the island).
Strong fluctuations are expected when a part of the grid become an island.
The induction motors and static loads can affect the frequency and voltage
in the isolated part of the grid. The event can last from tens milliseconds up
to hundred seconds
Monitor the inverter response to the changes in frequency, voltage and
power.

It should be pointed out that not all the tests in Table 7.2 should be carried out in model validation. As a
minimum, the model validation should be exercised for the studies it is designed for. For example, if a
model of a PV converter is designed to represent the PV converter’s anti islanding (AI) protection, the

167
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

model must be validated against the tests following the system requirements or according to the relevant
standard such as IEEE Standard 1547 in the US [18].
The IEC 61400-27-2 [155] which describes the validation of the RMS model for power system dynamic
studies can be referred to. This document is currently under review.
If there is another customer (third party) connected at or near the busbar where the PV generation plant
is connected to, the connection point becomes a point of common coupling (PCC) from the Grid
perspective. In this case the connection busbar must be compliant with the network connection and
operation requirements. The study cases should be defined according to the relevant Grid Code.

7.6 MODEL VALIDATION EXAMPLE

This sub-chapter presents an example of parameter identification of the PV inverter model in the process
of the model validation. The model consists of PV arrays, inverters, power generation unit network and
the plant control system.
The individual component is modelled analytically for its steady state and dynamic characteristics. As
the dominate parameters are not the same under different disturbances. A number of different tests are
proposed to produce enough useful data to identify these dominant model parameters. An actual electric
system was constructed as a platform to perform the tests on the PV system used in the example. It can
also be used to perform the grid compliance test and emulate AC system disturbances such as small
and large voltage and frequency deviations, system fault, large radiation variation (DC power input
variation) and so on. These tests can be used for model validation. The Chinese National Standard –
Model and Parameter Test Regulation for Photovoltaic Power System is based on this work.
7.6.1 Model validation based on laboratory test
Laboratory tests are convenient to validate the developed model at different scenarios so as to
understand the dynamic behaviours of inverter. Such a test was performed in the laboratory and the
results are shown in Figure 7.7.
The following RMS dynamic simulation model validation is used for a PV inverter of 500 kW capacity.
The first model validation scenario is large voltage disturbance on the grid side, the terminal voltage of
inverter is set to 0.05pu to simulate a fault. Due to the support effect of inverter reactive current, the
voltage is raised to 0.107pu. In Figure 7.7, the black line represents measurement data and the red line
represents simulation data. It can be seen that the measurement data and simulation data are in close
agreement.
There is a reactive power peak after fault is cleared in Figure 7.7. The voltage recovers quite quickly as
the short circuit fault is clear, but the current does not recover. When the inverter controller detects the
voltage recovery, it issues the command to reduce the reactive current. Upon this command the inner
current control loop controls the current with a delay. It is possibly this delay that causes the overshoot
in current. The possible reasons for the delay are:
 the time delay of the measurement including voltage measurement, PLL and abc-dq reference
transformation;
 the time delay of LVRT controller;
 the time delay of inner current control loop. In general, the duration time is approximately no longer
than 30ms.
The second model validation scenario is active power command disturbance. Set the same active power
command curve in test and simulation respectively, and the deviation between measured and simulated
data is calculated. The example of the second model validation test is shown in Figure 7-8.

168
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 7.7 Model validation of large voltage disturbance.

Figure 7.8 Model validation of active power command disturbance.

169
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

7.6.2 Model validation based on field test

Short-circuit or ground faults in the real network are useful in the validation of the developed model to
demonstrate the dynamic behaviour following such faults. The measurements following faults can be
utilized to then validate RMS dynamic simulation model.

750 kV substation

AC/DC convertor station

~
330 kV substation

110 kV substation

Heat power plant

Hydro power plant

Wind power plant

PV power plant

Load

PV power
plant A

Figure 7.9 Schematic diagram of Ningxia network.

170
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

In 2016, the single-line-to ground artificial faults were applied to a 750 kV transmission line in the Ningxia
province. The fault line / point was selected considering the renewable energy generation and HVDC
connected with the Zhejiang province system. The fault point was close to the 750kV substation shown
in Figure 7.9. PV plant A is interconnected to the secondary side of the 750/330 kV transformer. The
rated voltage of the POI is 35 kV and the rated capacity is 6.876 MW (See Table 7-3). Due to the large
scale of IBG in this area, not all of plants are marked in Figure 7.9.
The structure of PV plant A is shown in Figure 7.11, and the basic information is shown in Table 7.3.
POI of PV
power plant

35kV Bus

Generation Generation Generation

Generation unit 2# unit 3# unit 4#
Unit
unit 1# Transformer

Collective box 1-1#

Collective box 1-2#

Collective box 1-3#

Collective box 1-4#

Collective box 1-5#

Collective box 1-6#

Collective box 1-7#

Collective box 1-8#

Collective box 1-9#

AC AC AC AC AC AC

DC DC DC DC DC DC

Figure 7.10 Structure of PV power plant A.

Table 7.3 Major specification of PV inverter
The capacity of PV power plant (MW) 6.876
The capacity of each PV inverter (kW) 36
The number of PV inverters 191
The capacity of each unit transformer (kVA) 1.6
POI voltage of PV power plant (kV) 36.75
AC Voltage of PV inverter (V) 480

The dynamic behaviour at the POI of PV plant A was recorded and is shown in the black line of Figure
7.12. It can be seen that the ground fault occurred at 0.2 second, and the fault was cleared at 0.24
second. Immediately following the fault occurrence, the voltage at the POI dropped to 0.83 p.u., the
reactive power at the POI increased to 0.055 p.u., and the active power at the POI decreased to 0.2
p.u.
The PV power plants were modelled using a commercially available time-domain simulation environment.
The developed CEPRI model was used for the model validation. The simulated response and measured
response are shown in Figure 7.12. The identified models and their parameters are also shown in Figure
7.13 and Table 7.4. More detailed information of the model used is shown in Appendix 5-C-2. It can be
observed that the FRT controller model of PV power plant provided the reactive power to the grid
following the fault, while the FRT controller model reduced the active power following the fault.
Following the fault clearance, the voltage at the PV terminal recovers and the local controller model
displays the behaviour of the active and reactive power flowing into the grid with a fixed ramp limit rate
for the active power. Therefore, the local controller model and the FRT controller model can precisely
demonstrate the dynamic behaviour of the active power and the reactive power.

171
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Branch (1st line: Active power[MW], 2nd line: Reactive power[Mvar], 3rd line: Current [kA])
Node (1st line: Voltage [kV], 2nd line: Voltage [p.u.], 3rd line: Voltage phase [deg])

Figure 7.11 Representative structure of PV plant A [157].

172
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 7.12 Example of measured and simulated responses following faults at 750 kV [157].

173
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

1 Pmea
P
1 + sTm
Pord Delay dPord_ max
I max PQ _ flag
Active power control

(Tp_ord) ramp limit

dPord_ min - Reset if active
power recovers
Pref 1
Kp + after fault clear Current limit logic
Pm 1 dPm _ max min sTp
1 + sTmpp
ramp limit
- I max  I qmax  I max

Q priority :  I qmin  - I max
0 I pin
P _ flag
1 1  I p _ cmd
U term ÷ 1 ( PQ _ flag  0)  I p max  I max 2
- I q_2 cmd
1 + sTm 1 + sTinverter
I
 p min  0
 I p max  I max
1 I 0
÷ 1 + sTinverter I max P priority :  p min
Reactive power control

Pmea I
( PQ _ flag  1)  qmax  I max - I p _ cmd
2 2
dQord_ max 0
PFref 1 - PFref2 Q _ flag
× 0 Qref
Kq +
1  I q_ cmd
PFref  I qmin  - I max - I p _ cmd
2 2
PF _ flag 1 1
ramp limit sTq
dQord_ min
- 1 I qin I q_ cmd  lim( I qin , I qmin , I qmax )
Qord Delay I q _ flag
- I max I p _ cmd  lim( I pin , I p min , I p max )
(Tq_ord) Iq 0 0
1 Qmea
Q
1 + sTm

Figure 7.13 Chinese standard local controller module model [113].

Table 7.4 Derived model parameters
Local controller module of CEPRI PV power plant model
Name Description Values
dPord_max Active power reference up-ramp rate limit (pu/s) 0.03
dPord_min Active power reference down-ramp rate limit (pu/s) -0.03
dPm_max Active power up-ramp rate limit when irradiation increases (pu/s) 0.03
dQord_max Reactive power reference up-ramp rate limit (pu/s) 999
dQord_min Reactive power reference down-ramp rate limit (pu/s) -999
Tm Measurement lag time constant(s) 0.01
Tmpp Equivalent MPPT lag time constant (s) 2
Tp_ord Delay time of active power order (s) 0
Tq_ord Delay time of active power order (s) 0
Kp Active power regulator proportional gain 0.1
Tp Active power regulator integral time constant (s) 0.03
Kq Reactive power regulator proportional gain 0.1
Tq Reactive power regulator integral time constant (s) 0.03
Imax Maximum apparent current (pu) 1.1
Kreset Reset factor of PI controller in active power control part (pu) 0.9
FRT and protective controller module of CEPRI PV power plant model
Name Description Values
Iqmax_LV Maximum reactive current during LVRT (pu) 1.1
Kq_LV Reactive current factor during LVRT 3.6
Iq0_LV Initial reactive current during LVRT (pu) 0
ULV Low voltage threshold value (pu) 0.9
KIq0_flag Initial reactive current add in flag 0
Imax_FRT Maximum apparent current during FRT (pu) 1.08
Ip_flag Active current limit flag during FRT 2
Kp1_FRT Active current factor during FRT 0
Kp2_FRT Active current factor during FRT 0
Ip0_FRT Initial active current during HVRT (pu) 0.07
dIp_LV Active current up-ramp rate limit after low voltage fault clearance (pu/s) 1.25

It can be shown that the measured responses do not demonstrate the stepwise change at precisely the
fault occurrence, while the simulated responses do, which leads to a sort of mismatch between the
measured response and the simulated response. However, such mismatch is not a serious issue because
the measured response always includes a time delay when detecting the instantaneous V and I values
or during the calculation of RMS values from the instantaneous V and I values.

174
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

7.7 CONCLUSION
The model validation is an important and final step in the model development. It confirms the
assumptions made, the conditions and limitations of the model and gives confidence to the model
developers and users. The quality of the model validation can be only as good as the data used for the
model validation. Several data sources with their pros and cons were presented in this chapter including
the measurements taken from the field in real systems, the data produced on the laboratory test systems,
from real time digital simulators and from other simulation models and programs.
The chapter is limited to the validation work for RMS models. A procedure with a list of studies and test
events is proposed for the IBGs. They present most of the events that the RMS models are required to
perform and should be considered as a guideline in validation. They are by no means inclusive, and
equally, there is no intention to recommend all of them are performed in a model validation. A different
study may require not only a different model (that matches the purpose of the study) but different model
parameters, such as sampling rates, length of data etc.
The results of a model validation based on a laboratory test platform in response to large voltage
disturbance and the system response to power command, and the results of the FRT tests on PV plant
in a transmission grid system were provided in order to demonstrate the actions and how good the
model validation can be.

175
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

176
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

8. CONCLUSIONS AND FUTURE WORK

Any numerical models, i.e. mathematical simulation models have limitations. A mathematical simulation
model will provide correct simulation results to the type of phenomena to be observed or examined by
including appropriate individual model component (such as protections, controls and capabilities). Thus,
the proper selection of the mathematical simulation model needs to be performed by power system
engineers in academia as well as in industry. However, the proper selection of the mathematical
simulation model is not an easy task. The selection of the inverter-based generator (IBG) model is much
more difficult compared to the conventional mathematical simulation models such as the power plant
models and conventional power electronics models such as static var compensator models.
Considerable industry experience concerning power system dynamic studies and the dynamics of the
power electronics is required for the proper selection of the IBG model. The established CIGRE C4-
C6.35/CIRED JWG has successfully gathered a wide variety of experts which fully cover the required
industry experience. Thanks to this JWG, this TB provides the guidance and recommendation for the
IBG models to be used for the power system dynamic studies.
At the beginning of the CIGRE C4-C6.35/CIRED JWG activities, a questionnaire survey was performed
to clarify the currently used IBG models by utilities and system operators for the power system dynamic
studies. Through the analysis of the questionnaire survey, it is remarkable finding that around one third
of the utilities and system operators still use negative load models for the representation of IBG in power
system dynamic studies. According to the results of the questionnaire survey, the following reasons may
explain this approach, by the lack of:
 model requirements of IBG for specific power system phenomena
 well-validated detailed IBG models
 widely accepted generic IBG models
 widely accepted range of IBG model parameters
 specific grid code requirements
 information about dynamic performance of power system with IBGs
 agreed or well documented methodology to perform aggregation of embedded IBGs
 knowledge and experience of IBG impact on daily operation of power system
The dynamic characteristics of IBGs are different when compared to synchronous generators, but with
good control system design and increased functionalities of modern IBG technologies, they can provide
many of the same or even better services (e.g. voltage control, frequency response etc.). None the less,
they do need to be modelled differently and appropriately. Therefore, the development of the proper
computer simulation models for IBGs with such additional functionalities is vital for power systems
analyses.
DSOs have to some extent a representation of their networks and details about connected consumers
and producers. However, the limited data is generally not suitable for dynamic simulations for either the
distribution or the transmission system or, at least, has not been used for that purpose in the past, due
to the high level of detail required. From the point of view of the DSOs, time-domain simulations may
also now be necessary to assess protection system behaviour, distribution network automated operation,
unintentional islanding of part of distribution systems including IBG, voltage issues, etc. For these types
of power system dynamic studies detailed IBG models are needed.
The extraction of the model component, i.e., the necessary functionality for a specific dynamic
behaviour/phenomenon is a key to provide the proper selection of the model. Several major phenomena
such as 1) frequency deviation, 2) large voltage deviation, 3) small and long-term voltage deviation for
the power system dynamic studies with the IBGs are required to be obtained. In order to obtain the
appropriate simulation results for each type of dynamic stability study, more than sixty modelling experts
who were nominated as the members of CIGRE C4-C6.35/CIRED JWG around the world, organised
the necessary functionalities of the IBGs which should be implemented in the model in terms of the
power system phenomena instead of the power system dynamic studies. However, even unnecessary
functionalities may be essential under a special circumstance, even if the table says no need for the
functionality. That means, there can be exceptions. Therefore, it is suggested to carefully read the
descriptions on why the functionality is or was not necessary.
The functionalities are classified into three areas: 1) control, 2) protection, 3) capability. The detailed
clarification for the necessity of the use of the functionalities is also provided for each necessary
functionality. It should be highlighted that the protection model needs to be paid more attention to in

177
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

many power system dynamic studies, because the IBGs are disconnected by their protection systems
and because such disconnection can affect the power system stability.
It is noted that the complete functionality list provides clarification of the characteristics of IBGs focusing
on the difference between IBGs with minimum requirements and those of synchronous generators. The
four main characteristics
 System inertia
 Short-circuit current provision (which is large enough for protections to operate)
 Synchronization capability (existence of a synchronous torque component)28
 Constant internal voltage29 source
are inherently possessed by synchronous generators. However, they cannot easily be emulated (if at
all or if needed) by IBGs from technical or commercial perspectives although many of the characteristics
such as the frequency control capability and the reactive power control capability can be provided by
large-scale IBGs. It should be noted that the IBGs can be much better controlled with much shorter
delays than synchronous generators. In addition to that the IBGs have no critical clearing time, i.e. no
transient stability issue.
The example control block diagrams for representing each functionality are provided for the type of
model, RMS or EMT and depending on the level of the controller, either local level controller or plant
level controller. It should be emphasized that most of the functionalities are existing functionalities which
are currently implemented by commercial software tools for power system dynamic studies. On the other
hand, some functionalities such as the synthetic inertia are not required as a mandatory functionality
and the modelling of such functionalities are not presently offered.
Although the computer technology has been developed, it is becoming more challenging to model the
increasing number of individual generators as part of large-scale dynamic stability studies (mainly due
to the high computational burden and the limited assigned time for completing analysis activities). On
the other hand, correctly representing the aggregated response of the distributed IBGs will influence the
power system dynamic stability results at the bulk transmission system level and become more critical
as the penetration level of distributed IBGs increases. However, there is no generally agreed or accepted
methodology for aggregating IBGs and there are ongoing studies for developing such a methodology.
One of the latest aggregation methodologies which is proposed by WECC is discussed individually in
both power flow representation and dynamic simulation representation.
Other than the establishment of the generally accepted methodology of the aggregated IBGs, there are
several future technical challenges:
 The control and protection model parameter derivation for aggregated models.
 The aggregated active power recovery after the fault is cleared.
 Clarification of the aggregated dynamic behaviour of all IBGs during faults in the case of medium or
high penetration level.
The balance of the levels of details of the load model and the IBG model needs to be considered. The
loads and the IBGs can be connected to the same bus and they can interact with each other. For
example, the level of details of the load model is assumed to be low and the level of details of the IBG
model is assumed to be high. In this example, the sufficient levels of details of the whole system data
cannot be expected and an important behaviour could be missing.
The selection of the IBG model validation test condition and type of IBG model and IBG model elements
are vital for representing dynamic behaviour with IBGs in power system dynamic studies performed by
utilities and system operators. For the model validation, measurement is the key. The required levels of
the measurement data are not always the same for the phenomenon to be observed. The desirable
model validation test condition can vary depending on the type of power system dynamic studies. The
possible different levels of the validation test, such as the laboratory level and the field level are
illustrated. The suitable dynamic behaviour for each level of the tests is also clarified. Then, the
recommended model validation test conditions including the necessary electric quantities which are
generally captured by measurement devices are discussed. The general procedure for the model
validation is illustrated and the permissible model error which can be different depending on the type of

28
Note that this capability is NOT literally required for IBGs because they have no rotor-angle stability issue.
29
The voltage denotes the synchronous internal voltage. It does not denote the terminal voltage of synchronous generators.

178
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

power system dynamic studies is briefly discussed. Because the model error can be caused by three
possible reasons:
 Appropriate model with inappropriate model parameters,
 Inappropriate model with appropriate model parameters,
 Inappropriate model with inappropriate model parameters,
exploring the proper model parameters is usually performed as the first step. If any parameters cannot
obtain a good agreement between the simulated response and the model response, the model itself
needs to be changed. This procedure is generic and is for the model validation not only for IBGs but
also loads and other generator controllers.
It is ideal that the same IBG model parameters may be used for different/various power system
conditions. Identifying one set of model parameters for one incident is relatively easy. However,
identifying the same set of model parameters for more than two incidents is not at all easy. There are
two approaches
 Collect data for all incidents and identify the most reasonable set of model parameters.
 Identify one set of model parameters for all incidents one by one and collect all sets of model
parameters.
Then, the statistical process is applied to all the collected sets of model parameters. However, both
approaches have pros and cons. The derivation of a sort of the universal set of IBG model parameters
for various power system conditions can also be a technical challenge.
Both synchronous generators and IBGs have pros and cons. For example, thanks to the natural feature
or inherent characteristics which come from physics (flux linkage etc.), synchronous generators can
increase/decrease instantaneously at the moment when the fault occurs. In other words, IBGs cannot
attain such immediate/instant response. On the other hand, the ramping speed of voltage and current
provided by IBGs is much higher than that provided by synchronous generators. Therefore, such
prominent ramping speed of controllers of IBGs is expected to be utilised to provide more control
quantity for compensating the deficiency of control quantity at the moment when the fault occurs.
More advanced coordination between IBGs and synchronous generators can be a key to integrate more
IBGs into the grid without deteriorating power system security. In order to obtain the advanced
coordination, the modelling of IBGs become vital and the time-domain simulation using the
aforementioned models is required in order to figure out how those generations can be coordinated.
The time-domain simulation becomes more important especially when
 The interaction between one device/system and another device/system is not negligible.
 The (complicated) operation of protection devices is not negligible.
With the higher penetration of the IBGs, the above two points will be no longer negligible. Therefore,
more effort for better modelling of the IBGs will be required in the future. In other words, the continuous
improvement of simulation accuracy is never too old an issue to address.

179
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

180
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX A. DEFINITIONS, ABBREVIATIONS AND

SYMBOLS
A.1. GENERAL TERMS

App Table A.1 Abbreviations of general terms used in this TB

Abbreviation Full Text Chapter

ACE Area control error 3
ACR Automatic current regulator 2
AGC Automatic generation control 5
APFR Automatic power factor regulator 2
AQR Automatic reactive power regulator 2
AVR Automatic voltage regulator 2
BA Balancing authority 3
BAA Balancing authority area 3
BESS Battery energy storage system 4, 5
CAISO California ISO 3
CAPS Center for advanced power systems 7
CB Circuit breaker 3
CCT Critical clearing time 3
CEC California energy comission 3
CEPRI China Electric Power Research Institute 4, 5, 7
CDA Critical damping adjustment 4
CHIL Contoller hardware in the loop 7
CHP Combined heat and power 4, 5
CIGRE International council on large electric systems 1, 2, 3, 6, 8
International conference on electricity
CIRED 1, 3, 8
distribution
CPU Central processing unit 4
CPUC California public utility commission 3
COI California Oregon intertie 3
Central Research Institute of Electric Power
CRIEPI 7
Insutry
CT Current transformer 4
CVT Current and voltage transformer 4
DC Direct current 1, 2, 3, 4, 6
DEC Decremental energy 3
DFIG Doubly fed induction generator 3, 4
DFT Discrete Fourier transform 4
DG Distributed generator 3

181
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

DLL Dynamic link library 4

DN Distribution network 3
DSO Distribution system operator 1, 2, 3
DTRS Digital real time simulator 7
Defence university research instrumentation
DURIP 7
program
EMS Energy management system 3
EMT Electromagnetic transient 1, 2, 3, 4, 6
European network of transmission system
ENTSO-E 2
operators
EPRI Electric power research institute 3
ERCOT Electric reliability council of tewas 2
ESCR Effective short circuit ratio 4
EU European union 2
FACTS Flexible alternating current transmission system 3, 6
FDNE Flexible dependent network equivalent 4
FFT Fast Fourier transfor 4
FRO Frequency response obligation 3
(F)FRT (Frequency) Fault ride through 1, 2
FSM Frequency sensisitive mode 4
GPS Global positionning system 3
GTO Gate turn off 4
GUI Graphic user interface 4, 6
HUT Hardware under test 7
HV High voltage 1, 2, 3, 5
HVDC High voltage direct current 2, 3, 5, 6
HVRT High voltage ride through 2, 3, 4
IBG Inverter based generators 1, 2, 3, 4, 5
IEC International electrotechnical commission 1, 2, 6
IEEE Institute of electrical and electronics engineers 1, 2
IGBT Insulated gate bipolar transistors 4
IM Induction motor 3
IPS Internal protection system 2
JWG Joint working group 3
LCC Line commuted converter 4
Limited frequency sensitive mode-
LFSM-O 3
overfrequency
Limited frequency sensitive mode-
LFSM-U 3
underfrequency
LOM(D) Loss of main (detection) 3
LRS Load and resources subcommittee 3

182
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

LTC Load tap changer 3

LV Low voltage 1, 2, 3, 4, 5
LV(F)RT Low voltage (fault) ride through 1, 2, 3, 5
LVPL Low voltage power logic 5
MMC Multi-level converter 4
MPPT Maximum power point tracking 1, 2, 3, 4
MC Monte Carlo 3
MV Medium voltage 1, 2, 3, 4, 5
NA Nodal admittance 4
NDA Non disclosure agreement 4
NDZ Non detection zone 3
NERC North american electric reliability corporation 3
NYISO New York ISO 3
OFR Over frequency relay 4
OLTC On load tap changer 3
OOS Out of step 3
ORC ORankine cycle 2
OVR Over voltage relay 4
PCC Point of common coupling 1, 2, 3, 4, 7
PCU Power conditioning unit 4
PDCI Pacific DC intertie 3
PF Power factor 4
PHIL Power hardware in the loop 7
PI control Proportional and integral control 4, 6
PLL Phase locked loop 1, 2, 4
POD controller Power oscillation damping controller 2
POI Point of interconnection 3, 6, 7
PS Power system 2
PSS Power system stabilizer 2, 3
p.u. Per unit 3
PV Photovoltaics 1, 2, 3, 4, 5, 6
PWM Pulse width modulation 4, 6
RER Renewable energy resources 3
RES Renewable energy source 1, 2, 3, 4, 5
RMS Root means square 1, 2, 3, 4, 6
ROS Rest of system 7
RPS Renewable portfolio standard 3
ROCOF Rate of change of frequency 1, 2
RT Real time 4

183
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

RTDS Real time digital simulator 4, 6

SCUC Security-constrained unit commitment 3
SCUD Security-constrained unit dispatch 3
SCR Short circuit ratio 3, 4, 6
SOC State of charge 4, 5
SOFC Solid oxide fuel cell 4
STATCOM Static var compensator 3, 4, 5
SSCI Sub-synchronous control interaction 3, 4
SSR Sub-synchronous resonance 3, 4
SSTI Sub-synchronous torsional interaction 3, 4
SVC Static var compensator 2, 4, 5
SVS Static var system 3
TB Technical brochure 2
Transmission expansion planning policy
TEPPC 3
committee
TN Transmission network 3
TSO Transmission system operator 1, 2, 3, 4
UFR Under frequency relay 4
ULTC Under load tap changer 3
UVR Under voltage relay 4
VSC Voltage source converter 2, 4, 5
VT Voltage transformer 4
WECC Western electricity coordinating council 3, 4, 5
WPP Wind power plant 3, 5
WT(G) Wind turbine (generators) 1, 3, 4, 5
ZVRT Zero voltage ride through 5

184
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

A.2. SPECIFIC TERMS

App Table A.2 Special terms used in this TB

Terms Description Chapter

IBGs with minimum IBGs which were in their infancy and the size of the which
functionalities and was unlikely to be categorized as large utility scale. Only effecive in
with no advanced Functionalities and capabilities which are currently required Chapter 2
capability in some countries additionally were not required for IBGs.
Large capacity synchronous generators which is assumed
to be replaced with the IBGs where there is a high level of
Synchronous Only effecive in
penetration of renewables. It is usually connected to the
generator Chapter 2
high voltage network (e.g. over 200 kV). It is usually
equipped with a standard AVR and a turbine governor.
The synchronous generators inherently have the
synchronizing torque capability which is a very important
factor for angle stability. The synchronizing torque index is
proportional to the internal voltage of the synchronous
Synchronization
generator and the equivalent synchronous generators 2
Torque Capability
and/or the angle difference between the synchronous
generators and the equivalent synchronous generator.
This does not denote the synchronization capability shown
in below.
IBGs have this capability which tells how the IBGs in
Synchronization general capture the voltage angle through a Phase Locked
2, 8
Capability Loop (PLL) algorithm in order to output the active power
and reactive power in a correct phasor form.
Inner current PI control loop, creation of firing pulses
Low Level Control 2, 4, 5
(PWM), PLL.
High Level Control Id/Iq current reference 2, 4, 5
Frequency and voltage control, or P&Q control which are
Local Level Control 2, 4, 5
implemented to a single IBG.
Frequency and voltage control, or P&Q control which are
Plant Level Control implemented to multiple IBGs such as large utility scale 2, 4, 5
IBGs.
The internal control of the inverter which is performed at a
Control Category 2, 3, 5
local level.
The protection relay for the inverter. The grid protection
relay for the inverter is also included. The control for
Protection Category 2, 3, 5
protecting the internal devices is also categorised as the
protection.
The grid-friendly control or the control which owns an ability
Capability Category 2, 3, 5
for improving the grid stability.
It is the nearest point on the public utility network where the
Point of Common
customer’s electrical equipment operation has potential to 1, 2, 3, 4, 7
Coupling (PCC)
impact other customers.
Point Of It is the point on the network where a customer is physically
3, 6, 7
Interconnection (POI) connected to the public supply network.
A filter that consists of two sets of three series inductors i.e.
LCL Filter inverter side inductors and grid side inductors, and one set 2
of three capacitors which are placed between two phases.

185
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

186
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX B. LINKS AND REFERENCES

[1] P. Pourbeik, J. Sanchez-Gasca, J. Senthil, J. Weber, P. Zadehkhost, Y. Kazachkov, S. Tacke
and J. Wen, “Generic Dynamic Models for Modeling Wind Power Plants and other Renewable
Technologies in Large Scale Power System Studies”, IEEE Trans. on Energy Conversion, Sep.
2017, Vol. 32, No. 3, pp 1108 – 1116.
[2] Ö. Göksu, P. Sørensen, J. Fortmann, A. Morales, S. Weigel and P. Pourbeik, “Compatibility of
IEC 61400-27-1 Ed 1 and WECC 2nd Generation Wind Turbine Models,” Conference: 15th
International Workshop on Large-Scale Integration of Wind Power into Power Systems as well
as on Transmission Networks for Offshore Wind Power Plants, Nov. 2016.
[3] CIGRE TB 328, “Wind turbines - Electrical simulation models for wind power generation,” WG
C4.601, Aug. 2007.
[4] G. Lammert, K. Yamashita, L. D. Pab´on Ospina, H. Renner, S. Martinez Villanueva, P.
Pourbeik, F.-E. Ciausiu, and M. Braun, “International Industry Practice on Modelling and
Dynamic Performance of Inverter Based Generation in Power System Studies,” CIGRE Science
& Engineering Vol. 8, June 2017.
[5] G. Lammert, K. Yamashita, L. D. Pab´on Ospina, H. Renner, S. Martinez Villanueva, P.
Pourbeik, F.-E. Ciausiu, and M. Braun, “Modelling and Dynamic Performance of Inverter Based
Generation in Power System Studies: An International Questionnaire Survey,” CIRED -Open
Access Proceedings Journal, Vol. 1, 2017
[6] CIGRE TB 566, “Modelling and Aggregation of Loads in Flexible Power Networks”, WG C4.605,
2014.
[7] WECC, “WECC Guide for Representation of Photovoltaic Systems In Large-Scale Power flow
Simulations. WECC Modelling and Validation Work Group. Jan. 2011, URL:
https://www.wecc.biz/_layouts/15/WopiFrame.aspx?sourcedoc=/Reliability/WECC/PV/Plant/P
ower/Flow/Modeling/Guidelines/-/August/2010.pdf&action=default&DefaultItemOpen=1
[8] DGA Consulting. “International review of frequency control adaptation,” Oct. 2016, URL:
https://www.aemo.com.au/-/media/Files/Electricity/NEM/Security and
Reliability/Reports/FPSS---International-Review-of-Frequency-Control.pdf.
[9] N. Miller. “Technology Capabilities for Fast Frequency Response,” GE Final Report, March 2017,
URL:
https://www.aemo.com.au/-/media/Files/Electricity/NEM/Security and Reliability/Reports/2017-
03-10-GE-FFR-Advisory-Report-Final---2017-3-9.pdf.
[10] IEC “Solar photovoltaic energy systems -Terms, definitions and symbols,” IEC Technical
Specification, IEC/TS 61386, Ed. 2, Dec. 2007.
[11] P. Kundur, “Power System Stability and Control,” New York: McGraw-Hill, 1994.
[12] National grid, “Black Start from Distributed Sources,” Dec. 2017, URL:
https://www.nationalgrid.com/sites/default/files/documents/SOF/Report/Black/Start/2017.pdf
[13] H. Udal, S. M. Villanueva, J. Kilter and A. Baranaukas, “Future System Challenges in Europe.
Contributions to Solutions from Connection Network Codes,” 2016 CIGRÉ USNC International
Colloquium Evolution of Power System Planning to Support Connection of Generation,
Distributed Resources and Alternative Technologies, 2016.
[14] R. Teodorescu, M. Liserre and P. Rodriguez, “Grid Converters for Photovoltaic and Wind Power
Systems,” Wiley, 2011, ISBN: 978-0-470-05751-3.
[15] CENELEC, “Requirements for generating plants to be connected in parallel with distribution
networks - Part 2: Connection to a MV distribution network,” CLC/TS 50549-2:2015, 2015.
[16] ETRA, “Countermeasure for instantaneous voltage dips in power systems,” Electric Technology
Research Association Report No. 67, Vol. 2, 2011 (in Japanese).
[17] ENTSO-E, “Commission regulation (EU) 2016/631 of 14 April 2016 establishing a network code
on requirements for grid connection of generators,” URL: http://eur-lex.europa.eu/legal-
content/EN/TXT/?uri=OJ:JOL_2016_112_R_0001.
[18] IEEE standards coordinating comittee 21, “IEEE 1547-2003 - IEEE Standard for Interconnecting
Distributed Resources with Electric Power Systems,” July 2013.
[19] UL1741, “Standard for Inverters, Converters, Controllers and Interconnection System
Equipment for Use With Distributed Energy Resources,” Ed. 2, 2010.
[20] IEC TC95, “Measuring relays and protection equipment - Part 181: Functional requirements for
frequency protection,” IEC 60255-181 Ed.1, CDV status as of 2018.1.
[21] CIGRE/CIRED TB 613 “Protection of Distribution Systems with Distributed Energy Resources”,
JWG B5/C6.26/CIRED, March, 2015.

187
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

[22] CIGRE TB 421, “The impact of renewable energy sources and distributed generation on
substation protection automation,” WG B5.34, Aug. 2010.
[23] BDEW German Association of Energy and Water Industries,“Generating Plants Connected to
the Medium-Voltage Network,” BDEW medium-voltage guideline, Ed. 3, June, 2008.
[24] V. Lemort, “Sustainable energy conversion through the use of Organic Rankine Cycles for waste
heat recovery and solar applications,” Disseminations and Thesis of University of Liege,
Belgium, Oct. 2011.
[25] Energy Storage Association, “Variable Speed Pumped Hydroelectric Storage,” URL:
http://energystorage.org/energy-storage/technologies/variable-speed-pumped-hydroelectric-
storage
[26] B. Seal, “Common Functions for Smart Inverters, Version 3,” EPRI Report, 3002002233, Feb.
2014.
[27] CEI, Italian National Standards, “Standard-CEI 0-16: Reference technical rules for the
connection of active and passive consumers to the HV and MV electrical networks of distribution
Company,”.(in Italian), Dec. 2012.
[28] DIN, German National Standards, “Automatic disconnection device between a generator and
the public low-voltage grid,” VDE V 0126-1-1, Aug. 2013.
[29] CIGRE TB 458, “Electric Energy Storage Systems,” WG C6.15, 2011.
[30] P. Kundur, J. Paserba, V. Ajjarapu, G. Andersson, A. Bose, C. Canizares, N. Hatziargyriou, D.
Hill, A. Stankovic, C. Taylor, T. Van Cutsem, and V. Vittal, “Definition and Classification of power
system stability,” IEEE Trans. on power systems, Vol. 19, No. 2, May 2004.
[31] National Grid, “System Operability Framework 2016,” Nov., 2016. URL:
https://www.nationalgrid.com/sites/default/files/documents/8589937803-SOF/2016/-
/Full/Interactive/Document.pdf
[32] D. Lew et. al., “Western wind and solar integration study,” NERL Subcontract Report NERL-SR-
550-47434, May 2010.
[33] D. Lew et. al., “Western wind and solar integration study Phase 2,” NERL Technical Report
NREL/TP-5500-55599, Sep. 2013.
[34] N. W. Miller, “Western wind and solar integration study Phase 3,” NERL Subcontract Report
NERL/SR-5D00-62906, Dec. 2014.
[35] RG-CE System Protection & Dynamics Sub Group, “Frequency Stability Evaluation Criteria for
the Synchronous Zone of Continental Europe – Requirements and impacting factors –,”
ENTSO-E Report, March 2016.
[36] CIGRE TB 450, “Grid Integration of Wind Generation,” WG C6.08, Feb., 2011.
[37] P. Pourbeik, J. Sanchez-Gasca, J. Senthil, J. Weber, P. Zadehkhost, Y. Kazachkov, S. Tacke
and J. Wen, “Generic Dynamic Models for Modeling Wind Power Plants and other Renewable
Technologies in Large Scale Power System Studies”, IEEE Trans. on Energy Conversion,
September 2017, Vol. 32, No. 3, Page(s): 1108 – 1116.
[38] P. Tielens and D. V. Hertem, “Grid Inertia and Frequency Control in Power Systems with High
Penetration of Renewables,” Proceedings of the 6th IEEE Young Researchers Symposium in
Electrical Power Engineering, Delft, The Netherlands, April 2012.
[39] P. P Zarina, S. Mishra and P. C. Sekhar, “Deriving inertial response from a non-inertial PV
system for frequency regulation,” in Proceedings of the IEEE International Conference on Power
Electronics, Drives and Energy Systems, Karnataka, India, Dec. 2012.
[40] NEMMCO, “Assessment of Potential Security Risks due to High Levels of Wind Generation in
South Australia - Summary of DIgSILENT Studies (Stage1),” ABN 94 072 010 327, Dec. 2005.
[41] M. Pöller, H. Müller, K. Theron, P. Ravalli and A. Robertson. “Assessment of Potential Security
Risks due to High Levels of Wind Generation in South Australia,”. 4th World Wind Energy
Conference & Renewable Energy Exhibition 2005, Melbourne, Australia, Nov. 2005.
[42] ENTSO-E, “Continental Europe Operation Handbook Policy 1: Load-frequency control and
performance,” URL:
https://www.entsoe.eu/fileadmin/user_upload/_library/publications/entsoe/Operation_Handboo
k/Policy_1_Appendix/_final.pdf
[43] ECOFYS, “Impact of Large-scale Distributed Generation on Network Stability During Over-
Frequency Events and Development of Mitigation Measures,” Sep. 2011, URL:
http://www. ecofys.com/files/files/ecofys_ifk_2011_50_2_hz_summary.pdf
[44] GE Energy Consulting, “The effects of integrating wind power on transmission system planning,
reliability and operations,” March 4, 2005.
[45] California ISO, “Performance of ISO’s system during August 21, 2017 Eclipse,” Oct. 2017, URL:
http://www.caiso.com/Documents/Performance-ISOSystemsDuringSolarEclipse.pdf

188
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

[46] K. Misak, Blackout im Übertragungsnetz – den Ernstfall beherrschen, 2. Fachtagung Blackout -

Auswirkungen, Maßnahmen und Konsequenzen auf allen Spannungsebenen, 6.- 7.5. 2014,
Hannover (in German)
[47] K. Yamashita, H. Renner, S. M. Villanueva, G. Lammert, P. Aristidou, J. C. Martins, L. Zhu, L.
D. P. Ospina, and T. V. Cutsem, “Industrial Recommendation of Modeling of Inverter Based
Generators for Power System Dynamic Studies with Focus on Photovoltaic,” IEEE Power and
Energy Technology Systems Journal, Vol. 5, No. 1, pp. 1-10, March 2018.
[48] M Asmine and C-E Langlois, “Field measurements for the assessment of inertia response for
wind power plants based on Hydro-Québec TransÉnergie requirements,” IET Renewable Power
Generation, Vol.10, No. 1, pp. 25-32, 2016.
[49] J. Brisebois and N Aubut, “Wind Farm Inertia Emulation to Fulfill Hydro-Québec's Specific Need,”
in Proc. of IEEE PES GM 2011, Detroit USA, 2011.
[50] IEC “Short-circuit currents in three-phase a.c. systems - Part 0: Calculation of currents,” IEC
Standard 60909-0 Ed. 2, Jan. 2016.
[51] J. Keller and B. Kroposki, “Understanding Fault Characteristics of Inverter-Based Distributed
Energy Resources,” JTechnical Report NREL/TP-550-46698 Jan. 2010, URL:
http://www.nrel.gov/docs/fy10osti/46698.pdf.
[52] G. Lammert, J. C. Boemer, D. Premm, O. Glitza, L. D. Pabón Ospina, D. Fetzer, M. Braun,
“Impact of Fault Ride-Through and Dynamic Reactive Power Support of Photovoltaic Systems
on Short-Term Voltage Stability,” IEEE PowerTech 2017, Manchester, Great Britain, June, 2017.
[53] K Yamashita, et. al, “Experimental Verification of the Fundamental Influence of High Penetration
of PVs on Power Systems Following System Disturbance,” in Proc. of CIGRE AORC 2014
Tokyo, May 2014.
[54] I. Green, “CAISO Experience with Impact of High Penetration of Renewable Resources on
Short-term Voltage Stability,” IEEE PES GM 2015, Panel - Challenges of Voltage and Reactive
Power Control from Renewable Resources, July 2015.
[55] NERC, “NERC Reliability Standards for the Bulk Electric Systems of North America,” URL:
http://www.nerc.com/pa/Stand/Reliability/Standards/Complete/Set/RSCompleteSet.pdf.
[56] NERC, “Standard Models for Variable Generation,” NERC Special Report, May 2010.
[57] National Grid, “Performance of Phase-Locked Loop Based Converters,” URL:
https://www.nationalgrid.com/sites/default/files/documents/Phase/locked/loop/FINAL.pdf
[58] NERC, “1,200 MW Fault Induced Solar photovoltaic Resource Interuuption Disturbance Report
-Southern California Aug. 2016,” June 2017.
[59] T. Van Cutsem and C. Vournas, “Voltage Stability of Electric Power Systems,” Norwell, MA:
Kluwer, 1998.
[60] C. W. Taylor, “Power System Voltage Stability,” New York: McGraw-Hill, 1994.
[61] IEEE PSDP Committee, “Voltage Stability of Power Systems: Concepts, Analytical Tools, and
Industry Experience,” Aug. 1990.
[62] G.K. Morison, B. Gao, and P. Kundur, “Voltage Stability Analysis Using Static and Dynamic
Approaches,” IEEE Trans. on power systems, Vol. 8, No. 3, pp. 1159-1171, 1993.
[63] P. Aristidou. “Time-domain simulation of large electric power systems using
domaindecomposition and parallel processing methods,” Dissertation, Université de Liège,
June 2015.
[64] T. Van Cutsem and L. Papangelis, “Description, modeling and simulation results of a test system
for voltage stability analysis,” Internal Report, University of Liège (Nov. 2013). URL
http://hdl.handle.net/2268/141234
[65] P. Aristidou and T. Van Cutsem, “A Parallel Processing Approach to Dynamic Simulations of
Combined Transmission and Distribution Systems,” In International Journal of Electrical Power
& Energy Systems, Vol. 72, pp. 58–65, 2015.
[66] P. Aristidou, S. Lebeau, T. Van Cutsem, “Power System Dynamic Simulations using a Parallel
Two-level Schur-complement Decomposition,” IEEE Trans on Power Systems, Dec. 2015.
[67] M.J. Gibbard, P. Pourbeik and D.J. Vowles, “Small-signal stability, control and dynamic
performance of power systems,” 2015, URL:
https://www.adelaide.edu.au/press/titles/small-signal/
[68] S. Eftekharnejad, V. Vittal, G. T. Heydt, B. Keel, and J. Loehr, “Small Signal Stability
Assessment of Power Systems With Increased Penetration of Photovoltaic Generation: A Case
Study,” IEEE Trans on Sustainable Energy Vol. 4, No. 4, Oct. 2013.
[69] J. Machowski, J.W. Bialek and J.R. Bumby. “Power System Dynamics Stability and Control,”
2nd Edition, John Wiley & Sons, 2008.
[70] M. Stubbe, A. Bihain, J. Deuse and J.C. Baader. “STAG – A New Unified Software Program for
the Study of the Dynamic Behavior of Electrical Power Systems,” IEEE Trans. on Power
Systems, Vol. 4, No. 1, pp. 129-138, 1989.

189
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

[71] N. Fatabo, R. Rgnedal, and T. Carlsen. “Voltage Stability Condition in a Power Transmission
System Calculated by Sensitivity Methods,” IEEE Trans. on Power Systems, Vol. 5, No. 4., pp.
1286-1293, November 1990.
[72] C. Leaitre, J. P. Paul, J. M. Tesseron, Y. Harmand and Y.S. Zaho, “An Indicator of the Risk of
Voltage Profile Instability for Real-Time Control Applications,” IEEE Trans on Power Systems,
Vol. 5, No. 1, pp. 154-161, 1990.
[73] CIGRE TB 111, “Analysis and Control of Power System Oscillations”, TF 07 of Advisory Group
01 of SC 38, Dec. 1996.
[74] P.M. Anderson and A.A. Fouad, “Power System Control and Stability,” Second edition, Wiley-
IEEE Press, 2003.
[75] Ian A. Hiskens, “Dynamics of Type-3 Wind Turbine Generator Models, IEEE Trans. on power
systems,” Vol. 27, No. 1, Feb. 2012.
[76] CIGRE TB 671, “Connection of wind farms to weak AC networks,” WG B4.62, Dec. 2016.
[77] IEC, “IEC-61400-27-1:2015, Wind turbines - Part 27-1: Electrical simulation models - Wind
turbines,” Ed. 1, Feb. 2015
[78] J. Sun, “Modeling and Analysis of Harmonic Resonance Involving Renewable Energy Sources,”
International Conference on Power Systems Transients (IPST2013) in Vancouver, Canada July
18-20, 2013.
[79] H. Hu, Q. Shi, Z. He, J. He, S. Gao, “Potential Harmonic Resonance Impacts of PV Inverter
Filters on Distribution Systems,” IEEE Trans. on sustainable energy, Vol. 6, No. 1, Jan. 2015.
[80] F. D. Freijedo, S. K. Chaudhary, R. Teodorescu, J. M. Guerrero, C. L. Bak, L. H. Kocewiak,
and C. F. Jensen, “Harmonic resonances in wind power plants: modeling, analysis and active
mitigation methods,” Proc. of the IEEE PowerTech Eindhoven, pp. 1–6, June–July 2015.
[81] L. Harnefors, M. Bongiorno, and S. Lundberg, “Input-admittance calculation and shaping for
controlled voltage-source converters,” IEEE Trans. on Industrial Electronics, Vol. 54, No. 6, pp.
3323–3334, Dec. 2007.
[82] J. Sun, “Impedance-based stability criterion for grid-connected inverters,” IEEE Trans.on Power
Electronics, Vol. 26, No. 11, pp. 3075–3078, Nov. 2011.
[83] L. Sainz, J. J. Mesas, L. Monjo, J. Pedra and M. Cheah-Mané, “Electrical Resonance Instability
Study in Wind Power Plants”, in Proc. of Electric Power Quality and Supply Reliability (PQ),
Tallinn, 2016.
[84] P N. Papadopoulos and J. V. Milanovic, “Problematic Framework for Transient Stability
Assessment of Power Systems with High Penetration of Renewable Generation,” IEEE Trans.
on Power Systems, Vol. 32, No. 4, pp. 3078-3088, July 2017.
[85] P. N. Papadopoulous, A. Adrees and J. V. Milanović, “Probabilistic Assessment of Transient
Stability in Reduced Inertia Systems,” in Proc. of IEEE PES GM, Boston, USA, July, 2016.
[86] A. Adees, P. N. Paoadopoulos and J. V. Milanovic, “A Framework to Assess the Effect of
Reduction in Inertia on System Frequency Response,” in Proc. of IEEE PES GM, Boston, USA,
July, 2016.
[87] H.W. Dommel, “Digital Computer Solution of Electromagnetic Transients in Single- and
Multiphase Networks”, IEEE Trans. On Power Apparatus and Systems, Vol. PAS-88, No. 4, pp.
388-399, April 1969.
[88] Dommel H. W., Scott-Meyer W. S. ‘‘Computation of electromagnetic transients,’’ Proc. IEEE.;
62, pp. 983–93 Jul. 1974.
[89] Woodford D. A., ‘‘Electromagnetic design considerations for fast acting controllers,’’ IEEE
Transaction on Power Delivery. Vol. 11, No. 3, pp. 1515–21, 1996.
[90] Gole A. M., Woodford S. A., Nordstrom J. E., Irwin G. D., ‘‘A fully interpolated controls library
for electromagnetic transients simulation of power electronic systems,’’ Proc. of International
Conference on Power System Transients; Rio de Janiero, Brazil, June 24–28, 1995.
[91] A. Ametani, “Numerical Analysis of Power System Transients and Dynamics,” The Institution of
Engineering and Technology 2015.
[92] U. D. Annakkage et.al., “Dynamic System Equivalents: A Survey of Available Techniques,” IEEE
Trans. on Power Delivery, Vol. 27, No. 1, Jan. 2012.
[93] W. Tinney and W. Meyer, ‘‘Solution of large sparse systems by ordered triangular factorization,’’
IEEE Trans. on Auto. Control. Vol. 18, No. 4, pp. 333-346, Aug 1973.
[94] J Lin; J. R. Marti “Implementation of the CDA procedure in the EMTP,” IEEE Trans. on Power
Systems, Vol. 5, No. 2, pp 394 - 402, DOI: 10.1109/59.54545, 1990.
[95] C. W. Ho, A. E. Ruehli and P. A. Brennan, “The modified nodal approach to network analysis,”
IEEE Trans. on circuits and systems, Vol. 22, No. 6, pp. 504-509, June 1975.
[96] T. Maguire, B. Warkentin, Y. Chen and J. Hasler “Efficient Techniques for Real Time Simulation
of MMC Systems,” International Conference on Power Systems Transients (IPST2013), 2013.

190
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

[97] T. Maguire, P. Forsyth and R. Kuffel, “Small Time-step (<2uSec) VSC Machine Drives for the
Real Time Digital Simulator,” presented at the 8th International Conference on Modeling and
Simulation of Electric Machines, Converters and System (Electromacs 2005), in Hammamet,
Tunisia, April 2005.
[98] S. Hui and C. Christopoulos, “A Discrete Approach to the Modeling of Power Electronic
Switching Networks”, IEEE Trans. on Power Electronics, Vol. 5, No. 4, pp. 398-403, Oct. 1990.
[99] P. Pejovic and D. Maksimovic, “A Method of Fast Time-Domain Simulation of Networks with
Switches”, IEEE Trans. on Power Electronics, Vol. 9, No. 4, pp. 449-456, July 1994.
[100] F. Plumier, P. Aristidou, C. Geuzaine and T. Van Cutsem, “Co-Simulation of Electromagnetic
Transients and Phasor Models: A Relaxation Approach,” IEEE Trans. on Power Delivery, Vol.
31, No. 5, pp. 2360 – 2369, 2016.
[101] V. Jalili-Marandi, V. Dinavahi, K. Strunz, J. A. Martinez and A. Ramirez, “Interfacing Techniques
for Transient Stability and Electromagnetic Transient Programs IEEE Task Force on Interfacing
Techniques for Simulation Tools,” IEEE Trans. on Power Delivery, Vol.24, No. 4, pp.2385 - 2395,
2009.
[102] O. Venjakob, S. Kubera, R. Hibberts-Caswell, P.A. Forsyth and T.L. Maguire, “Setup and
Performance of the Real-Time Simulator used for Hardware-in-Loop-Tests of a VSC-Based
HVDC scheme for Offshore Applications,” International Conference on Power Systems
Transients (IPST2013), 2013.
[103] M. Davies, M. Dommaschk, J. Dorn, J. Lang, D. Retzmann and D. Soerangr, “HVDC PLUS –
Basics and Principle of Operation,” Siemens Energy Sector, E T PS SL/DSoe/Re – 2008-08-10
– HVDC PLUS V3.
[104] O. Mo, “Average model of PWM converter,” Project memo AN 03.12.103, Sintef energy
research, 2003.
[105] J. G. Slootweg, S. W. H. de Haan, H. Polinder and W. L. Kling, “Modeling new generation and
storage technologies in power system dynamics simulations,” in IEEE PES Summer Meeting,
Vol. 2, pp. 868–873, 2002.
[106] URL: https://wiki.openelectrical.org/index.php?title=Power_Systems_Analysis_Software.
[107] CIGRE Task Force 38.01.10, “Modeling New Forms of Generation and Storage,” 2001.
[108] X. Mao and R. Ayyanar, “Average and Phasor Models of Single Phase PV Generators for
Analysis and Simulation of Large Power Distribution Systems,” 2009 Twenty-Fourth Annu. IEEE
Appl. Power Electron. Conf. Expo., pp. 1964–1970, Feb. 2009.
[109] A. Nagarajan and R. Ayyanar, “Dynamic phasor model of single-phase inverters for analysis
and simulation of large power distribution systems,” in 2013 4th IEEE International Symposium
on Power Electronics for Distributed Generation Systems, pp. 1–6, 2013.
[110] N. Pogaku, M. Prodanović and T. C. Green, “Modeling, analysis and testing of autonomous
operation of an inverter-based microgrid,” IEEE Trans. on Power Electron., Vol. 22, No. 2, pp.
613–625, 2007.
[111] A. Ellis, M. Behnke, and R. Elliott, “Generic Solar Photovoltaic System Dynamic Simulation
Model Specification,” Sandia National Laboratories, 2013.
[112] S. Winjnbergen, S. W. H. de Haan and J. G. Slootweg, “A System for Dispersed Generator
Participation in Voltage Control and Primary Frequency Control of the grid,” in IEEE 36th
Conference on Power Electronics Specialists, pp. 2918–2924, 2005.
[113] L. Ge, L. Zhu, L. Qu, L. Zhang, L. Zhao and N. Chen, “Model study of photovoltaic inverter for
bulk power system studies based on comprehensive tests,” proceedings of IET renewable
power generation 2015.
[114] K. Clark, R. A. Walling, and N. W. Miller, “Solar photovoltaic (PV) plant models in PSLF,” in
Proc. of IEEE PES GM 2011, Detroit, USA, 2011.
[115] K. Yamashita, K. Shirasaki and T. Ohno, “Dynamic Behavior of Photovoltaic Single-phase
Inverter Following Faults in Medium-voltage Networks Using CRIEPI's Power System Simulator,”
in Proc. of IEEE PES APPEEC 2013, Hong Kong, China. Dec. 2013.
[116] P. Aristidou and T. Van Cutsem, “A parallel processing approach to dynamic simulations of
combined transmission and distribution systems,” Int. J. Electr. Power & Energy Syst., Vol. 72,
pp. 58–65, 2015.
[117] G. Chaspierre, P. Panciatici, and T. Van Cutsem, “Dynamic Equivalent of a Distribution Grid
Hosting Dispersed Photovoltaic Units,” IREP’2017 Symposium: X Bulk Power Systems
Dynamics and Control Symposium, Espinho, 2017.
[118] Chinese National Standard Committee, “Guide for modeling PV power system,” GB/T 32826,
2016 (in Chinese).
[119] IEEE PES Energy Development and Power Generation Committee, “IEEE Std. 421.5-2005,
IEEE Recommended Practice for Excitation System Models for Power System Stability
Studies,” April 2006.

191
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

[120] JEA “Grid Interconnection Code,” JEAC 9701-2016, 2016 (in Japanese).
[121] J. Schlabbach, “Low voltage fault ride through criteria for grid connection of wind turbine
generators,” in Proc. of 5th International Conference on the European Electricity Market, 2008.
[122] Á. Ortega and F. Milano, “Generalized Model of VSC-based Energy Storage Systems for
Transient Stability Analysis,” IEEE Trans. on Power Systems, Vol. 31, No. 5, pp. 3369-3380,
2016.
[123] L. Schwartfeger and D. Santos-Martin, “Review of Distributed Generation Interconnection
Standards,” in EEA Conference & Exhibition 2014, 2014.
[124] K. Yamashita and H. Sato, “Small-scale Three-Phase Photovoltaic Inverter Model for Grid
Interconnection Studies,” in Proc. of CIGRE AORC Auckland New Zealand, Sep. 2017.
[125] K Tokumitsu, K Shirasaki and K. Yamashita, “Development of Photovoltaic Model for CPAT -
Modeling and Experimental Verification of Suppressive Function of Voltage Rise -,” in Proc. of
IEEJ Annual Conference of Power and Energy Society, Sep. 2012 (in Japanese).
[126] T. Neumann and I. Erlich, “Short Circuit Current Contribution of a Photovoltaic Power Plant,”
Institute of Electrical Power Systems, University Duisburg Essen, 2012, URL:
https://www.uni-due.de/ean/downloads/papers/neumann2012.pdf
[127] A. Ulbig, T. Rinke, S. Chatzivasileiadis and G. Andersson, "Predictive control for real-time
frequency regulation and rotational inertia provision in power systems," in 2013 IEEE 52nd
Annual Conference on Decision and Control (CDC), pp. 2946-2953, Dec. 2013.
[128] N. Miller, “Final Report: Technology Capabilities for Fast Frequency Response,” GE Report,
March 2017, URL: https://www.aemo.com.au/-
/media/Files/Electricity/NEM/Security_and_Reliability/Reports/2017-03-10-GE-FFR-Advisory-
Report-Final---2017-3-9.pdf
[129] D. Maksimovic, A. M. Stankovic, V. J. Thottuvelil and G. C. Verghese, “Modeling and simulation
of power electronic converters,” Proc. IEEE, Vol. 89, No. 6, pp. 898-912 Jun. 2001.
[130] E. van Ruitenbeek, J. C. Boemer, J. L. Rueda, M. Gibescu and M. A. van der Meijden, “A
Proposal for New Requirements for the Fault Behaviour of Distributed Generation Connected
to Low Voltage Networks,” in 4th International Workshop on Integration of Solar Power into
Power Systems, Berlin, Germany, 2014.
[131] J. C. Boemer, “On Stability of Sustainable Power Systems. Network Fault Response of
Transmission Systems with Very High Penetration of Distributed Generation,” Dissertation, Delft
University of Technology, June 2016, ISBN 978-94-6186-646-2.
[132] S. Altschäffl and R. Witzmann, “A modelling approach for dynamic short-circuit analysis of the
German power system considering all voltage levels,” in Die Energiewende - Blueprint for the
new energy age International ETG Congress 2015, Bonn, 2015.
[133] I. Maqbool, G. Lammert, A. Ishchenko and M. Braun, “Power System Model Reduction with
Grid-Connected Photovoltaic Systems Based on Hankel Norm Approximation,” 6th Solar
Integration Workshop, Vienna, Austria, Nov. 2016.
[134] NERC, “Reliability Guideline on Modeling Distributed Energy Resources in Dynamic Load
Models: NERC DER Modeling Guideline,” Dec. 2016, URL:
http://www.nerc.com/pa/RAPA/rg/Pages/Reliability-Guidelines.aspx
[135] WECC, “WECC Wind Power Plant Power Flow Modelling Guide,” May 2008, URL:
https://www.wecc.biz/_layouts/15/WopiFrame.aspx?sourcedoc=/Reliability/WECC/Wind/Plant/
Power/Flow/Modeling/Guide.pdf&action=default&DefaultItemOpen=1
[136] J. Langstädtler, A. Schnettler, A. Roehder and M. Mittelstaedt, “Investigation and Application of
Aggregated Wind Farm Models for Large Scale Power System Stability Analyses,” presented
at the 1. Conference for Wind Power Drives, Aachen, March, 2013.
[137] CIGRE TB 575, “Benchmark Systems for Network Integration of Renewable and Distributed
Energy Resources,” TF C6.04.02, Paris (21 rue d'Artois, 75008): CIGRÉ, 2014.
[138] E. Muljadi, C. P. Butterfield, A. Ellis, J. Mechenbier, J. Hochheimer, R. Young, N. Miller, R.
Delmerico, R. Zavadil and J. C. Smith, “Equivalencing the Collector System of a Large Wind
Power Plant,” in Proc. of IEEE PES GM 2006, Montreal, Quebec, June, 2006.
[139] E. Muljadi and A. Ellis. “WEEC Wind Generator Development. Final Project Report Appendix III
– Wind Power Plant Equivalencing (Appendix II - Typical Values of Collector System
Impedance),” NREL. March, 2010. URL: https://uc-ciee.org/downloads/WGM_Final_Report.pdf
[140] WECC, “WECC Steady State and Dynamic Data Requirements MOD-(11 and 13)-WECC-CRT-
1 Regional Criterion”, June 2013. URL
http://www.wecc.biz/library/Documentation/Categorization/Files/Regional/Criteria/MOD-
11/and/13-WECC-CRT-1.pdf
[141] NERC, “Distributed Energy Resources: Connection, Modeling and Reliability Considerations,”
NERC, Feb. 2017, URL,

192
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

http://www.nerc.com/comm/Other/essntlrlbltysrvcstskfrcDL/Distributed_Energy_Resources_R
eport.pdf
[142] WECC “WECC Data Preparation Manual for Steady State and Dynamic Base Case Data”,
System Review Work Group, Technical Studies Subcommittee, November 2012. URL:
http://www.wecc.biz/committees/StandingCommittees/PCC/TSS/SRWG/Shared/Documents/
WECC/Data/Preparation/Manual.pdf.
[143] CIGRE TB 527, “Performance Coping with Limits for Very High Penetrations of Renewable
Energy,” JWG C1/C2/C6.18 2013.
[144] VDE, “Power generation systems connected to low voltage distribution network -Technical
minimum requirements for the connection to and parallel operation with low-voltage distribution
networks,” VDE-AR-N 4105, Aug. 2011.
[145] O. Tsukamoto, “Grid Code Development in Japan,” Session 2 of IRED 2014, URL:
http://www.nedo.go.jp/english/ired2014/program/pdf/s2/s2_2_osami_tsukamoto.pdf.
[146] J. C. Boemer, E. Vittal, M. Rylander, B. Mather, “Determination of WECC Distributed PV System
Model Parameters for Partial Loss of Generation Due to Bulk System Voltage Sags” in Proc. of
IEEE PES GM 2017, Chicago July 2017.
[147] “The DER_A Model Proposal,” WECC memo, URL:
https://www.wecc.biz/Administrative/DER_A_Final_021518.pdf.
[148] IEC-61400-27-1:2015, Wind turbines - Part 27-1: Electrical simulation models - Wind turbines.
[149] M. Steurer, S. Woodruff, T. Baldwin, H. Boenig, F. Bogdan, T. Fikse, M. Sloderbeck, and G.
Snitchler, “Hardware-in-the-Loop Investigation of Rotor Heating in a 5 MW HTS Propulsion
Motor,” IEEE Trans. on Applied Superconductivity, Vol. 17, No. 2, Part 2, pp. 1595-1598, 2007.
[150] J. Langston, M. Steurer, K. Schoder, J. Hauer, F. Bogdan, I. Leonard, T. Chiocchio, M.
Sloderbeck, A. Farrell, J. Vaidya, and K. Yost, “Megawatt Scale Hardware-in-the-Loop Testing
of a High Speed Generator,” in Proc. ASNE Day, May 9-10, 2012, Arlington, VA.
[151] C. Schacherer, J. Langston, M. Steurer, and M. Noe, “Power Hardware-in-the-Loop Testing of
a YBCO Coated Conductor Fault Current Limiting Module,” IEEE Trans. on Applied
Superconductivity, Volume 19, No. 3, Part 2, June 2009 Page(s):1801 – 1805.
[152] T. Chiocchio, R. Leonard, Y. Work, R. Fang, M. Steurer, A. Monti, J. Khan, J. Ordonez, M.
Sloderbeck, and S. L. Woodruff, “A Co-Simulation Approach for Real-Time Transient Analysis
of Electro-Thermal System Interactions on Board of Future All-Electric Ships,” in Proc. of
Summer Computer Simulation Conference, July 15-18, 2007.
[153] M. O. Faruque, M. Sloderbeck, M. Steurer, and V. Dinavahi, “Thermo-electric co-simulation on
geographically distributed real-time simulators,” in Proc. of IEEE PES GM, Calgary, AB, Canada,
July, 2009.
[154] IEC 61400-21-1 Ed. 1: “Measurement and assessment of electrical characteristics -Part 1 -Wind
Turbines,” CDV as of July 2017.
[155] IEC 61400-27-2 Ed. 1: “Wind energy generation systems – Part 27-2: Electrical simulation
models – Model validation,” CDV as of Feb. 2017.
[156] FGW, “Technical guidelines for power generating Units and Systems Part 4, -Demands on
Modelling and Validating Simulation Models of the Electrical Characteristics of Power
Generating Units and Systems,” FGW -TR4, Rev-08, Jan. 2016.
[157] L. Qu, L. Zhu, L. Ge, and M. Sun, “Research on multi-time scale modelling of photovoltaic power
plant,” in proc. of IET renewable power generation 2015, Beijing, 2015.

193
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

194
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX C. EXAMPLE STUDIES

APPENDIX 1-A INTERNATIONAL INDUSTRY PRACTICE ON MODELLING AND
DYNAMIC PERFORMANCE OF INVERTER BASED GENERATION IN
POWER SYSTEM STUDIES
Chapter 1-A.1 Introduction
Presently, many regions around the world are seeing ever increasing penetration of Inverter Based
Generation (IBG), such as type 3 and 4 wind generation systems and PhotoVoltaic (PV) generation.
The dynamic characteristic of IBG is different from conventional synchro-nous generators, and so this
ever-increasing penetration of IBG means that the dynamic performance of the power system following
disturbances may change. Therefore, analyses of the high penetration levels of IBG require much focus
on the type of the IBG models being used in power system dynamic studies.
There are many levels of models used for all types of power equipment. At one end of the spectrum are
stability models, sometimes also referred to as positive sequence or Root Mean Square (RMS) models.
At the other end of the spectrum are very detailed equipment level Electro-Magnetic Transient (EMT)
based models. However, since all models have limitations, the selection of the model type is crucial
based on the objectives of the study to be performed with the model. Therefore, the adequate model
type could change depending on the type of power system dynamic study and the system conditions to
be studied.
Static load models are still widely used around the world, except in North America [1], where reliability
standards now mandate that dynamic load models be used at least when studying heavy summer
loading conditions [2]. Thus, because most of the PV generation in many regions around the world is
connected to the distribution level, such as medium and low voltage networks, it is more likely that
negative static load models are being used in many countries for the representation of IBG.
RMS models for IBG have been developed over the last years for generation connected at the bulk
electric system level that is, at the transmission level. Two industry working groups were established,
one within the Western Electricity Coordinating Council (WECC) [3] in North America, and another within
the International Electrotechnical Commission (IEC) [4], in order to develop generic models of different
types of IBG for power system dynamic studies. Some of those models have been already implemented
in widely used commercial power system analysis software tools [5], [6]. However, these generic models
are still not widely used yet by the industry, especially in Europe, as they are still relatively new. With
respect to IBG, connected at the medium and low voltage distribution level (e.g. residential PV), there
are still no widely accepted aggregated dynamic models. In this context, the most recent work has been
done by WECC [7], but it is presently under review and discussion for further changes. Some of this is
driven by new grid code requirements that need to be considered in the development of generic models
for IBG, e.g., the changing IEEE 1547 standard in North America or the newly proposed German grid
code for the high voltage network VDE-AR-N 4120. This might be another reason that generic models
for IBG are not widely used yet.
In October 2013, CIGRE and CIRED established the new Joint Working Group (JWG) CI-GRE
C4/C6.35/CIRED: “Modelling and dynamic performance of inverter-based generation in power system
transmission and distribution studies”, to look at some of these evolving issues. One of the tasks of the
JWG is to identify the present industry practice on modelling of IBG, with the focus on PV systems, for
power system dynamic studies. For that purpose, a comprehensive questionnaire was developed and
distributed during the spring of 2015 to 63 utilities and system operators in 21 countries on five
continents. This Sub-Chapter summarizes some of the key findings from 45 responses to the survey,
which have been received by the summer of 2016.
The aim of this Sub-Chapter is not necessarily to recommend the application of any specific dynamic
model for a specific power system dynamic study, but, rather, to identify what dynamic models are
presently applied and to provide some fundamental information on their use. Based on the key findings
and observations, this Sub-Chapter states a clear message for the necessity and importance of the use
of IBG models.

195
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Chapter 1-A.2 Survey

Chapter 1-A.2.1 Organization
The original questionnaire that was distributed consists of four categories, whereas this Sub-Chapter
focuses on the results of the two main categories. The main categories and the corresponding questions
are listed in Table 1-A- I. Further results of the original questionnaire can be found in [8].
Category 1 is about the characterization of the studied power system and gives a basic, but
comprehensive, overview of the analysed grids. The participants of the survey describe the overall
system they study, e.g., the system connectivity, the installed capacity of their generation and the
highest percentage of the penetration level of IBG.
Category 2 describes the type of model that is used by the utility and system operator for a specific type
of power system dynamic study. In this category two types of models are distinguished, namely RMS
and EMT models. A detailed comparison between RMS and EMT models is given in Chapter 1-A.3.
Furthermore, category 2 refers to 14 different types of studies, including well known phenomena such
as frequency, voltage, transient and small-signal stability. In this context, it should be noted that the
evaluation of voltage fluctuations at steady state is not considered as long-term voltage stability.
Table 1-A-1: Survey categories and questions

No. Category No. Question

1.1 What system is the model intended to be used for
(transmission/distribution/both)?
1.2 What is the system connectivity (isolated small
grid/interconnected grid)?

Characterization of the 1.3 What is the main purpose of the power system dynamic
1 study?
studied power system
1.4 What is the total generation capacity?
1.5 What is the total inverter-based generation capacity?
1.6 What is the highest percentage of penetration level of
inverter-based generation?
What type of model do you use for a specific type of
power system dynamic study?
2.1 Frequency stability (large system)
2.2 Short-term voltage stability (seconds)
2.3 Short-circuit provision from inverter-based generation
2.4 Low voltage ride-through
2.5 High voltage ride-through
Type of model used for a 2.6 Transient stability with balanced faults
2 specific type of power
2.7 Transient stability with unbalanced faults
system dynamic study
2.8 Long-term voltage stability (minutes)
2.9 Small-disturbance angle stability
2.10 Unintentional islanding
2.11 Transients including switching transients
2.12 Control system interactions (high frequency)
2.13 Control system interactions (low frequency)
2.14 Protection coordination

196
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Chapter 1-A-2.2 Participants

Potential survey participants were identified by the JWG members and contacted via e-mail. The
considered utilities and system operators were either Transmission System Operators (TSOs), or utilities
and system operators that operate both, transmission and distribution systems. However, Distribution
System Operators (DSOs) were not considered for this Sub-Chapter, because most of them replied that
they do not perform power system dynamic studies.
As depicted in Table 1-A- II, the questionnaire was sent to 63 utilities and system operators around the
world between spring 2015 and summer 2016. Out of these 63 contacted utilities and system operators,
45 replied to the JWG. Hence, the response rate of 71% was reached. The 45 received questionnaires
from utilities and system operators came from 21 countries on five continents.
Table 1-A-2: Survey participants and response rate

Participants Total
Sent questionnaires to utilities and system operators 63
Received questionnaires from utilities and system
45
operators
Response rate from utilities and system operators [%] 71
Received questionnaires from continents 5
Received questionnaires from countries 21

It should be noted that software vendors, consultancies, research organizations and academia are out
of scope and only responses from utilities and system operators were considered for the survey. Out of
the 45 received questionnaires, 35 came from TSOs, and 10 came from utilities and system operators
that operate both, transmission and distribution systems.
Chapter 1-A.3 Type of models
Chapter 1-A-3.1 RMS models
RMS models are mainly used to study power system stability for large interconnected systems, this
includes electromechanical oscillations (small-signal stability), as well as rotor-angle stability of
synchronous generators and voltage and frequency stability [9]. Phasor simulation methods are used
when only the magnitude and phase of the voltages and currents are of interest. It is not necessary to
solve all the differential equations resulting from the interaction of R, L, and C elements. Hence, the
network is simulated with fixed complex impedances in-stead of differential equations [10]. RMS
simulation tools consider phenomena with a band-width of typically 0.1 to 3 Hz, since the network
model’s fidelity diminishes rapidly for phenomena with frequencies significantly outside of this range.
However, the control loops modelled on individual power plants may cover phenomena up to 10 Hz.
Another commonly used terminology for RMS type models is to call them “positive-sequence stability”
models. In essence, such models assume a perfectly balanced network and consider only the positive
sequence components of all phenomena.
Converters are included in RMS programs using their averaged models. They provide the behaviour of
the converter ignoring fast switching transients and any control with very small-time constants compared
to the time steps and the phenomena considered. For instance, in stability analysis there is usually no
need to model the inverter in detail since its transients are much faster than the dynamics being studied.
Therefore, only the fundamental frequency outputs of the converter are modelled, which are mainly
reflected in the electrical control model. Due to this averaged representation, the modelling of different
technologies of IBG can be unified using sub-models for each basic component according to their
characteristics.
Chapter 1-A-3.2 EMT models
When the phenomena to be studied are significantly outside of the range of fidelity of RMS models (for
example studying electro-magnetic transients, or sub-synchronous torsional interactions, etc.), then
EMT simulation software should be used, together with detailed equipment specific models. EMT
analysis programs solve the differential-algebraic equations of a three-phase electrical network [11].
This distinction means that EMT analysis is capable of representing electro-magnetic transients (hence
the name EMT), the frequency dependence of network components (e.g. change in transmission line

197
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

resistance/reactance with frequency), harmonics, unbalanced networks, power electronic devices,

including the switching transients as well as the detailed controls and protection systems.
Users need to be careful on the selection of the type of models between RMS and EMT. RMS models
usually do not include the inner current control loops of the converter, the detailed phase-locked loop
model and various other details of the converter. Nevertheless, RMS simulations are much faster to
execute than EMT and are thus dominant for specific types of power system dynamic studies.
EMT models may also be required under the following special system conditions (where RMS time-
domain simulation programs may not be accurate or suitable anymore):
 Weak system conditions (with a very low short-circuit ratio)
 Detailed inverter and collector system design
 Detailed equipment and system interaction studies
 Unbalanced faults (note that many RMS models are positive sequence models)

Chapter 1-A.4 results

The survey results are introduced for each question in this Sub-Chapter. Most of the survey results
include the mean value inserted at the bottom of each table. It is noted that such mean value is an
invented index that makes no real sense since each type of study weights the same but allows us to
discover which model is more likely to be selected.
Chapter 1-A-4.1 Category 1: Characterization of the studied power system
Question 1.1 (What system is the model intended to be used for (transmission/distribution/both)?) in this
category is about the type of power system that is studied. For this Sub-Chapter only utilities and system
operators that perform studies with either the transmission system only, or with both, the transmission
and the distribution system, were considered. Out of the considered participants, 78% perform
transmission system studies only, whereas 22% per-form studies for both, the transmission and the
distribution system.
Question 1.2 (What is the system connectivity (isolated small grid/interconnected grid)?) analyses the
system connectivity. Out of the considered participants, 87% of the utilities and system operators
operate an interconnected grid. Furthermore, there are 9% of the utilities and system operators, which
operate both, an interconnected and an isolated small grid. This is the case for countries that also consist
of islands. Just 4% of the survey participants perform studies with an isolated small grid only.
Question 1.3 (What is the main purpose of the power system dynamic study?) in this category is about
the predominant studies that utilities and system operators usually perform and the results are shown
in Figure 1-A-1. The participants could select between different types of studies given in category 2 in
Table 1-A- I. The results indicate that the utilities and system operators mainly perform stability studies,
such as frequency stability (study 2.1) with 89%, short-term volt-age stability (study 2.2) with 89% and
transient stability with balanced faults (study 2.6) with 89%. It should be noted that recently announced
grid code requirements for IBG, like low voltage ride-through (study 2.4), are also very common with
82%. Other power system dynamic studies, such as transients including switching transients (study
2.11) with 40%, control system interactions (studies 2.12 and 2.13) for high frequencies with 44% and
low frequencies with 47% as well as unintentional islanding (study 2.10) with 53% are less commonly
performed by the utilities and system operators. However, even for the less commonly performed studies
it should be noted that almost every second utility and system operator performs these types of studies.
The main observation of the results is that 69% of the participants perform the listed power system
dynamic studies (studies 2.1 to 2.14), as seen in the last row of Figure 1-A-1.

198
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

All 2.14 2.13 2.12 2.11 2.10 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1
Frequency stability (large system) 89

Short-term voltage stability (seconds) 89

Short-circuit provision from IBG 78

Purpose of the power system dynamic study

Low voltage ride-through 82

High voltage ride-through 76

Transient stability with balanced faults 89

Transient stability with unbalanced faults 76

Long-term voltage stability (minutes) 69

Small-disturbance angle stability 71

Unintentional islanding 53

Transients including switching transients 40

Control system interactions (high frequency) 44

Control system interactions (low frequency) 47

Protection coordination 62

Average value of studies 2.1 to 2.14 69

0 10 20 30 40 50 60 70 80 90 100
Responses of the participants [%]

Figure 1-A-1: Question 1.3: What is the main purpose of the power system dynamic study?
Question 1.4 (What is the total generation capacity?) analyses the size of the system regarding the
installed generation capacity. The average value of the system sizes of all of the utilities and system
operators is about 50 GW. The system size starts with 84 MW grids and ranges over large-scale power
systems with 780 GW.
Question 1.5 (What is the total inverter-based generation capacity?) is about the amount of installed
IBG. The average value of installed IBG of all of the participants is about 5 GW, which is 10% of the
average value of the total installed generation capacity. However, there are also utilities and system
operators with no IBG and a few with a high amount of IBG of about 44 GW.
Question 1.6 (What is the highest percentage of penetration level of inverter-based generation?)
investigates the penetration level. The results to this question are depicted in Figure 1-A-2 as a box plot.
The present penetration level is calculated considering the values given by questions 1.4 and 1.5, and
therefore represents the installed capacity. The main characteristics of the box plot shown in Figure 1-
A-2 are: the median, shown as the black bar inside the grey box; the 25%- and 75%-quantile,
represented as the top and the bottom of the grey box; and the min. and max. value, depicted as the
whisker. The results can be interpreted as follows. The median of all of the utilities and system operators
reaches 12%. The min. and max. values are 0% and 57%, respectively. This means there are some of
the utilities and system operators with no IBG, and some of the utilities and system operators reach a
high share of IBG with 57%.

199
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

60%

50%

40%

30%

20%

10%

0%
Present penetration level
(installed capacity)

Figure 1-A-2: Question 1.6: What is the highest percentage of penetration level of inverter-based
generation?

Chapter 1-A-4.2 Category 2: Type of model used for a specific type of power system
dynamic study
Category 2 is about the type of model that is used for a specific type of power system dynamic study.
The results for this category are divided into three parts. Part 1 analyses whether an IBG model or a
negative load model is used. If an IBG model is applied, a further distinction is made between RMS and
EMT model, which is investigated in part 2. Part 3 is similar to part 2 while the focus is on the expected
application of RMS and EMT models in the future. For all of the three parts, the answers were separated
into 14 different power system dynamic studies, as shown in Table 1-A-1. The results are discussed
below.
Part 1 (distinction between IBG model and negative load model) describes a fundamental decision by
the utilities and system operators, whether they represent IBG with an associated model or with a
negative load model. The results of this distinction are depicted in Figure 1-A-3. It can be seen that for
13 out of 14 studies the participants decide for the IBG model instead of the negative load model. Only
for transients including switching transients (study 2.11) the participants prefer the negative load model.
Furthermore, the studies 2.1 to 2.7 in Figure 1-A-3 show that the application of IBG models is
predominant (about 75% in average) compared to the negative load model (about 25% in average). For
the studies 2.8 to 2.14 except 2.11 the share of the negative load model increases, whereas for study
2.11 it exceeds the IBG model. The main finding of the results is that one third of the utilities and system
operators still use the negative load model for power system dynamic studies, as seen in the last row of
Figure 1-A-3.

200
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

IBG model Negative load model

All 2.14 2.13 2.12 2.11 2.10 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1 Frequency stability (large system) 71 29
Short-term voltage stability (seconds) 70 30
Short-circuit provision from IBG 72 28
Low voltage ride-through 79 21
Type of power system dynamic study

High voltage ride-through 77 23

Transient stability with balanced faults 75 25
Transient stability with unbalanced faults 71 29
Long-term voltage stability (minutes) 65 35
Small-disturbance angle stability 61 39
Unintentional islanding 60 40
Transients including switching transients 47 53
Control system interactions (high frequency) 60 40
Control system interactions (low frequency) 52 48
Protection coordination 54 46
Average value of studies 2.1 to 2.14 65 35

0 10 20 30 40 50 60 70 80 90 100
Responses of the participants [%]

Figure 1-A-3: Question 2: What type of model do you use for a specific type of power system
dynamic study? Distinction between Inverter Based Generation (IBG) model and negative load
model.
Part 2 (distinction between RMS and EMT model) deals with the decision of utilities and system
operators with the type of IBG model that is used for a specific type of power system dynamic study.
The results of the comparison between RMS and EMT models are presented in Figure 1-A-4. In general,
it can be concluded that in the studies 2.1 to 2.10 and 2.13 as well as 2.14 RMS models are predominant.
Only for transients including switching transients (study 2.11) utilities and system operators prefer the
EMT model instead. For control system interactions (high frequency) (study 2.12) the participants apply
equally both, RMS and EMT models with 50%, respectively. It should be noted that for stability studies,
such as frequency stability (study 2.1), voltage stability (studies 2.2 and 2.8) and rotor angle stability
(studies 2.6, 2.7 and 2.9), RMS models are widely (about 90% in average) used by the utilities and
system operators. Furthermore, for power system dynamic studies that analyse transients, like
unintentional islanding (study 2.10), transients including switching transients (study 2.11), control
system interactions (high and low frequency) (studies 2.12 and 2.13) as well as protection co-ordination
(study 2.14), the share of EMT models used by the utilities and system operators is considerable
increased. The main observation of the results is that 78% of the participants apply RMS models (if they
use IBG models at all (refer to Figure 1-A-3)) instead of EMT models for power system dynamic studies,
as seen in the last row of Figure 1-A-4.

201
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

RMS model EMT model

All 2.14 2.13 2.12 2.11 2.10 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1 Frequency stability (large system) 91 9
Short-term voltage stability (seconds) 88 13
Short-circuit provision from IBG 81 19
Low voltage ride-through 83 17
Type of power system dynamic study

High voltage ride-through 84 16

Transient stability with balanced faults 91 9
Transient stability with unbalanced faults 85 15
Long-term voltage stability (minutes) 95 5
Small-disturbance angle stability 95 5
Unintentional islanding 74 26
Transients including switching transients 38 62
Control system interactions (high frequency) 50 50
Control system interactions (low frequency) 62 38
Protection coordination 70 30
Average value of studies 2.1 to 2.14 78 22

0 10 20 30 40 50 60 70 80 90 100
Responses of the participants [%]

Figure 1-A-4: Question 2: What type of model do you use for a specific type of power system
dynamic study? Distinction between Root Mean Square (RMS) and Electro-Magnetic Transient
(EMT) model.
Part 3 (distinction between RMS and EMT model in the future) analyses in the same way as part 2
(distinction between RMS and EMT model) the type of IBG model that is used by the utility and system
operator, with the difference that part 3 is focused on the expected application of RMS and EMT models
in the future. The results are shown in Figure 1-A-5. At the first sight, it can be seen that RMS models
are still prevailing compared to EMT models. The majority of the participants use RMS models for 11
out of 14 power system dynamic studies. Only for transients including switching transients (study 2.11)
as well as control system inter-actions (high and low frequency) (studies 2.12 and 2.13) EMT models
are predominant. It should be noted that for stability studies, such as frequency stability (study 2.1),
voltage stability (studies 2.2 and 2.8) and rotor angle stability (studies 2.6, 2.7 and 2.9), RMS models
are still widely used (about 75% in average) by the utilities and system operators. Furthermore, similar
to Figure 1-A-4, for power system dynamic studies that analyse transients, like unintentional-al islanding
(study 2.10), transients including switching transients (study 2.11), control system interactions (high and
low frequency) (studies 2.12 and 2.13) as well as protection coordination (study 2.14), the share of
expected EMT models in the future used by the participants is considerable increased. The main finding
by analysing the results is that 63% of the participants apply RMS models instead of EMT models for
power system dynamic studies in the future, as seen in the last row of Figure 1-A-5. By comparing the
results, the last row of Figure 1-A-5 with the last row Figure 1-A-4, it is important to mention that the
share of EMT models used for power system dynamic studies will increase in the future from 22% to
37%.

202
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

RMS model in future EMT model in future

All 2.14 2.13 2.12 2.11 2.10 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1 Frequency stability (large system) 78 22
Short-term voltage stability (seconds) 72 28
Short-circuit provision from IBG 65 35
Low voltage ride-through 65 35
Type of power system dynamic study

High voltage ride-through 68 33

Transient stability with balanced faults 74 26
Transient stability with unbalanced faults 73 27
Long-term voltage stability (minutes) 81 19
Small-disturbance angle stability 76 24
Unintentional islanding 57 43
Transients including switching transients 32 68
Control system interactions (high frequency) 46 54
Control system interactions (low frequency) 38 62
Protection coordination 58 42
Average value of studies 2.1 to 2.14 63 37

0 10 20 30 40 50 60 70 80 90 100
Responses of the participants [%]

Figure 1-A-5: Question 2: What type of model do you use for a specific type of power system
dynamic study? Distinction between Root Mean Square (RMS) and Electro-Magnetic Transient
(EMT) model in the future.

Chapter 1-A.5 discussion

Chapter 1-A-5.1 negative load model
From the results it is remarkable that still around one third of the utilities and system operators use
negative load models for the representation of IBG in power system dynamic studies. According to the
results of the questionnaire survey, the following reasons may explain this approach:
 Lack of model requirements of IBG for specific power system phenomena:

As the penetration level of such IBG technologies increases, various aspects of power system stability
and dynamic performance in the grid may change. Therefore, requirements that address the necessary
functionalities that need to be modelled of IBG for specific power system phenomena need to be
developed. These functionalities include various aspects, such as control, protection and the capability
of IBG. Considering these requirements, utilities and system operators can select specific models for
each power system phenomenon.
 Lack of well-validated IBG models:

In recent years there has been much effort on the development of validated models for IBG. This work
has been primarily related to wind generation. Now, further attention is starting to be devoted to PV
systems and other technologies. In general, there is still a lack of well-validated and generally accepted
dynamic simulation models, particularly for distributed PV systems, for the use in power system dynamic
studies.
 Lack of widely accepted generic models for IBG:

203
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Usually utilities and system operators do not create their own (user-written) models. They request
validated models from manufacturers, either proprietary or adjusted generic models. This request poses
two main disadvantages: 1) the manufacturer wants to keep the confidentiality of their proprietary user-
written model; and 2) the extra effort for the manufacturer for tuning the parameters of the generic model
including the validation of the simulations against the field measurements. Thus, the importance of
developing reliable and flexible generic models for different technologies and manufacturers of IBG
should be noted. Advantages of generic models are: vendor and manufacturer independent, grid code
compatible, public model structure (control block diagram), software simulation tool independent, etc.
For some technologies, like wind generation, these models are already being widely used, however, the
latest generic models had only recently been developed at the time the questionnaire survey was
conducted.
 Lack of widely accepted range of model parameters for IBG:

Although widely accepted generic models are provided, the control model parameters are crucial for
power system dynamic studies. Because many grid codes do not define the detailed
specification/characteristics of the inverter control, the control model parameters could be different
depending on the manufacturer of the inverter. Even if the control model parameters of one inverter can
be identified through validation, it is al-most infeasible to identify the parameters of all inverters
connected to the power system. Therefore, a set of realistic control model parameters need to be
provided.
 Lack of grid code requirements:

Due to the lack of grid code requirements in the past, specifying detailed control functionalities of IBG,
the approach of using negative load models for power system dynamic studies was justified. However,
with the development of new grid codes, certain functionalities of IBG are required (e.g. voltage control,
frequency response etc.) and therefore, the negative load model is not adequate anymore.
 Lack of information about the power system:

The aforementioned increased penetration level of IBG also makes system operation, both for TSOs
and DSOs, more challenging than in the past. Already, in some areas the consumers’ demand is mostly
covered by generation which is connected directly to the distribution system. TSOs routinely run time-
domain simulations to assess the stability of the power system. Models, which are currently used to
represent distribution systems, are only based on a limited amount of information, generally related to
the high voltage network.
 Lack of agreed methodology for the aggregation of IBG:

Present trends towards the integration of an increasing range of IBG technologies, widely differing in
size and number, poses serious concerns in the industry on how to represent these new technologies
in power system dynamic studies. There is not only a lack of validated dynamic computer models of
individual distributed generating technologies, such as distributed PV systems, fuel cells, micro turbines
etc., but also there is no agreed methodology on how to represent or aggregate the enormous number
of distributed generation, embedded in very low voltage grids, for power system dynamic studies,
focusing on both, local (distribution level) and widespread (transmission level) studies.
Chapter 1-A-5.1 Necessity of IBG models
The high penetration level of IBG has resulted in the displacement of conventional synchro-nous
generators. Therefore, the impact of IBG on the dynamic performance of the system in-creases. The
dynamic characteristic of IBG is different compared to synchronous generators, and with proper control
system design and functionalities of modern IBG technologies, they can provide many of the same or
even better services (e.g. voltage control, frequency response etc.). None the less, they do need to be
modelled differently and properly. Therefore, the development of the proper computer simulation models
for IBG with such additional functionalities is vital for power systems analyses.
Moreover, DSOs have a representation of their networks and have details about connected consumers
and producers to some extent. However, the limited data is generally not suitable for dynamic
simulations for either the distribution or the transmission system or, at least, has not been used for that
purpose in the past, due to the high level of detail. From the point of view of the DSOs, time-domain
simulations may also now be necessary to assess, e.g., protection system behaviour, distribution
network automated operation, unintentional islanding of part of distribution systems including IBG,
voltage issues, etc. For these types of power system dynamic studies detailed IBG models are needed.

204
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Therefore, the necessity of IBG models should be clarified for each type of power system dynamic study.
A few examples are given as follows:
 Frequency stability (refer to study 2.1 in Table 1-A- I):

Frequency stability studies often involve looking at the frequency response of the grid to a large
disturbance, such as the loss of the largest generating unit (or facility) on the system and assessing if
the resulting frequency response of the system is stable and avoids under frequency load shedding.
Such studies are typically performed using RMS (positive-sequence) simulation models and tools.
 Short-circuit provision from IBG (refer to study 2.3 in Table 1-A- I):

Short-circuit studies are performed, typically by protection engineers, to identify the setting for protection
relays as well as to assess the capability of existing (or new) circuit breakers to be able to withstand and
interrupt the short-circuit levels that will be seen on the network during fault conditions. Such studies are
typically performed in EMT tools and other software tools specifically designed for short-circuit studies.
Thus, appropriate models of IBG are needed in such tools.
 Low voltage ride-through (refer to study 2.4 in Table 1-A- I):

To properly assess the actual low voltage ride-through capabilities of IBG, analyses have to be
performed using detail vendor specific EMT type models and simulations. Furthermore, such simulations
may be verified by factory tests of the equipment. It usually then suffices to translate the observed
performance to simplified RMS models that will mimic the low voltage ride-through capabilities of the
equipment for large-scale power system dynamic studies. Thus, both appropriate RMS and EMT models
of IBG are needed.
 Transient stability (refer to study 2.6 in Table 1-A- I):

For bulk power system transient stability studies validated RMS (or positive-sequence) models are
needed for IBG. However, in some cases EMT models may also be needed when investigating, e.g.,
the connection of a large IBG power plant to a very weak part of the power system.
 Long-term voltage stability (refer to study 2.8 in Table 1-A- I):

Long-term voltage stability is often analysed using continuous power flow analysis, such as P-V or Q-V
analyses. For such studies, the dynamics of IBG are often neglected, and a quasi-steady-state model
that respects the real and reactive power limits of the IBG is sufficient. If mid-term time-domain dynamic
simulations are performed, then a suitable RMS (positive-sequence) IBG model is needed.
 Unintentional islanding (refer to study 2.10 in Table 1-A- I):

For this type of study, either an RMS or EMT model for IBG is needed. Some anti-islanding protection
systems utilize the harmonics of the voltage for the islanding detection and therefore, the EMT model
for IBG is the only option to analyse unintentional islanding. On the other hand, if the used anti-islanding
protection system does not consider voltage harmonics and if the power systems model includes many
IBG models, the RMS model for IBG may be adequate to analyse unintentional islanding.
 Protection coordination (refer to study 2.14 in Table 1-A- I):

Many protection systems need to be coordinated. Depending on the type of protection systems to be
studied, either an RMS or EMT model for IBG is needed.
Chapter 1-A.6 conclusion
The aim of this Sub-Chapter is not necessarily to recommend the application of any dynamic model for
a specific power system dynamic study, but, rather, to identify what dynamic models are presently
applied and to provide some fundamental information on their use.
The main contributions and key findings of this Sub-Chapter are:
 Prevalent type of power system dynamic studies:

The most dominant type of studies performed by TSOs are stability studies, such as frequency stability,
short-term voltage stability and transient stability with balanced faults. This is done by 89% of those who
responded to the survey.
 Prevalent type of model (IBG/negative load) for power system dynamic studies:

205
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

around one third of the utilities and system operators still use a negative load model for power system
dynamic studies.
 Prevalent type of model (RMS/EMT) for power system dynamic studies:

78% of the utilities and system operators apply RMS models instead of EMT models for power system
dynamic studies. EMT models are more likely to be used for various high-frequency transient studies
(e.g. switching transients).
 Prevalent type of model (RMS/EMT) in the future for power system dynamic studies:

63% of the utilities and system operators expect to continue to apply RMS models in-stead of EMT
models in the future for power system dynamic studies. It is important to mention that the share of EMT
models used for power system dynamic studies will increase in the future from 22% to 37%.
Based on the results of the questionnaire, the following reasons for the approach of the negative load
model are identified:
 Lack of model requirements of IBG for specific power system phenomena
 Lack of well-validated IBG models
 Lack of widely accepted generic models for IBG
 Lack of widely accepted range of model parameters for IBG
 Lack of grid code requirements
 Lack of information about the power system
 Lack of agreed methodology for the aggregation of distributed IBG

Some reasons have been resolved mainly for models of wind generation. The IEC 61400-27-1 and
61400-27-2 standards under development are presently in the process of providing more refined generic
wind turbine and generic wind power plant models and the procedures for validating those models. A
standard range of parameters, similar to WECC [12], is also expected to be illustrated in the future IEC
documents. The methodology for the aggregation of distributed IBG has been discussed in the WECC
and is presently under review [7]. The second-generation generic renewable energy system models,
developed in the WECC, as well as the generic wind turbine models, developed in the IEC, are now
implemented by several commercial software vendors and utilities and system operators start using
these models for power system dynamic studies with IBG. None the less, much learning as well as
technical challenges still remain, such as the methodology of the aggregation of distributed IBG
considering the diversified control parameters, and therefore, further effort is needed.
In light of the observations of the survey, the necessity of the IBG model for the several representative
power system dynamic studies is discussed highlighting the need of IBG models. Furthermore, the
recommended type of model is also emphasized providing the approach of RMS or EMT model for IBG.
It can be concluded that every type of model has advantages and disadvantages and the proper model
type needs to be selected depending on the type of power system dynamic study and the system
condition.
The results of the questionnaire emphasize the clear message for the necessity and importance of the
use of IBG models. Furthermore, the final technical brochure of the CIGRE JWG C4/C6.35/CIRED will
give guidance in selecting adequate models for IBG for specific power system dynamic studies.
With these contributions the Sub-Chapter supports utilities and system operators as well as research
institutes and academia to benchmark their approach against the prevailing international industry
practice.

206
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 1-B MODELLING AND DYNAMIC PERFORMANCE OF INVERTER-BASED

GENERATION IN POWER SYSTEM STUDIES: AN INTERNATIONAL
QUESTIONNAIRE SURVEY
Chapter 1-B.1 Introduction
Nowadays, the electrical power system is undergoing fundamental changes due to the ever-increasing
penetration of IBG. The dynamic characteristic of IBG, such as wind or PV generation, is fundamentally
different from conventional synchronous generators, and thus the dynamic performance of the system
following disturbances might change due to the increasing impact of IBG. Therefore, studies are required
that analyse the impact of high penetration levels of IBG on the dynamic performance of the system. In
this context, the selection of the IBG model type is crucial for the analysis.
There are many levels of models used for all types of power system components. On the one hand,
there are detailed models of IBG, also referred to as individual models. On the other hand, there are
simplified models that represent an aggregation of IBG, also referred to as aggregated models.
However, since all models have limitations, the selection of the model type is crucial based on the
objectives of the study.
Different types of models for IBG have been developed over the last years. Two industry working groups
were established, one within the Western Electricity Coordinating Council (WECC) [3], and another
within the International Electrotechnical Commission (IEC) [4], in order to develop generic models of
different types of IBG. Some of those models have been already implemented in widely used commercial
power system analysis software tools [5], [6]. However, these generic models are still not widely used
yet by the industry.
In 2013, the new Joint Working Group (JWG) CIGRE C4/C6.35/CIRED: “Modelling and dynamic
performance of inverter-based generation in power system transmission and distribution studies” was
established, to look at some of these evolving issues. One of the tasks of the JWG is to identify the
present industry practice on modelling of IBG, with the focus on PV generation. For that purpose, a
comprehensive questionnaire was distributed to utilities and system operators around the world. This
report summarizes some of the key findings.
The aim of this report is not necessarily to recommend the application of any specific dynamic model
for a specific power system dynamic study, but, rather, to identify what dynamic models are presently
applied and to provide some real pictures on their application.
Chapter 1-B.2 Survey
Chapter 1-B-2.1 Organization
The original questionnaire that was distributed to the utilities and system operators consists of four
categories, whereas this report focuses on the results of two of them. The two categories and the
corresponding questions to each of them are listed in Table I. Further findings and observations of the
original questionnaire can be found in [13].
Category 1 sorts the IBG technology and its modelling by type. The participants of the survey could
select between different technologies, such as wind or PV generation, micro turbines, fuel cells or battery
energy systems. This category reveals which type of IBG technology is modelled by utilities and system
operators for power system dynamic studies.
Category 2 describes the type of model that is presently used by the utility and system operator for a
specific type of power system dynamic study. In this category two types of models are classified, namely
individual models and aggregated models. A detailed comparison between individual and aggregated
models is provided in Chapter 1-B.3, titled: “Type of models”. Furthermore, category 2 refers to 14
different types of power system dynamic studies, including well known phenomena such as frequency
stability, short-term and long-term voltage stability as well as transient and small-disturbance angle
stability.
Chapter 1-B-2.2 Participants
Potential survey participants were selected by the JWG members and contacted via e-mail. The
considered utilities and system operators were either Transmission System Operators (TSOs), or utilities
and system operators that operate both, transmission and distribution systems. However, Distribution
System Operators (DSOs) were not included in the results of this report, because the responses show
that power system dynamic studies are barely conducted by DSOs.

207
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The developed questionnaire was sent to 63 utilities and system operators around the world between
spring 2015 and summer 2016. Out of these 63 contacted utilities and system operators, 45 replied to
the JWG. Hence, the response rate of 71% was reached. The 45 received questionnaires from utilities
and system operators came from 21 countries on five continents. The participants mostly operate an
interconnected grid, and only a few operate an isolated small grid. The system sizes vary from small-
scale power systems (a few megawatts) to large-scale power systems (several gigawatts).
It should be noted that software vendors, consultancies, research organizations and academia are not
included in the interviewees. Out of the 45 received questionnaires, 35 came from TSOs, and 10 came
from utilities and system operators that operate both, transmission and distribution systems.
Table 1-B-1: Survey categories and questions
No. Category No. Question

Which of the following inverter-based generation

technologies do you model for power system dynamic
studies?

Type of the inverter-based 1.1 Wind

1 generation technology and its 1.2 Photovoltaic
modelling
1.3 Micro turbine
1.4 Fuel cell
1.5 Battery energy system
What type of model do you use for a specific type of
power system dynamic study?
2.1 Frequency stability (large system)
2.2 Short-term voltage stability (seconds)
2.3 Short-circuit provision from inverter-based generation
2.4 Low voltage ride-through
2.5 High voltage ride-through

Type of model used for a specific 2.6 Transient stability with balanced faults
2 type of power system dynamic
2.7 Transient stability with unbalanced faults
study
2.8 Long-term voltage stability (minutes)*
2.9 Small-disturbance angle stability
2.10 Unintentional islanding
2.11 Transients including switching transients
2.12 Control system interactions (high frequency)
2.13 Control system interactions (low frequency)
2.14 Protection coordination
* Voltage fluctuations at steady state is not considered as long-term voltage stability

Chapter 1-B.3 Type of Models

Usually, power system dynamic studies are examined in various system sizes, such as studies in a local
power system or in a bulk power system. Therefore, various types of models are applied. In general,
there are many levels of models used for all types of IBG. This report focuses on individual and
aggregated models of IBG, as explained in the following sub-chapters.

208
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Chapter 1-B-3.1 Individual Model

In this report, individual models of IBG are defined as: 1) one model for each generator/inverter; or 2)
one model for a group of generators/inverters within the same power plant/park/farm. For a dynamic
study in a local power system, an individual model is more likely to be applied.
(a) Individual models of large-scale power plants of inverter-based generation
Individual models of large-scale power plants which consist of a group of generators/inverters are not
likely to be modelled individually even for power system dynamic studies in the local network. Because
not only the generator/inverter type but also the manufacturer can be the same, the single-machine
equivalent model, as recommended by, e.g., WECC [14], can be used for local power system dynamic
studies. Therefore, the IBG plant is represented by the collector system equivalent, the step-up
transformer equivalent and the generator/inverter equivalent [15], as shown in Figure 1-B-1. Additionally,
this model may also be used for bulk power system studies.
(b) Individual models of small-scale distributed systems of inverter-based generation
If the system is not too large, individual models of small-scale distributed systems may explicitly
represent each generator/inverter, load, line, etc. (the distribution system) individually for the local power
system dynamic study, as depicted in Figure 2. It should be noted that the structure of the network in
Figure 1-B-2 is just an example.

Real power plant Model used for power system dynamic studies

PV PV PV PV PV

Figure 1-B-1: Individual model of a large-scale photovoltaic plant (adapted from WECC [6]).

Real power system Model used for power system dynamic studies

PV PV
PV PV

Figure 1-B-2: Individual model of small-scale distributed photovoltaic systems.

Chapter 1-B-3.2 Aggregated Models

In this report, aggregated models of IBG are defined as the lumped model, which represents the dynamic
behaviour of the system in an aggregated manner. For a dynamic study in a bulk power system, an
aggregated model is more likely to be applied.
(a) Aggregated models of small-scale distributed systems of inverter-based generation
Small-scale distributed systems of IBG, e.g., residential PV, can be represented as an aggregated
model, as recommended by, e.g., WECC [14], for bulk power system dynamic studies. Therefore, the

209
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

distribution system is modelled with the equivalent impedance, the load equivalent and the
generator/inverter equivalent, as illustrated in Figure 1-B-3. The goal is to capture the impact of small-
scale distributed IBG, lumped at the transmission level, on the bulk system performance. However, there
are still no widely accepted aggregated dynamic models for small-scale distributed IBG. In this context,
the most recent work has been done by WECC [16], but it is presently under review and discussion for
further changes.
Real power system Model used for power system dynamic studies

PV
PV

PV

Figure 1-B-3: Aggregated model of small-scale distributed photovoltaic systems (adapted from
WECC [6]).

Chapter 1-B-4 Results

Chapter 1-B-4.1 Category 1: Type of the inverter-based generation technology and its
modelling
Category 1 is about the predominant IBG technologies that are modelled by utilities and system
operators and the results are shown in Figure 1-B-4. It can be concluded that wind and PV generation
are predominantly modelled for power system dynamic studies with 76% and 67%, respectively.
Furthermore, 24% of the participants model battery energy systems, and only 9% and 7% model micro
turbines and fuel cells, respectively.
1.5 1.4 1.3 1.2 1.1
based generation

Wind 76
Type of inverter

Photovoltaic 67
Micro turbine 9
Fuel cell 7
Battery energy system 24
0 20 40 60 80
Responses of the
participants [%]

Figure 1-B-4: Question 1: Which of the following inverter-based generation technologies do you
model for power system dynamic studies?

Chapter 1-B-4.2 Category 1: Type of model used for a specific type of power system
dynamic study
Category 2 is about the type of model that is used by utilities and system operators for a specific type
of power system dynamic study and the results are depicted in Figure 1-B-5. In general, it can be seen
that the application of individual models for IBG is slightly higher than the application of aggregated
models, and reaches 59% in average, as indicated in the last row of the figure.
The share of aggregated models is slightly higher for high voltage ride-through and transient stability
studies, compared to the average value. These results are identical to the aforementioned observations
because those dynamic studies are typically performed in a bulk power system.
The share of individual models is slightly higher for voltage stability and control system interactions
studies, compared to the average value. These results are identical to the aforementioned observations
because those dynamic studies are typically performed in a local power system.

210
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Individual model Aggregated model

ALL 2.14 2.13 2.12 2.11 2.10 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1 Frequency stability (large system) 59 41
Short-term voltage stability (seconds) 62 38
Short-circuit provision from IBG 60 40
Low voltage ride-through 59 41
Type of power system dynamic study

High voltage ride-through 57 43

Transient stability with balanced faults 57 43
Transient stability with unbalanced faults 58 42
Long-term voltage stability (minutes) 64 36
Small-disturbance angle stability 63 37
Unintentional islanding 54 46
Transients including switching transients 57 43
Control system interactions (high frequency) 63 37
Control system interactions (low frequency) 59 41
Protection coordination 54 46
Average value of studies 2.1 to 2.14 59 41

0 10 20 30 40 50 60 70 80 90 100
Responses of the participants [%]

Figure 1-B-5: Question 2: What type of model do you use for a specific type of power system
dynamic study?

Chapter 1-B-4 Conclusions

The aim of this report is not necessarily to recommend the application of any dynamic model for a
specific type of power system dynamic study, but, rather, to identify what dynamic models are presently
applied and to provide some fundamental information on their use.
The main contributions and key findings of this report can be summarized as:
 Wind and PV generation is very likely to be modelled by utilities and system operators for power
system dynamic studies with 76% and 67%, respectively.
 59% of the utilities and system operators apply individual models instead of aggregated models for
power system dynamic studies.

In general, an individual model for IBG is typically used for the analysis in the local power system of,
e.g., protection coordination, unintentional islanding, short-circuit provision from IBG, transients
including switching transients, long-term and short-term voltage stability and control system interactions
(low and high frequency).
On the other hand, an aggregated model for IBG is typically used for the analysis in the bulk power
system of, e.g., frequency stability, low voltage and high voltage ride-through, transient stability (with
balanced and unbalanced faults) and small-disturbance angle stability.
However, for specific power system dynamic studies the aforementioned application of the different
types of models might also be used vice versa.
It can be concluded that every type of model has advantages and disadvantages and the adequate
model type needs to be selected depending on the type of power system dynamic study and the power
system condition (e.g. weak system conditions with very low short-circuit ratio or strong system
conditions with high short-circuit ratio).

211
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The results of the questionnaire emphasize the clear message for the necessity and importance of the
use of IBG models. Furthermore, the final technical brochure of the CIGRE JWG C4/C6.35/CIRED will
give guidance in selecting adequate models for IBG for specific power system dynamic studies.
With these contributions the report supports utilities and system operators as well as research institutes
and academia to benchmark their approach against the prevailing international industry practice.

212
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 2-A DYNAMIC VOLTAGE BEHAVIOUR BEFORE AND AFTER ISLANDING

IN RESPONSE TO REACTIVE POWER VARIATION
Most of the distributed synchronous generators connected to MV or lower network do not operate in the
AVR mode, but in the AQR mode. The major reasons are that the capacity of such synchronous
generators is relatively small and that they are not directly connected to HV network, i.e. the
transmission-level voltage network. In other words, the reactive support coming from such distributed
generators is not sufficient to contain the voltage.
Figure 2-A-1 shows an example how the change in the system voltage via the small capacity generators
is demanding. The measured data were obtained in the field test which was originally performed for
examining an anti-islanding performance. The type of anti-islanding protection is the reactive power
variation type (See also Clause 5.4.8-5). In this example, a 150 kW/185kVA class synchronous
generator is connected at the end of a 6.6 kV distribution feeder (See Figure 2-A-2). A sinusoidal wave
signal is added to the voltage reference of the AVR of the 185 kVA synchronous generator continuously
(See Figure 2-A-3) in order to generate the growing oscillation and to increase the frequency deviation
after the islanding is formed, which results in the tripping of the synchronous generator using the anti-
islanding protection. It is noted that the islanded grid consists of the 185 kVA synchronous generator
and a resistive load.
51 System Sepration
F[Hz]

50

49
170

160
PG[kW]

150

140

130
40

20
QG[kVar]

-20

-40
420

410
V[V]

400

390

380
-3 -2 -1 0 1 2
Time[s]

Figure 2-A-1: Image of generator output and frequency responses in case of generator tripping [17]
In this example field text, the system separation occurred at 0 second and the reactive power injection
which comes from the active anti-islanding protection successfully generate the growing oscillation of
the terminal voltage after the system is separated. Then, the synchronous generator was tripped at
around 2.05 seconds in Figure 2-A-1. It can be considered that this example also shows the difficulty of
the change in the terminal voltage using the small capacity synchronous generator while it stays
connected to the grid. It is also highlighted that the significant reactive power variation is observed while
the small capacity generator stays connected to the grid.
Generally speaking, it is expected that the terminal voltage of the generator can change using the
synchronous generator regardless of the capacity. However, when the small capacity synchronous
generator stays connected to the distribution feeder, the terminal voltage of the 185 kVA generator was
hardly increased/decreased, while the reactive power output clearly shows the sinusoidal wave. Once
the synchronous generator and the induction motor load are isolated from the distribution feeder, the
terminal voltage clearly shows the sinusoidal wave, while the reactive power output hardly shows the
sinusoidal wave. These measured dynamic behaviour reveals that the small capacity generator does

213
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

not have much ability to control the terminal voltage while it is connected. The terminal voltage is
regulated by the MV (or HV) grid. Therefore, it can be concluded that the transmission-level voltage
support is generally not to be achieved by any small capacity generators.

Synchronous Measured Values

Generator PG, QG, V, F 400V / 20 kV
187.5 kVA PCC
150 kW G Distribution
Network
Resistive Load
j0.017 p.u.

Figure 2-A-2: Testing system [17]

AC exciter
1.5 0.7

1 - + + 1 1+0.866s
Terminal Field
27.8
Voltage 1+0.02s + - + 0.24s 1+0.40s Voltage
+
1 1
1+0.79s 1+0.12s
0.337
Active anti-
+ Reactive Power
islanding Voltage
protection + Reference Reference
+
control signal
- + 1 Reactive
0.03 Power
1+0.02s
Cross current compensator

Figure 2-A-3: Control block diagram of exciter with anti-islanding protection [17]

214
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 3-A EXAMPLE PROTECTION FOR INVERTER-BASED GENERATOR

CONNECTING IN MV AND LV NETWORK
Chapter 3-A.1 Connection scheme
Generators connected to MV or LV networks usually have a connection scheme as shown in Figure 3-
A-1.
This scheme allows:
 the generator’s disconnection from the network in case of a fault in the network itself through the
interface switch opening;
 the possibility for the generator to continue to operate on its own loads (auxiliaries and others)
foreseen to operate in islanding conditions after the interface switch opens.

This connection scheme is usually associated to radial operation of electric network, which is widely
adopted in distribution systems. Sometimes, however, the scheme may be used also on transmission
systems, even if the feeder where the generator is connected is radial or meshed operated (See also
Chapter 3-A.3.1). In these cases, additional considerations may be taken with reference to the specific
transmission feeder.

Radial configuration

POC

General device
General protection

Interface device Interface protection

Loads for
non island
operation Generator device Generator protection

Load for
island
operation

Figure 3-A-1: Example of MV and LV generators connection scheme

Chapter 3-A.2 Protection system

Chapter 3-A.2.1 General protection
The general protection has the main objective to detect faults within the producer plant so separating
the entire producer plant from the distribution network in case of a fault internal to the producer plant
itself.
The type of protection, its sensitivity and operating times depend upon the protection and the
characteristics of the distribution network. A typical general protection system which is used in many
countries consists of:
 Over current protection (50/51) or fuses
 Zero sequence over current (51.0) (depending on the network operation criteria)
 Directional zero sequence over current (67.0) (depending on the network operation criteria)

It is noted that all device numbers of the protections (See Table 3-A-3) are followed by IEEE standard
[18], “C37.2-2008 - IEEE Standard Electrical Power System Device Function Numbers, Acronyms, and
Contact Designations.”

215
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Chapter 3-A.2.2 Interface protection system (IPS)

A typical IPS consists of:
 Under Frequency protection (81U)
 Over Frequency protection (81O)
 Under Voltage protection stage 1 (27.1)
 Under Voltage protection stage 2 (27.2) (optional)
 Over Voltage protection stage 1 (59.1)
 Over Voltage protection stage 1 (59.2) (optional)
 Zero sequence over voltage (59.0) (depending on the network operation criteria)
 Additional loss of mains passive function (optional).

IPS functions may be also supported by anti-islanding active detection methods (which are not to be
considered as IPS functions). It’s up to the electric system operator to accept or not these additional
methods and to define their interactions with IPS.
Chapter 3-A.2.3 Generator protection
In general, the generator protection has the main objective to detect faults into the generator in order to
protect the generator or within the producer plant in order to eliminate the fault.
The type of protection, its sensitivity and operating times depend upon the generator technology, the
characteristics of the plant and the distribution network they are connected to. A typical, non-exhaustive,
generator protection system is illustrated in Table 3-A-1. In case of inverter-based generators,
generators protections are, of course, different (See Clause 2.2.3-1).
Table 3-A-1: Comparison of typical generator protections
Protection Synchronous Asynchronous Static converter
Out-of-Step / Pole Slip (78) X - -
Overexcitation (Volts per Hertz) (24) X - -
Loss of Field (40) X - -
Field Ground (64F) X - -
Overcurrent (50/51) X X -
Over/under frequency (81) X X X
Negative Sequence Overcurrent (46N) X X -
Differential (87G) X - -
Over/under voltage (59/27) - X X
Others X X X
Chapter 3-A.2.4 Protection coordination
The generating plant and the electric system protection systems should be coordinated in order to detect
and eliminate faults isolating the minimum network section. The connection scheme and the protection
system adopted should be coordinated in such a way that:
1. start up, operation and stop of the generator under normal network operating conditions, i.e. in the
absence of faults or malfunctions, should be assured;
2. faults and malfunctions within the generating plant should not impair the integrity of the distribution
network;
3. co-ordinated operation of the interface switch with the generator switch, the general switch, and the
distribution network switch, for faults or malfunctions during operation in parallel with the distribution
network itself should be assured;
4. reliable disconnection (with reference to the specific operation rules and distribution network) of the
generating plant from the distribution network by tripping the interface device should take place in
the following cases:
 intentional opening of the distribution network switch,
 faults at the distribution network level (local event),
 abnormal voltage or frequency (i.e. excursions outside of set limits);
5. no disconnection of the generating plant due to IPS/loss of main functions should take place in
case of perturbations at transmission system level (system event), unless these perturbations
exceed voltage and frequency thresholds defined from TSOs, even if these thresholds are much

216
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

wider than those necessary in case of local events (different solutions are possible with the aim of
discriminating local events from system events).

In order to satisfy the above functions, coordinated but independent switches and protection equipment
may be applied to each of the following sections of the generating plant:
 generator;
 part of the producer’s network designed to run as an island (if required);
 the remaining part of the producer’s (i.e. all the remaining part of the producer’s network except the
section able to be operated in island);
 distribution network.

Therefore, with reference to Figure 3-A-2, the coordinated sequence illustrated in Table 3-A-2 is
requested in case of faults located in different positions.

C B

POC

General
General protection
device

Interface Interface protection

device
Loads for E
non island
operation Generator Generator protection
device

Load for
island
operation

Figure 3-A-2: Possible locations of MV and LV faults

Table 3-A-2: Sequence of operation of switches with reference to Figure 3-A-2
Fault position Distribution switch General switch Interface switch Generator switch
A 1st level No 2nd level 3th level
B 1st level No 2nd level 3th level
C 1st level No 2nd level 3th level
D 2nd level st
1 level 3th level 3th level
E 2nd level 1st level 3th level 3th level
Chapter 3-A.3 Example protection for inverter-based generators connected to the HV network
For simplicity, in this Chapter HV networks are assumed to be meshed. Of course, also radial operated
networks may be used at HV level but these conditions may be referred to MV/LV connection (see
Chapter 3-A.2.4)
Chapter 3-A.3.1 Connection scheme
Generators connected to HV networks may have:
1. same scheme used for MV/LV connection (see Figure 3-A-3) in case of generators not called to
contribute to network operation, that is generators that may be disconnected from the network in

217
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

case of fault in the network itself (usually of not of appreciable power with reference to voltage level
of Point of Common Coupling)
2. a connection scheme without the interface device and the relevant interface protection system in
case of generators called to contribute to network operation, that is generators that cannot be
disconnected from the network in case of fault in the network itself unless a really important
deviation from standard voltages and frequencies values in detected for relatively long time.

In this case, a typical connection scheme is illustrated in Figure 3-A-4.

Meshed configuration

POC

General device
General protection

Interface device Interface protection

Loads for
non island
operation Generator device Generator protection

Load for
island
operation

Figure 3-A-3: Typical HV generators connection scheme (type A)

Meshed configuration

POC

General device
General protection

Generator device Generator protection

Auxiliaries

Figure 3-A-4: Typical HV generators connection scheme (type B)

Chapter 3-A.3.2 Connection Type A

This connection is equal to that used in MV connection (See Chapter 3-A.2.4). Of course, the main
difference concerns the generator protections that may be different due to the generator size, but they
have the same behaviour in case of fault as described in Chapter 3-A.2. In particular, the following
additional protection could be installed (referred to traditional synchronous generators):
 Impedance (21),
 Power (32),

218
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 Power swing blocking (68).

Chapter 3-A.3.3 Protection coordination

It is worth remembering that the generator protection should both detect faults into the generator in order
to protect the generator and within the producer plant in order to eliminate the fault. But, at the same
time, it should allow the generator to continue to operate during faults in the HV network (within certain
limits in voltage, frequency and duration).
Therefore, with this connection scheme the protection coordination is quite similar to that described in
app. 3-A.2.4. The difference is present only in case of faults on the line where the generator is connected
(fault position B or C in Figure 3-A-5) because MV (LV) network is usually radial operated, whereas HV
is usually meshed operated.
With reference to Figure 3-A-5 and Figure 3-A-6, the coordinated operation sequence driven from
protection system in case of faults located in different positions of the line is:
 Fault position B (see Figure 3-A-5): B is intended to be not very close to the busbar where the
generating plant is connected.

In this case CB1 and CB2 switches will open and the fault will be cleared. No protection inside the
producer plant shall trip due to the coordination of the regulations (in terms of values of input quantities
and of intentional operating delay): in the example the undervoltage interface protection is properly
delayed.
CB1 and CB2 switches will re-close if automatic reclosing is foreseen and, again, no protection of the
producer plant shall trip.
 Fault position B’ (see Figure 3-A-6): B’ is intended to be very close to the busbar where the
generating plant is connected.

In this case three switches, CB1, CB2 and CB3, will open because no pilot scheme is used, and the
fault will be cleared. No protection of the producer plant shall trip during the fault because the
undervoltage interface protection is delayed. After the three switches open, the plant is disconnected
from the HV network and the interface protection will trip.
It’s worth noting that in case of pilot protection scheme (POTT, PUTT), only two switches will open and
the co-ordinated sequence is equal to fault position B.
Chapter 3-A.3.4 Connection Type B
This connection is completely different from that used in distribution systems, because the goal is to
maintain the generator connected as long as possible.
Chapter 3-A.3.5 Protection system
Since the goal is to maintain the generator connected as long as possible, there is no interface protection
included. In addition, the generator protections may be extremely different depending on the importance
of the generator in terms of rated power.

219
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

POC

General
General protection
device

Interface Interface protection

device
Loads for E
non island
operation Generator Generator protection
device

Load for
island
operation

Figure 3-A-5: Locations of HV faults

B’

POC

General
General protection
device

Interface Interface protection

device
Loads for E
non island
operation Generator Generator protection
device

Load for
island
operation
Figure 3-A-6: HV faults

Chapter 3-A.3.6 Protection coordination

Also, in this case, the generator protection should both detect faults into the generator in order to protect
the generator and within the producer plant in order to eliminate the fault. At the same time, it should
allow the generator to continue to operate during faults in the HV network. The important difference with
the connection scheme A (See Chapter 3-A.2.4) is that faults are eliminated in a completely selective
way by using pilot a protection scheme (POTT, PUTT); or using a protection scheme that provides zone-

220
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

extension in combination with single-pole reclosing procedure. Therefore, the generator remains always
connected to the network and in case of fault:
 continues to supply the fault before completely selective fault elimination,
 continues to operate after the fault elimination.

Table 3-A-3: Device number of protections shown in Appendix 3-A

Device number Name of protection (IEEE Std C37.2) Sign used in some countries
21 Distance relay DZ
24 Voltage per Hertz relay V/F
27 Undervoltage relay UVR
32 Directional power relay
40 Field (over/under excitation) relay
46 Reverse phase or phase balance current relay
50 Instantaneous overcurrent relay
51 AC time overcurrent relay OCR
59 Overvoltage relay OVR
64 Ground detector relay
67 AC directional overcurrent relay
68 Blocking relay or out-of-step relay
78 phase-angle measuring relay
81 Frequency relay
81O Overfrequency relay OFR
81U Underfrequency relay UFR

221
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 3-B CALIFORNIA ISO EXPERIENCE: OVER-GENERATION ASSESSMENT

Chapter 3-B.1 Over-generation issues and metrics
More and more conventional resources are being displaced with renewable resources as renewable
penetration increases, and inverter-based renewable resources may not be equipped to provide
frequency response. The California ISO (CAISO) is concerned that during periods of light load and high
renewable production, the system may require reserving headroom on governor responsive resources
or primary frequency control capability to meet frequency response obligations as proposed under the
NERC Standard BAL-003-1 (Frequency Response and Frequency Bias Setting [19]). As described in
Sub-Chapters 2.2, 3.2.4 and 3.2.5-7, unlike conventional generation, current IBGs do not provide inertia
control to arrest frequency decline following the loss of a generating resource. As more wind and solar
resources displace conventional synchronous generation, the mix of the remaining synchronous
generators may not be able to adequately meet the CAISO’s frequency response obligation (FRO) under
BAL-003-1 for all operating conditions.
The objectives of this study were to assess the potential risk of over-generation conditions in the 2020
timeframe under 33 percent Renewable Portfolio Standard (RPS), evaluate the CAISO’s frequency
response during light load conditions and high renewable production, assess factors affecting frequency
response, and evaluate mitigation measures for operating conditions during which the FRO couldn’t be
met.
Over-generation occurs when there is more internal generation and imports into a balancing area than
load and exports. The risk of over-generation is illustrated on the curve in Figure 3-B-1. This curve
represents net load for multiple years during a spring day with light load and high renewable production.
Although load is the true demand that must be served moment by moment, net load is the demand met
by dispatchable resources.
Before an over-generation event occurs, the system operator will exhaust all efforts to send dispatchable
resources to their minimum operating levels and will have used all the decremental energy (DEC) bids
available in the imbalance energy market. If no DEC bids or insufficient DEC bids are received, the
system operator may declare an over-generation condition if high system frequency and associated high
Area Control Error (ACE) can no longer be controlled. With a high ACE, the energy management system
(EMS) will dispatch regulation resources to the bottom of their operating range. Also, operators will make
arrangements to sell excess energy out of the market to the extent bids to balance the system are
exhausted.

Figure 3-B-1: Duck-shaped curve shows steep ramping needs and over-generation risk.
 real-time energy market prices may be negative — the CAISO must pay internal or external entities
to consume more or produce less power;
 ACE is higher than normal and can result in reliability issues;
 grid operators may have difficulties controlling the system due to insufficient flexible capacity;

222
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 insufficient frequency responsive generation on-line may reduce the system ability to quickly arrest
frequency decline following a disturbance;
 inability to shut down a resource because it would not have the ability to restart in time to meet
system peak;
 need to commit more resources on governor control; and
 possible curtailment of resources that cannot provide frequency response.
 Frequency response is the overall response of the power system to large, sudden mismatches
between generation and load. The study focused on light spring conditions, because the relatively
low level of conventional generation may present a challenge in meeting the FRO. NERC
developed the frequency response obligation of the Western Interconnection based on the loss of
two fully loaded Palo Verde nuclear power station units (2,750 MW). This is a credible outage that
results in the most severe frequency excursion post-contingency.
 The following frequency performance metrics that were proposed by the CAISO and General
Electric Energy were used in the study and are illustrated in Figure 3-B-2.

Figure 3-B-2: Frequency performance metrics.

where,
Cf — Frequency Nadir (Hz)
Ct — Frequency Nadir Time (sec)
Bf — Settling Frequency (Hz)
Δ MW/Δfc *0.1 — Nadir-Based Frequency Response (MW/0.1HZ)
Δ MW/Δfb*0.1 — Settling-Based Frequency Response
Cp — Nadir-based governor response (MW)
Bp — Settling frequency-based governor response (MW)
The system frequency performance is acceptable when the frequency nadir post-contingency is above
the set point for the first block of the under-frequency load shedding relays, which for most loads are set
at 59.5 Hz. Another metric is the actual CAISO’s frequency response following a contingency. The
Western Interconnection Frequency Response Obligation is updated annually, according to the NERC
BAL-003-1 standard. The NERC-established annual interconnection frequency response obligation for
the Western Interconnection is currently set at 949 MW/0.1Hz, which was the value used for this study.
Frequency response of the Interconnection is calculated as:
Δ𝑃
𝐹𝑅 = [𝑀𝑊 ⁄0.1𝐻𝑧 ]
Δ𝐵𝑓
Equation 3-B-1

223
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

where 𝛥𝑃 is the difference in the generation output before and after the contingency, and 𝛥𝐵𝑓 is the
difference between the system frequency just prior to the contingency and the settling frequency. For
each balancing authority within an Interconnection to meet the NERC Standard BAL-003-1, the actual
frequency response should exceed the FRO of the balancing authority. FRO is allocated to each
balancing authority and is calculated using the formula below.
PgenBA + Pload BA
FROBA  FROint
Pgenint + Pload int
Equation 3-B-2
where BA stands for Balancing Authority, Int for interconnection, Pgen for annual generation output
(MWH) and Pload for annual demand (MWH). For the CAISO, annual FRO obligation is approximately
30 percent of WECC FRO, which is 285 MW/0.1 Hz.
The ratio of generation that provides governor response or primary frequency control capability to all
generation running on the system is used to quantify overall system readiness to provide frequency
response. This ratio is introduced as the metric Kt; the lower the Kt, the smaller the fraction of generation
that will respond. The exact definition of Kt is not standardized. For this study, it is defined as ratio of
power generation capability of units with responsive governors to the MW capability of all generation
units. For units that don’t respond to frequency changes, power capability is defined as equal to the MW
dispatch rather than the nameplate rating because these units will not contribute beyond their initial
dispatch.
The other metric that was evaluated was the headroom of the units with responsive governors. The
headroom is defined as a difference between the maximum capacity of the unit and the unit’s output.
For a system to react most effectively to changes in frequency, enough total headroom must be available.
Block loaded units have no headroom.
Chapter 3-B.2 Study assumptions
The power-flow base case selected for the study was based on the results of production simulations for
the year 2024. Production simulations represent the system performance considering security-
constrained unit commitment (SCUC) and security-constrained unit dispatch (SCUD) for each hour of
the year. The model for production simulation was obtained from the WECC Transmission Expansion
Planning Policy Committee (TEPPC) Study Program. The latest 2024 Common Case was used. The
Common Case is the first base case for the 10-year timeframe from which additional portfolio cases can
be developed. The production simulation case selected for the study modelled 33 percent of renewable
resources in California and had the latest updates on the new transmission and generation projects.
The model used the California Energy Commission (CEC) load forecast for California for the year 2024
developed in 2013 and the load forecasts for other areas from the latest WECC Load and Resources
Subcommittee (LRS) data developed in 2012. New renewable generation projects were modelled
according to the California Public Utility Commission (CPUC) renewable resources portfolios. All other
assumptions were consistent with the CAISO 2014 Unified Study Assumptions and the latest TEPPC
database.
The production simulation was run for the year 2024 using ABB Grid View software. The hour of the
year selected for the detailed transient stability studies modelled low load and high renewable generation
that usually occurs in spring. Based on the production simulation results, the hour of 11 am April 7, 2024
was selected because it represents a low load and high renewable production scenario. Power flow
case was created for the 11 am, April 7, 2024 with the generation dispatch and load distribution from
the results of the production simulation study. The power flow case was created by exporting the results
of the Grid View production simulation for the selected hour and solving the case in a commercially
available time-domain simulation tool. Due to high voltages because of low load in the selected hour,
reactive support was adjusted by turning off shunt capacitors and turning on all available shunt reactors.
Dynamic stability data file was created to match the power flow case. The latest WECC Master Dynamic
File was used as a starting dataset. Missing dynamic stability models for the new renewable projects
were added to the dynamic file by using typical models according to the type and capacity of the projects.
The latest models for IBG recently approved by WECC were utilized. For the new wind projects, the
models for type 3 (double-fed induction generator) or type 4 (full converter) were used depending on the
type and size of the project. For the solar PV projects, three types of models were used: large PV plant,
small PV plant and distributed PV generation. It was assumed that the large plants (20 MW and higher)
have centralized plant control. More detailed description of the RMS dynamic stability models for
renewable generation used in WECC is provided in the Appendix 5-C.

224
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3-B-1: Over generation base case assumptions for hour of 11 a.m. April 7, 2024

Base Case Assumptions WECC CAISO

Load, MW 100,410 24,117
Losses, MW 3,162 510
Generation, MW 103,580 22,650
Wind and solar output, percent of total dispatch 25.8 percent 48.6 percent
COI (California-Oregon PDCI (Pacific DC
Intertie) Intertie)
Flow, MW 1170, north-to-south 620, north-to-south
Import to the CAISO, MW 1977

Table 3-B-2: Generation by Type, April 7, 2024 11 a.m. (in MW)

Natural
Area Nuclear Geothermal Biomass Coal Hydro Storage Solar Wind
Gas

Capacity 2300 2005 1350 404 7125 34214 3718 18143 7000
CAISO
Dispatch 1150 948 605 0 1169 6914 -786 8621 2946

California Capacity 0 795 158 1640 3193 8826 1370 1812 437
Municipal
Utilities Dispatch 0 627 77 328 1038 449 392 1499 245

Rest of Capacity 5380 1431 1563 30814 56827 68281 985 5523 20165
WECC Dispatch 3976 1131 1053 22490 23459 12360 -451 4710 8713

Total Capacity 7680 4232 3071 32858 67145 111321 6073 25478 27602
WECC Dispatch 5126 2706 1735 22818 25666 19723 -845 14830 11904

The power flow case was adjusted to better match the case from production simulation and to ensure
that all generation is dispatched within the units’ capability. As a result, load, generation and flows in the
power flow case closely matched those from the production simulation study. The power flow base case
assumptions are summarized in Table 3-B-1.
Table 3-B-2 shows the capacity and dispatch levels of different types of generation technology modelled
in the study case.
The simultaneous loss of two Palo Verde generation units was studied because it results in the lowest
post-contingency frequency nadir. In this case, the generation loss was approximately 3% of total
generation dispatch in the Western Interconnection. The contingency of generator tripping was
simulated at 1 second and the transient stability and frequency stability simulation were run for 60
seconds.
In addition to evaluating the system frequency performance and the WECC and CAISO governor
response, the study evaluated the impact of unit commitment and the impact of generator output level
on governor response. For this evaluation, such metrics as headroom or unloaded synchronized
capacity, speed of governor response and number of generators with responsive governors were
estimated.
Chapter 3-B.3 Study results
The dynamic simulation results for an outage of two Palo Verde generation units showed the frequency
nadir of 59.708 Hz at 6.5 seconds (5.5 seconds after the contingency) and the settling frequency after
60 seconds at 59.882 Hz. The frequency plot for the six 500 kV buses (three buses in the north and
three in the south) with the largest frequency deviations is shown in Figure 3-B-3. The power swing
oscillation, the frequency component of which is about 0.3 Hz, is clearly observed in the same figure.

225
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-B-3: Frequency on 500 kV buses with an outage of two Palo Verde units.
As can be seen from the plot, the frequency nadir was above the first block of under-frequency relay
settings of 59.5 Hz. Figure 3-B-4 illustrates voltage at the same buses that was within the limits.

Figure 3-B-4: Voltage on 500 kV buses with an outage of two Palo Verde units.
The study evaluated governor response of the units that had responsive governors. Governor response
for the units with responsive governors varied from 4% of the rated capacity of the unit to 12% of the
rated capacity. Total kinetic energy (inertia) of the generators in WECC in this case was 589 GWs with
104 GW of generation dispatched. The calculated metrics of the frequency response and headroom for
the WECC and the CAISO are summarized in Table 3-B-3.

226
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3-B-3: Frequency response and headroom, April 7, 2024 11 a.m.

As can be seen from the table, the total WECC frequency response was within the BAL-003-1 standard
and well above the FRO: 2292 MW/0.1 Hz compared with the WECC FRO of 949 MW/0.1 Hz. However,
the CAISO frequency response was below its FRO: 269 MW/0.1 Hz when the ISO FRO is 285 MW/0.1
Hz. Thus, this study showed that although the total system performance was stable with no criteria
violations and the WECC frequency response was within the standard, the CAISO may not meet the
BAL-003-1 standard because its frequency response was below the frequency response obligation.
The metric Kt for this case was 56 percent for WECC and 31 percent for the ISO. Due to the large
amount of IBG within the ISO Balancing Authority Area (BAA), which is not responsive to changes in
frequency, the Kt metrics for the ISO was significantly lower than for the WECC as a whole. The
headroom of the frequency responsive generation at the ISO was relatively large (4420 MW), but it still
wasn’t sufficient to meet the frequency response obligation.
Sensitivity studies were performed to evaluate the system performance in case of reduced headroom in
the CAISO and in WECC. The original April 7, 2024 11 a.m. case had high headroom at the frequency-
responsive generation due to low dispatch of the generators that were modelled on-line. The sensitivity
case was created by turning off some units that had low dispatch and re-dispatching their output to other
on-line units in the same geographical zone, or at the same river for hydro power plants. The CAISO
generation headroom was reduced in this case from 4420 MW to 1925 MW and the total WECC
headroom was reduced to 12,000 MW. This case had 486 GWs of inertia with the same 104GW of
generation dispatched as in the initial case, which is 17.5% reduction in inertia.
The same contingency of an outage of two Palo Verde units was studied. Frequency on 500 kV buses
in the sensitivity case is shown in Figure 3-B-5.

227
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-B-5: Frequency on 500 kV buses with an outage of two Palo Verde units in case of reduced
headroom.
The study results showed the frequency performance that still was acceptable (nadir at 59.557 Hz and
settling frequency at 59.754 Hz), but it was close to the margin. 27 MW of load in British Columbia that
had under-frequency relay settings at 59.7 Hz was tripped. These loads did not include distributed PV
models. Also, 579 MW of distributed solar PV generation in WECC and CAISO was tripped due to the
under-frequency relays built in the distributed PV models, which had additional negative impact on
frequency.
For distributed solar PV generation, the WECC model PVD1 described in the Sub-Chapter 4.1 on the
RMS models was used, which is a simplified model of aggregated distributed PV units. This model
provides for tripping of some distributed units for low or high voltage or frequency. The range of
frequencies and voltages for which certain amount of distributed units may be tripped is specified by the
user. It was assumed that distributed generation will start tripping when the system frequency falls below
59.7 Hz and all distributed generation will be tripped at 59.5 Hz. The fraction of generation tripped by
the PVD1 model when the system frequency is between these set points is proportional to the frequency
deviation in the transient stability simulation. In this sensitivity study, it was assumed that the tripped
distributed solar PV will not reconnect during the timeframe of the simulation (60 seconds). These
assumptions are consistent with the IEEE 1547 standard. Since the PVD1 is a simplified model, it
doesn’t include the timer circuit element, so the reduction in the distributed generation output starts as
soon as the frequency reaches the first set point (in this case, 59.7 Hz).
The case had the total of 826 MW of distributed solar PV generation. The transient stability simulation
with frequency excursion of the case with the 12,000 MW headroom showed that 70 percent of
distributed generation was tripped. WECC frequency response was 1219 MW/0.1 Hz, which is within
the BAL-003-1 standard. However, the ISO frequency response was only 154 MW/0.1 Hz, which is
significantly below its frequency response obligation. Amount of responsive governors in this sensitivity
case was 46 percent for WECC and 26 percent for the ISO.
More sensitivity studies were performed to determine at which unit commitment and dispatch the system
performance may become unacceptable. Therefore, the headroom was reduced even more; however,
generation dispatch was changed only outside the CAISO since the CAISO already had insufficient
governor response.
The study with the headroom in WECC reduced to 10,100 MW identified frequency nadir at 59.5 Hz and
significant under-frequency load shedding: 537 MW of load, including pumping loads was lost, as well

228
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

as one 19 MW solar PV plant. In this case, also 100 percent of distributed generation (826 MW) was
tripped for low frequency. Due to the under-frequency load shedding, the frequency nadir did not go
below 59.5 Hz. This case had 476 GWs of inertia in WECC for the 104 GW of dispatched generation.
Total generation output was the same as in two other cases, but the inertia was reduced by 19%
compared with the starting case. Figure 3-B-6 illustrates frequency on selected 500 kV buses in the
case with the 10,100 MW headroom.

Figure 3-B-6: Frequency on 500 kV buses with an outage of two Palo Verde units in case of
headroom in WECC reduced to 10 GW.
The case with the critical reduction in headroom (to 10,100 MW) was studied in more detail to determine
impact of IBG on frequency response. The following additional scenarios were studied:
1. all distributed generation that was tripped for low frequency is reconnected when frequency
recovers to 57.0 Hz or higher;
2. no load shedding is available – this case was studied to determine frequency performance without
load shedding and to investigate if the frequency nadir would go below 59.5 Hz;
3. Large wind and solar PV plants are providing frequency response regulated by the power plant
controller. The upside droop (gain) was assumed at 20 per unit for each IBG plant with plant
controllers, as was recommended by GE. See block diagram of plant controller in Sub-Chapter 4.1
on the RMS models. Majority of the large solar PV plants were not dispatched to the maximum
output in the power flow case, therefore, they could provide frequency response.

Frequency plots on a 500 kV bus in Central California are shown for these cases and compared with
the case with the critically reduced headroom in Figure 3-B-7. As can be seen from these plots, ability
of the plants with IBG to provide frequency response allowed to maintain acceptable frequency
performance and to avoid significant load shedding. Recovery of the tripped distributed generation did
not have an impact on frequency nadir since it started to recover after the nadir was reached, but
resulted in higher settling frequency after the disturbance.

229
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Legend:

- Case with the

WECC headroom reduced to
10,100 MW (case 1)

- Case 1 with the

tripped distributed generation
reconnected

- Case 1 with under-

frequency load shedding
disabled

- Case 1 with large

inverter-based generation plants
providing frequency response

Figure 3-B-7: Frequency on Midway 500 kV buses with an outage of two Palo Verde units in case of
10 GW headroom and different scenarios.
In the case with the large wind and solar PV plants providing frequency response, there still was some
distributed solar PV generators that were tripped, since the frequency nadir was at 59.626 Hz which is
below the first set point for distributed generation tripping (59.7 Hz). Tripped distributed generation in all
of WECC amounted to 314 MW, or 38% of the dispatched distributed generation. The response from
large inverter-based plants was 919 MW, which is 8.4% of their installed capacity. Thanks to frequency
response from the IBG, the settling frequency after the contingency increased from 59.75 Hz to 59.83
Hz and the frequency nadir increased from 59.50 Hz to 59.63 Hz. Frequency response from total WECC
increased from 1063 MW/0.1 Hz to 1603 MW/0.1Hz and frequency response from the CAISO increased

230
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

from 154 MW/0.1 Hz to 511 MW/0.1 Hz. For this case, frequency response from the CAISO exceeded
its frequency response obligation.
However, it should be noted that the frequency response from the large IBG plants with centralized plant
controllers in this case was so high because majority of these plants were modelled as not being
dispatched to their maximum capability. Thus, it was assumed that they may increase their output up to
their rated capacity. The speed of the increase in their output depends on the droop specified in the
power controller model, which is demonstrated by the slopes of the curves in Figure 3-B-8.
Figure 3-B-8 illustrates response from selected large inverter-based power plants (wind and solar PV).
The generators that have a flat output were already dispatched up to the maximum capacity in the
starting case, therefore, their output didn’t increase in response to low frequency. The generator that
didn’t reach its maximum capability (the lower curve) was dispatched significantly below its capacity.

Figure 3-B-8: Power output from large wind and solar PV plants with frequency control.
Figure 3-B-9 illustrates output of selected aggregated units that represent distributed solar PV
generation in the case when it was assumed that the distributed generation tripped for low frequency
reconnects when frequency recovers. As can be seen from the plots, all distributed solar PV generation
was tripped at approximately 15 second which corresponds to the frequency nadir of 59.5 Hz, and then
started to reconnect with all distributed generation reconnected at 24 seconds when frequency
recovered to 59.7 Hz. Since the PVD1 is a simplified model and it doesn’t include the time delay (or
timer) component, the rate at which the distributed solar PV units reconnect is proportional to the rate
of the frequency recovery, the same way as the rate of their tripping is proportional to the frequency
decline. The frequency plot for this scenario is shown in Figure 3-B-7.

231
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-B-9: Power output from distributed PV generation assuming that tripped units reconnect
when frequency recovers.
The impact of reduced inertia and reduced headroom is illustrated in Figure 3-B-10. The plot shows
frequency on a 500 kV bus in Central California for an outage of two Palo Verde units in the starting
case with 30 GW of headroom and 589 GWs of inertia in WECC and in the case with 10 GW of headroom
and 476 GWs of inertia. The second case is shown with load shedding enabled, as well as with load
shedding disabled. As can be seen from this plot, frequency declines faster in the second case due to
the lower inertia and the frequency nadir and settling frequency are lower in this case due to the lower
headroom. With load shedding enabled, the frequency nadir is at 59.5 Hz, since the load shedding
doesn’t allow it to go below this point.

232
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-B-10: Frequency on the Midway 500 kV bus with an outage of two Palo Verde units in the
cases with high and low headroom and inertia.

Chapter 3-B.4 Conclusions

 The initial study results indicated acceptable frequency performance within WECC. However, the
CAISO’s frequency response was below the CAISO frequency response obligation specified in
BAL-003-1.
 Compared to the CAISO’s actual system performance during disturbances, the study results seem
optimistic because actual frequency responses for some contingencies were lower than the
dynamic model indicated. Optimistic results were partly due to large headroom of responsive
generation modelled in the study case.
 Amount of headroom on responsive governors is a good indicator of the Frequency Response
Metric, but it is not the only indicator. Higher available headroom on a smaller number of governor
responsive resources can result in less frequency response than lower available headroom on a
larger number of governor responsive resources for the same contingency.
 Requirement to provide frequency response from IBG plants may be a good solution to the problem
of insufficient frequency response.

The models for IBG used in the study appeared to show reasonable performance that would be expected
from such models. However, more work is needed to validate the model parameters by comparing
simulation results with measurements.

233
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 3-C LVRT STUDY FOR FREQUENCY STABILITY IN TASMANIAN GRID

The blue trace is a simulation of network frequency that follows a simple trip of the largest single
generating system (a radially connected CCGT unit). The red trace shows the exact same contingency,
i.e. the largest generator tripping, only this time as a result of a transmission line fault. The FRT (FRT)
response of the Tasmanian HVDC link and wind farms causes a very large transient energy deficit which
results in a more rapid decline in frequency than if a fault did not occur. The concern is that PV inverters
will add to this issue, especially under low short circuit conditions when voltage disturbances (due to
network fault events) will propagate more widely and therefore have an effect on more individual PV
installations. The protection characteristics of embedded inverters (pickup setting and time delays),
along with the modelling assumptions applied for the distribution network, will affect how significant this
issue becomes with more PV connected to the network.

Figure 3-C-1: Impact of low-voltage FRT response on network frequency control.

Understanding the potential range of under/over voltage settings (that might cause an FRT type
response), and being confident about how transmission voltage disturbances propagate downward into
the distribution network, are some of the challenges going forward. Note that the response of load
devices to the same voltage disturbance means that there is probably a growing need to understand the
appropriateness of existing load models because of the feedback effect this has on voltages which the
inverter is then exposed to.

234
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 3-D IMPACT OF CCGT TRIPPING WITH PV ON FREQUENCY STABILITY

IN ISRAELI GRID
During the period of 2012 -2013, the system has suffered from the numerous power plants outages,
mostly CCGTs, which caused to frequency drop below 49.4 Hz and load shedding activation.
In the past, the primary spinning reserve was enough to cope with such power plant outages in order to
prevent load shedding activation. Since 2011 the high level of PV penetration, mostly on LV, affected
frequency system ability to cope with frequency disturbances. Because of the isolated system (Capacity:
16.0 GW, Peak Load: 12.0 GW, Average Load: 7 – 8 GW, PV: 0.6 GW), frequency may vary in range
of 49.8 – 50.2 Hz on regular base, the system is very sensitive to frequency behaviour during
disturbances, especially daylight hours, when PV generations is effective.
Based on the information gathered from different load shedding activations, it was unreasonable amount
of shed load compared to the power station load before its shutdown. The facts talked about themselves
– close to 200MW of installed PV have not being calibrated and mostly disconnected when the frequency
dropped below 49.2 Hz. Note that load shedding started to activate at 49.4 Hz and the system frequency
could continue to drop until it reaches at frequency nadir. If the frequency nadir below 49.2 Hz, PVs
would trip due to their underfrequency calibration. This could be the second consecutive disturbance
which causes to exacerbation of the frequency drop in addition to power station outage.
Since 2013, the new rules for PV calibrations were defined for new PV installations – frequency level
for disconnection changed from 49.2 Hz to 47 Hz together with the change in time delay (from 0 second
to 1 second). Finally, the problem with recalibration of old installations was completed in 2015. Currently,
during the off-season peaks PV penetration is up to 7-8% of from Max daily peak. During this period –
some coal fired power plants are on maintenance and primary spinning reserve are secured mainly from
CCGTs which are not the best tool to cope with frequency drops.
In order to cope with such PV donation to load shedding activations, the primary system reserve
increased and a lot of activities performed on CCGTs in order to improve their primary response. The
practice shows that the best way was to cope with frequency drops by thermal coal and natural gas
power plants.
The facts similar to "50.2 Hz" [20] phenomena, known from Germany, have taken place in the system.
The 5 new High Voltage PV Plants (200 MW) are operating starting 2014. (By 2020 the total PV installed
capacity should reach the level of 3000 MW, compare to the current 900 MW (see Table 3-D-1). The
reason for that growth is the new government targets for renewable energy generation, mostly PV
installations – 10% by 2020, 13% by 2024, and 17% by 2030. The ability of the system to cope during
frequency drops with number of those plants in nearest future will be a challenge to the system stability
when system inertia is going down at time of PV installations going up.

Figure 3-D-1: Installed PV capacity in Israel.

Regarding compliance with various voltages:
 The facility shall continue to generate when the voltages at the point of connection to the
transmission system lie in the permitted range (150 to 170 KV).
 The facility shall continue to generate during and after a disturbance to the transmission system,
that causes an irregular voltage at the point of connection of the facility to the transmission system,
in accordance with the following LVRT/HVRT (low/high voltage ride through) curve:

235
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-D-2: LVRT AND HVRT functionality.

where
Region 1: The facility shall not disconnect from the grid and shall continue to generate.
Region 2: The facility shall be entitled to disconnect for a short time until removal of the disturbance.

If the facility did not disconnect from the grid, after removal of the disturbance, the active power shall be
increased at a rate that shall be not less than 20% of the available power per second.
Regarding compliance with frequency regulation:
 Continuous operation in range 47 Hz to 51.5 Hz
 The plant should be disconnected from the grid in case of grid frequency will be:

Table 3-D-1: Amended grid requirement for PV from voltage point of view in Israel

Max time before disconnection Frequency level at the point of grid connection
1 sec F< 47 Hz
0.2 sec 51.5 Hz < F
Continuous Operation 47<F<51.2 Hz

Other than the above, the frequency response requirement is also regulated:
Pavailable f - 50.2 Hz
DP  -  ( 50.2 Hz  f  51.5 Hz )
R fbase
where, f denotes system frequency, fbase denotes nominal system frequency, Pavailable denotes available
capacity, R denotes droop (3%), Pgenerated denotes power generated in the range of 50.2 Hz and 51.5 Hz,
DP denotes power deviation in case of the aforementioned frequency deviation.
The new challenge of the last year is coping with single and multiple phase short circuits in the system.
The LVRT and HVRT functionality of the HV PV plants is still under the cloud of uncertainty. The different
Grid Codes define those requirements in general and leave the place to inverter's and control system
manufacturers to implementations based on hardware and equipment limitations. The system Operator
challenge is to define those features from the best proximity to system stability.
Presented underneath couple diagrams from LVRT activations triggered by SC close to HV PV plant –
40 MW installed capacity.

236
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-D-3: Voltage and current behaviour during disturbance.

Figure 3-D-4: Voltage, Active and reactive power during disturbance.

237
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-D-5: Current behaviour during disturbance.

Figure 3-D-6: Harmonics voltage during disturbance.

238
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 3-E CALIFORNIA ISO EXPERIENCE: IMPACT OF PV ON SHORT-TERM

VOLTAGE STABILITY
Chapter 3-E.1 Study background
In recent years, the penetration of renewable resources in the transmission systems in North America
and everywhere in the world significantly increased and it is expected to be even higher in the near
future. In the transmission system of the California Independent System Operator (CAISO) which serves
80% of the California customers, 33% of annual energy is expected to be covered by renewable
resources by the year 2020. Most of the renewable resources are electronically‐ coupled to the grid,
which presents some new and different technical challenges.
There is a need for generic, standard and publicly available models for IBG technologies for the purpose
of power system planning studies. There are several industry-wide efforts to develop such models, one
of which is Western Electricity Coordinating Council (WECC) Modelling and Validation Working Group
[21]. This group, with participation of Electric Power Research Institute (EPRI), National Renewable
Energy Laboratory (NREL) and other entities has developed models, including the ones for wind and
solar PV generation that were implemented by software vendors and approved for use in the Western
Interconnection of North America. These models were validated by EPRI on the individual unit level
using EPRI tool and on the plant level using PMU recordings from several system events. The validation
results are described in the EPRI report available on the EPRI website and cited in [22].
The current report is devoted to evaluating impact of solar photovoltaic (PV) generation and modelling
of this type of generation. Most of PV system are small size (up to several kW) residential and larger
size (up to several MW) commercial installations connected to distribution networks. However, many PV
systems are large power plants that connect to transmission systems mainly at 115 kV and 230 kV
voltage levels, but some large plants may be connected at 500 kV. Some PV plants connect to sub-
transmission systems at 60-70 kV voltages. Currently, the CAISO transmission system has around 4100
MW installed capacity of solar PV generation with another 12600 MW in the generation interconnection
queue which was proposed to be installed by the year 2020.
A study was performed to determine impact of high penetration of IBG on voltage stability in the
California transmission system. The study used a model that included all Western Interconnection
(WECC) transmission system. It was performed for a future year (2023) under peak load conditions with
high output of solar PV generation. It was assumed that 33% of the CAISO generation is provided by
renewable resources.
Chapter 3-E.2 Study assumptions and models
Power flow case selected for the study was 2023 Heavy Summer with 33% of renewable resources in
the California ISO system. Total WECC load was assumed at 191GW, CAISO load at 62GW. Output of
the Solar PV generation in the CAISO system was modelled at 2900 MW and output of wind generation
was modelled at 1650 MW. The rest of renewable generators were solar thermal, small hydro,
geothermal and biomass power plants.
Solar PV plants were modelled as aggregated generation units connected to the transmission system
through step-up transformers, equivalent collector system and substation transformers to step up from
the collector system voltage to the transmission voltage. This representation is shown in Figure 3-E-1.
Dynamic stability models for solar PV generation were assumed to be mostly generic. A commercially
available time-domain simulation tool was used in the study with the models available in this software.
Description of these models is provided in Chapter 5 and detailed in Appendix 5-B. Figure 3-E-2 shows
the modules included in the model.
In this short-term voltage stability study the following models were used for large solar PV plants
connected to the sub-transmission (60-70 kV) and transmission systems (115 kV and higher):
 REGC_A module, used to represent the Generator/Converter (inverter) interface with the grid.
 REEC_B module, used to represent the Electrical Controls of the inverters.
 REPC_A module, used to represent the Plant Controller. It processes voltage and reactive power
output to emulate volt/var control at the plant level.

Small distributed solar PV plants were modelled as aggregated at the sub-transmission level. Since
small solar PV installations don’t have central plant control, they were modelled with only the
generator/converter and electrical control models.

239
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

In addition to this representation, over- and under-voltage and frequency protection relays were
modelled for large plants connected to the transmission system. It was assumed that all PV units are
capable of riding through the faults. Therefore, generators’ tripping settings in the models were assumed
to have at least 0.2 second delay for low voltage and 0.1 second delay for high voltage.
The system load was modelled with a composite load model available in the commercially available
time-domain simulation tool (cmpldw). This model was applied to the entire WECC system, using the
WECC load model data tool (LMDT). The load composition for a peak summer hour was assumed with
default parameters of the individual induction motors modelled according to the research data provided
by WECC. The study was performed for two options of the conduct of the single-phase air-conditioner
motor load components: 1) single-phase air-conditioners stall with low voltage and 2) single-phase air-
conditioners don’t stall. The under-voltage tripping settings for induction motors were assumed based
on the recommendations from WECC. The relay settings modelled used the combination of the voltage
sag amount and the delay timer. The exact relay settings varied according to the type of load, type of
the feeder and the climate zone in which the load was located. The load model used in the study is
shown in Figure 3-E-3. It didn’t include the Solar PV element, since the solar PV units in the distribution
system were modelled explicitly as aggregated units connected to the transmission.

Figure 3-E-1: WECC composite load model.

In Figure 3-E-1, the load is divided to the following types of induction motors:
 Motor A – 3-phase compressor in air-conditioners and refrigerators
 Motor B - fan.
 Motor C – industrial pumps
 Motor D – single phase residential air conditioner

Chapter 3-E.3 Study example [23]

The study investigated impact of solar PV generation on voltage stability with transmission system faults.
The study examples included a three-phase fault at the 230 kV Point of Interconnection (POI) of a large
(800 MW net output) Solar PV project. This plant is located in Central California. The plant
interconnection is shown in Figure 3-E-4 as a part of a larger system. The system model included all
WECC interconnection (20100 buses, 16800 transmission lines and 8100 transformers). The solar PV
units of the plant were aggregated into five equivalent generators. The step-up transformers and the
34.5 kV collector systems were modelled as their aggregated equivalents. The Figure 3-E-5 illustrates
the adjacent substations and the load areas.

240
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Note: Fault currents at POI 1 (15571 A 3-phase, 14976 A single-phase) and

POI 2 (15327 A 3-phase, 14808 A single-phase).
Figure 3-E-2: Studied network diagram with PV plant.

Figure 3-E-3: Area of studied PV plant.

241
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 3-E-1: Test Scenarios

Scenario 1 2 3 4 5 6
Generator power factor +/-0.95 +/-0.95 1 1 N/A N/A
Generator Q and V control Yes Yes No No N/A N/A
Network Motor Stalling No Yes No Yes No Yes
Inverter Q priority Yes Yes No No N/A N/A
Inverter P priority No No Yes Yes N/A N/A
Source PV PV PV PV AGCC AGCC

A three-phase 6-cycle fault was modelled at the 230 kV double-circuit transmission line between the
Solar Switching Station and the Bus 2 next to the bus of the Solar Switching Station cleared by opening
both circuits between the Solar Switching Station and Bus 2. The study monitored dynamic stability
performance of the solar PV units and the surrounding system, as well as performance of all Western
Interconnection. Several scenarios were studied. They are summarized in the following table and also
described below.
1. The Solar PV plant is capable of maintaining power factor between 0.95 lead and 0.95 lag and
controls reactive power and voltage. The inverters have reactive power priority selection in the
current limit logic because in this scenario generator regulates voltage. Plant controller controls
voltage. Stalling of the single-phase air conditioner motor load components is disabled.
2. Same as 1, but stalling of the single-phase air conditioner motor load components is enabled.
3. The Solar PV plant holds unity power factor and provides neither reactive support nor voltage
regulation. The inverters have real power priority in the current limit logic, which means that real
power output of the units is maintained as much as possible at the initial output level. Stalling of the
single-phase air conditioner motor load components is disabled.
4. Same as 3, but stalling of the single-phase air conditioner motor load components is enabled.
5. The Solar PV plant is replaced by a fictitious combined-cycle plant of the same size to compare the
response and to evaluate impact of IBG plant. Stalling of the single-phase air conditioner motor
load components is disabled.
6. Same as bullet point No. 5, but stalling of the single-phase air conditioner motor load components
is enabled.

Chapter 3-E.4 Study results

The following plots illustrate the study results. Figure 3-E-6 shows voltage at the 230 kV bus of the Solar
Switching station where the three phase fault was applied. Voltage is shown per unit; only first 5 seconds
of the simulation are shown. As can be seen from the plots, voltage recovered to the pre-fault values in
all scenarios, except for the solar PV plant with unity power factor. In this case, voltage recovered to a
slightly lower value. The inverter-based plant showed even faster recovery than the combined-cycle
plant. Whether single-phase air conditioning load stalled or not did not have almost any impact on the
voltage on this bus.

242
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-E-4 Voltage at the 230 kV bus of the Solar Switching station with a 3-phase fault.

243
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The Figure 3-E-5 shows voltages at the terminals of one of the equivalent generators and terminals of
the generator that was modelled as a combined-cycle unit replacing the solar PV plant. Voltage is shown
in per unit; only first 5 seconds of the simulation are shown.
As can be seen from Figures 3-E-5, both solar PV with reactive power and voltage control and the
thermal unit were able to hold scheduled voltage. Solar PV with unity power factor and without voltage
control still had the voltage recovered, but voltage recovered to a slightly lower value. Single-phase air-
conditioner load stalling did not have any impact except for very slightly lower voltages in case of solar
PV with unity power factor. It is noted that the significant transient overvoltage is observed in Scenarios
1 and 2 right after the fault is cleared.

Figure 3-E-5 Voltage at equivalent generator terminals.

244
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The Figures 3-E-6 and 3-E-7 show real and reactive power response from the solar PV equivalent
generator and the replacement combined cycle plant unit. As can be seen from these plots, inverter-
based generators had better damping; however, they had slightly slower real power recovery after a
fault. Generator in scenarios 3 and 4 (unity power factor) didn’t provide any reactive power output, thus,
didn’t provide voltage support. This can be seen in Figure 3-E-7. The studies didn’t show any impact
from the single-phase air-conditioners stalling, except for a very slight increase in reactive power output
from the generators in scenarios 2 and 6 to compensate for the induction motor stalling

Figure 3-E-6 Active power output from PV equivalent or thermal generator.

245
MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-E-7 Reactive power output from PV equivalent or thermal generator.

246
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The Figure 3-E-8 illustrates the voltages in per unit at a load bus close to the fault (Load Bus 1 in Figure
3-E-3). As can be seen from these plots, voltage recovery on the load bus was not impacted by whether
the plant was solar PV or combined-cycle. In the cases when stalling of air-conditioners was modelled,
the plots showed delayed voltage recovery in all three scenarios.

Figure 3-E-8 Voltage at load bus closed to fault (Load Bus 1, 230 kV).
All six scenarios showed some loss of load due to tripping of induction motor load and sensitivity of
some loads to voltage and frequency. The loss of load was rather insignificant and was not much
different between the scenarios. In an assumption that single phase air-conditioners don’t stall, loss of
load was 66 MW in the case with solar PV plant with 0.95 power factor and voltage control, 67 MW in
the case of a solar PV with unity power factor without voltage control and 69 MW in the case of the
combined-cycle plant. With stalling of the air-conditioners, the loss of load was respectively 91MW, 96
MW and 97MW. This loss of load was only local, in the area close to the fault.
Chapter 3-E.5 Conclusions
Studies of transient voltage stability performance of a large interconnected transmission system with
high penetration of IBG during faults showed that the IBG did not have adverse impact on the system
performance. On the contrary, system voltages recovered faster with IBG than with conventional
generation and scenarios with IBG showed better damping after a fault. The reasons for better damping
and faster voltage recovery are low inertia of the inverter-based generators and fast control of the
inverters.
The study was performed in an assumption that the IBG units have FRT capability and are not tripped
with transmission system faults such as three phase ground fault on HV network. The IBG models used

247
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

in the study are not intended to evaluate the details of the FRT capability because the control actions
that affect the behaviour of the inverter during the span of a short fault are not modelled in detail in the
generic dynamic models.
The plant representation included protection models with voltage and frequency thresholds and time
delays to indicate the minimum disturbance tolerance requirements.
The study conclusions for the 6 scenarios studied may be summarized in Table 3-E-2.
Even if this study did not identify any voltage stability issues, the issues that may be of a concern are
the following.
1) If an IBG doesn’t have FRT capability (for example small units installed in the distribution system),
a concern is a large loss of generation that may be tripped with transmission faults.
2) A preliminary study performed by the CAISO on the WECC transmission system did not show that
loss of IBG with faults caused any problems in the system voltage stability performance. However,
it may appear to be an issue in other systems or under other system conditions than those assumed
in the study.
3) Inverter-base generation connected to distribution systems may be a cause of high voltages due to
real power injections and lower net load as seen from the transmission system. In case of
underground collector systems that have large charging capacity, voltages in the transmission
system may be even higher (especially if fast voltage support is activated). It is important to evaluate
voltage performance of IBG in each particular case and develop mitigation measures if the study
results indicate unacceptably high voltages caused by these generation units.

In the system model used by the CAISO in this study, the results showed that many of the IBG units
need to have capability to absorb reactive power to avoid excessively high voltages.
More studies need to be performed to investigate the system performance with higher output of IBG,
including studies under light-load conditions that may result in over-generation and adverse impacts on
voltage and frequency.
Table 3-E-2: Summary of dynamic voltage response
Slow voltage
Transient voltage Post-fault
recovery on
Scenario on generator steady state
adjacent load
terminals voltage
buses
1 Solar PV, voltage control, induction motors
High Normal No
don’t stall
2 Solar PV, voltage control, induction motors stall High Normal Yes
3 Solar PV, no volt and Q control, induction
Normal Low No
motors don’t stall
4 Solar PV, no volt and Q control, induction
Normal Low Yes
motors stall
5 Thermal, induction motors don’t stall Low Normal No
6 Thermal, induction motors stall Low Normal No

248
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 3-F EXAMPLE MODEL COMPONENTS FOR FREQUENCY STABILITY

STUDY
Focusing on frequency control and stability, a model for IBG should provide the following optional
functionalities:
 Droop characteristics Δ𝑃 = 𝐾1 Δf
df
 Differential characteristics Δ𝑃 = 𝐾1
𝑑𝑡
 Upper/lower limit, range of control power UL1, LL1
 Upper/lower limit, rate of change of power UL2, LL2
 Additional restrictions due to limited energy available UL3, LL3 battery SOC, rotating energy
 This leads to the following basic structure of the model:

protection
(chapter 3.x)

K2s 0

W=0
UL1 UL 2
dead band current
-Df controller I
+ DP P set
K1 + 0<W<W max + (RMS or EMT,
fdb + chapter 4)
LL1 LL2
W=W max P0

0 selector

W
1
s

W0

Figure 3-F-1 Example model for frequency stability study.

Note: the listed functionalities might not satisfy extreme scenarios like islanded grids exclusively
supplied by inverter-based generation (micro grid)

Table 3-F-1 Model parameters for frequency stability study.

Micro-
description PV PV + battery Fuel cell
turbine
fdb droop characteristic Hz 0.2 0.2
K1 droop characteristic pu/Hz 0…0.3 0…0.3
K2 differential component pu/(Hz/s) 0 0
UL1 upper limit DP pu 0 0…1
LL1 lower limit DP pu 0… -1 0… -1
UL2 upper limit dP/dt pu/s 0 1…10
LL2 lower limit dP/dt pu/s -1…-10 -1…-10
P0 operating point pu 0…1 0…1
W0 initial condition (SOC s 0 0…5000
battery, 0otating energy, …)
Wmax maximum energy content s 0 1000…5000

249
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 3-G HIGHLIGHT OF GRID-CODE REQUIREMENTS IN STEADY STATE

For steady-state analysis, such as power flow analysis, the operating point of IBGs needs to be
considered. Recently the reactive power capability has changed mainly due to the evolvement of the
grid interconnection code. Figures 3-G-1, 3-G-2, and 3-G-3 shows an example of two commercially
available 2500 kW and 3000 kW PV inverters.

(a) 2.5 MW PV inverter (b) 3.0 MW PV inverter

Figure 3-G-1 Example reactive power capability of commercially available inverters
(for nominal voltage)[24] .

(a) 2.5 MW PV inverter (b) 3.0 MW PV inverter

Figure 3-G-2 Example reactive power capability of commercially available inverters
(for 90% of nominal voltage) [24].

Figure 3-G-3 Example reactive power capability of commercially available inverters

(for 110% of nominal voltage) [24].

250
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Chapter 3-G.1.2 Reactive power capability at maximum capacity

Reactive power provision capability requirements in the context of varying voltage. It shall be defined a
U-Q/Pmax-profile within the boundaries of which the generator shall be capable of providing reactive
power at its maximum capacity. The defined U-Q/Pmax profile may take any shape, having regard to
the potential costs of delivering the capability to provide reactive power production at high voltages and
reactive power consumption at low voltages.
Chapter 3-G.1.3 Reactive power capability below maximum capacity
Reactive power provision capability requirements. It shall be defined a P-Q/Pmax-profile that may take
any shape within the boundaries of which the generator is capable of providing reactive power below
maximum capacity.

251
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 3-K PROBABILISTIC APPAROCH FOR DYNAMIC STABILITY STUDY

Probabilistic studies can be used as an additional tool to provide information considering the impact of
IBGs on the dynamic behaviour of power systems. The inherent uncertainties of Renewable Energy
Sources (RES) operation coupled with electricity market and load (including conventional and new types
of load, e.g., electric vehicles) driven uncertainties lead to an overall increase in operating condition-
based uncertainties that are becoming one of the key attributes of modern power systems.
It has been acknowledged that the increasing participation of RES in the power generation mix of the
system can significantly affect its dynamic behaviour following a disturbance due to multiple reasons: i)
RES exhibiting different dynamic behaviour than conventional synchronous generators; ii) the increasing
amount of connected RES results in synchronous generators displacement either by de-loading or
disconnection; iii) changing power flows in the network depending on RES availability, both temporal
and spatial. It has not yet been clarified, let alone quantified, though to what extent each of these
influences affects overall system dynamic behaviour [25]. Due to spatial and temporal uncertainties
associated with RES operation, in particular, it is neither practical nor feasible to identify easily “worst
case scenarios” that are representative enough to capture the overall impact of RES on power system
dynamics. Probabilistic methods can be used as a tool to assess various aspects of the dynamic
behaviour of power systems with RES [26]-[29].
Chapter 3-K.1 Probabilistic methodology
A number of Ns Monte Carlo (MC) simulations are performed following the procedure explained below.
A power system dynamic model suitable for the respective stability studies is required, taking into
consideration the connected RES with the associated controllers. Since system stability can be also
significantly affected by pre-fault operating conditions, the uncertainties concerning system loading and
wind/PV generation for a 24-hour period are considered. Moreover, the uncertainties related to the
disturbance (i.e. fault location and duration) are also accounted for. All random variables are sampled
according to appropriate probability distributions describing the uncertain parameter behaviour. The
sampling of the respective distributions is performed separately for each load and each RES unit in the
system to consider independent behaviour of loads and RES units. After the uncertainties in loading and
RES contribution to power generation have been accounted for, an Optimal Power Flow (OPF) problem
is solved to determine the output of conventional generators. The dispatch obtained from OPF also
determines the amount of disconnection of conventional generation and consequently system inertia
reduction due to increased RES penetration.
The proposed framework offers full flexibility when accounting for relevant uncertain parameters. The
sampling of uncertain factors can be done according to any probability distribution based on historical
data, prior knowledge or forecast and solving the OPF problem can include any number of additional
security constraints associated with RES, as for example proposed in [30].
Statistical analysis of the obtained results can be useful in drawing conclusions and identifying
tendencies considering the dynamic behaviour of the system, from the abundance of available simulated
data. For example, plotting Cumulative Distribution Functions (CDFs) of the random variables (i.e.
relevant stability indices) can reveal the associated probabilities for the whole range of values each
index takes.
Chapter 3-K.2 Uncertainty modelling
The effect of the intermittent behaviour of wind generators and Photo-Voltaic (PV) units is modelled
using appropriate Probability Distribution Functions (PDFs) [28],[29]. The uncertainties related to the
conventional part of the system such as system loading as well as the investigated disturbances are
also included in this study. For system loading and PV generation, typical daily curves are initially used,
obtained from national grid data and the literature [29], respectively. The system load is considered to
vary from 0.6 to 1 p.u. The hour of the day is sampled randomly and for each hour, an appropriate PDF
is used to model the uncertainty within the hour. For system loading a normal distribution with mean
value 1 p.u and standard deviation 3.33% is used, while for PV generation, a beta distribution with a
and b parameters 13.7 and 1.3. For wind generation, a Weibull distribution is used with φ =11.1 and
k=2.2, assuming constant mean wind speed for every hour within the day. For each load within the
system and each DER unit, the PDFs are sampled separately, to consider independent behaviour of
each unit. For example, assuming the total load is 0.8 p.u. for a given hour of the day, one load might
have a value of 0.75pu and another 0.85pu.
Frequency stability studies are performed for three operating conditions of the network, nominal loading,
60% and 40% loading of the network.

252
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Considering transient stability, the fault location is also sampled randomly following a uniform distribution
along each line within the system. Moreover, the uncertainty of the fault duration is modelled using a
normal distribution with mean value 13 cycles and standard deviation 6.67%. Three phase self-clearing
faults are considered as disturbances in all simulated cases for transient stability studies.
More information considering the modelling of uncertainties for transient and frequency stability studies
can be found in [31] and [32] respectively.
The number of simulations Ns is chosen by keeping the error of the sample mean less than a certain
threshold (for example 5% for transient stability studies), for 99% confidence interval. The error of the
sample mean is calculated using Equation 3-K-1, where Φ-1 is the inverse Gaussian CDF with a mean
of zero and standard deviation one, σ2 is the variance of the sampled random variable, δ is the
confidence level (i.e. 0.01 for this study) and XN is the sampled random variable with N samples [33].
δ σ2 (XN)
Φ-1 (1- 2 )√ N
eX̅N =
̅̅̅̅
𝑋𝑁
Equation 3-K-1

Chapter 3-K.3 Test system

The test network used, is a modified version of the IEEE 16 68 bus, 16 machine reduced order equivalent
model of the New England Test System and the New York Power System (NETS – NYPS). The
conventional part of the test network is adopted from [34] and RES are added at buses shown in Figure
3-K-1 for transient stability studies. For frequency stability studies renewables are only added in NETS
and NYPS area. Two types of RES units are connected on each bus: Doubly Fed Induction Generators
(DFIGs), representing wind generators and Full Converter Connected (FCC) units, representing both
wind generators and PV units.
NEW ENGLAND TEST SYSTEM NEW YORK POWER SYSTEM
G7 G5 G3 G2 G12 G16
G13
3 2 16
G6 G4 13 39 50
59 43
7 5 12 18
62 63 44 L71
6 4 60 17 45
65 58 36
23 20 35 51
64 L41
22 34 L66
24 19 66 61 33 46
21 67 56 L42 G15
57
G9 68 32 15
52 30 38 49
55 31
9 37 L43 11 42
28
27 L44 G11
10
29 26 25 54
L45 G10 G14
47
1 53 48 40 14
8 L69
G8 G1
41

Figure 3-I-1 Modified IEEE 68 bus test network.

The test network consists of 16 generators (G1-G16) in five interconnected areas. NETS consists of G1
- G9, NYPS of G10 to G13 and the three areas are represented by equivalent generators G14, G15 and
G16, respectively. Standard 6th order models are used for all synchronous generators. G1-G16 are
equipped with either slow IEEE DC1A dc exciters or fast acting static exciters type IEEE ST1A and G9
is equipped with a Power System Stabilizer (PSS). All generators are also equipped with generic
governors, representing gas, steam and hydro turbines.
A generic type 3 model, suitable for large scale stability studies is used in this study to represent DFIGs.
The model has a structure similar to the one proposed by WECC [35] and IEC [36] and is available in a
commercially available software. It takes into consideration the aerodynamic part and the shaft of the
wind turbine/generator as well as the pitch control of the blades. The rotor side converter controller is
also modelled including relevant limitations, ramp rates and protection mechanisms, such as the
crowbar. The DFIG is represented by a typical 2nd order model of an induction machine neglecting the
stator transients and including the mechanical equation. Therefore, the model represents all the relevant
parts that influence the dynamic behaviour of DFIGs.

253
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Similarly, a type 4 wind generator model is used to represent all FCC units. Both wind generators and
PV units can be represented by a type 4 model in stability studies, since the converter can be considered
to decouple the dynamics of the source on the dc part. This is also suggested by the WECC Renewable
Energy Modelling Task Force [37], which develops a PV model by slightly modifying the type 4 wind
generator model. The FCC model used in this study has a similar structure to [35],[36] and is available
in a commercially available software.
Both DFIGs and FCC units are treated as aggregate units. Each RES unit model has a 2 MW power
output and the number of connected units is varied to determine the output of the aggregate unit.
Furthermore, all RES units are considered to have Fault Ride Through (FRT) capabilities and remain
connected during the fault. The amount of connected RES for each area of the system, i.e. the installed
capacity of RES, is given as a percentage of the total installed conventional generation capacity of that
area before adding any RES. Considering the total RES installed capacity, approximately 66.67% are
assumed to be DFIG wind generators and the remaining 33.33% are FCCs. FCCs are further considered
to be 30% wind generators and 70% PV units.
Chapter 3-K.4 Frequency stability
The replacement of conventional synchronous generation with converter connected generation reduces
system inertia and primary frequency response of the system [38]-[40]. In this reduced inertia system,
the most challenging periods can be off-peak hours when fewer conventional synchronous generators
are scheduled. During these hours system inertia is low and few power plants are available to provide
frequency response services. This can have a significant impact on frequency dynamics and power
system security.
It is a general practice to study the collective performance of all generators in the network for frequency
stability. In such cases all generators in the network are typically aggregated into an equivalent
generator. This equivalent generator has an inertia constant equal to the sum of the inertia constants of
all generators, and it is driven by the combined mechanical outputs of the individual turbines. The effects
of the system loads are lumped into a single damping constant D. The speed of the equivalent generator
represents the system frequency. Frequency nadir is dictated by the inertia of the system and primary
frequency response.
These studies present a framework to perform a robust analysis of frequency excursion in a reduced
inertia system taking into account the stochastic and intermittent patterns of renewable generation. In
the network, all generators (synchronous and renewable) and associated controls are modelled using
full dynamic models for generators and appropriate models for controllers. The methodology establishes
the critical penetration levels of RES and inertia limits in the studied system for the grid frequency
stability. The change in frequency nadir due to a reduction in system inertia is also quantified.
Chapter 3-K.4.1 Case study
All power systems exhibit frequency excursions following an active power disturbance in the network.
Frequency nadir (the minimum frequency following an active power disturbance) is governed by the
system inertia and governors response. In the absence of speed governors, frequency nadir is
determined by the size of active power disturbance and inertia of the system.
To establish the effect of reduction in inertia and primary frequency response on frequency nadir in a
reduced inertia system following case studies are developed.
i. Nominal loading and no renewable energy sources in the network
ii. At the nominal loading of the network, 30% of synchronous generation is replaced by renewables
in two areas of the network. This reduces the inertia of the system and primary frequency response.
iii. With 30% renewables and 70% synchronous generation in the network, loading of the network is
reduced to 60%. 28% reduction in the load is balanced by disconnecting synchronous generation,
and remaining 12% reduction in the load by de-loading synchronous generators. This further
reduces the inertia and primary frequency response, and increases the nominal penetration of
renewables to 46%.
iv. With 30% synchronous generation replaced by renewables, loading of the network is reduced to
40%. 42% reduction in the load is balanced by disconnecting synchronous generation and 18% by
de-loading generators.

This reduces the inertia of the system even further, and the nominal penetration level increases to 52%.

254
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The nominal penetration level ( 〖 NPL 〗 _a) of RES is defined as installed capacity (the
nominal/maximum power) of renewables based on the total generation (synchronous+renewables) in
the network, and is calculated using Equation 3-K-2.
𝒅
𝚺𝒏=𝟏 𝑷𝟎𝑹𝑬𝑺,𝒏
𝑵𝑷𝑳𝒂 = 𝒈 𝒅
𝚺𝒎=𝟏 𝑺𝑺𝑮,𝒎 . 𝒑𝒇𝑺𝑮,𝒎 + 𝚺𝒏=𝟏 𝑷𝟎𝑹𝑬𝑺,𝒏
Equation 3-K-2
where S_(SG,m)^ is the toal synchronous generation in the network at particular operating condition
and P_(RES,n)^0 is the nominal/ maximum power output of renewables. Table 3-I-1 shows the nominal
penetration level of RES for each case study and average inertia of the system for each area.
Table 3-I-1 Inertia values and nominal penetration levels.
𝑵𝑷𝑳 Average ‘H’ sec
NETS & NYPS HNETS HNYPS HEq HSys
Case study i 0 3.9 7.9 11.1 7.95
Case study ii (nominal loading) 30% 2.7 5.5 11.1 6.8
Case study iii (60% loading) 45% 1.64 3.32 7.8 4.1
Case study iv (40% loading) 52% 1.28 2.26 6.6 2.86

Simulations including the uncertainties with RES generation and loads are performed to establish the
effect of reduction in inertia with and without speed governor system.
Chapter 3-K.4.2 Results and analysis
The inertia constant of a synchronous generator defines its response to any changes in power balance.
The inertia constant of a generator (H) can be viewed as the time the generator can maintain full
electrical power output without any mechanical power input.
The average inertia of the network H_sys is defined by Equation 3-K-3, where H_i, S_i and n denote the
inertia constant of each generator, the generator rating and the number of units respectively
𝒏
𝚺𝒊=𝟏 𝑺𝒊 𝑯 𝒊
𝑯𝒔𝒚𝒔 = 𝒏
𝚺𝒊=𝟏 𝑺𝒊
Equation 3-K-3
The system inertia H_sys of 16 machine system, with nominal loading as given in [41] is calculated
using Equation 3-K-4. Inertia of three areas, inertia of NETS H_NETS, inertia of NYPS 〖 H〗_NYPS
and inertia of equivalent areas H_Eq (part of the test system with generators G14, G15 and G16) is also
calculated.
Replacement of 30% of synchronous generation with renewables in NETS and NYPS area reduces the
overall inertia H_sys of the system, H_NETS and H_NYPS.
Inertia constant H can be expressed in terms of moment of inertia J
𝑱𝝎𝟐𝒎
𝑯 =
𝟐𝑽𝑨𝒃𝒂𝒔𝒆
Equation 3-K-4
where J is the moment of inertia of generator in kg.m2 and
ωm is the angular velocity of the rotor in mech. rad/s.
Total moment of inertia J of the system is calculated for each case study case. As synchronous
generation reduces in the network, total moment of inertia of synchronous generation also reduces.
H_sys calculated using (3) is reduced in the same proportion as J is reduced in each study case.
Table 1 illustrates the impact of increased penetration of RES on the network inertia. It can be seen that
the network has low, medium and high inertia areas in the network. H_NETS is only 3.9 s without RES
and 52% penetration of RES reduces H_NETS to 1.28 s. H_NYPS is 7.9 s without RES and this reduces
to 2.26 s (67% reduction) with 52% RES penetration. The overall inertia of the system is 7.95 s and this
decreases to 2.86 s when the penetration of RES increases to 52%.
1) Defining critical inertia levels and penetration levels of RES

255
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

As the inertia of the system reduces due to disconnection of synchronous generation and increased
nominal penetration of renewables, small variations in the active power cause bigger variation in
frequency nadir. In these studies, a threshold of 500mHz for frequency nadir variation is considered.
When ±10% variation in active power disturbance leads to a variation in frequency nadir greater than
500 mHz; the inertia of the system and the nominal penetration level of RES in the network is considered
to be critical. Variation in active power disturbance and threshold for frequency nadir can be customized.
2) Impact of increased penetration of RES on system inertia for each studied operating condition
Modelling the stochastic and intermittent behaviour of the wind and solar generation, and uncertainties
in the loading forecast changes the instantaneous penetration level of RES in NETS and NYPS at each
operating condition in continuous manner as shown in Figure 3-K-2. Instantaneous penetration of
renewables PLia of the area is calculated as
𝒅
𝚺𝒏=𝟏 𝑷𝑹𝑬𝑺,𝒊𝒏
𝑷𝑳𝒊𝒂 = 𝒈 𝒅
𝚺𝒎=𝟏 𝑷𝑺𝑮,𝒊𝒎 + 𝚺𝒏=𝟏 𝑷𝑹𝑬𝑺,𝒊𝒏
Equation 3-K-5
The subscript i=1…1000 denotes the dynamic simulations number, a =1…5 the system area, g the
generator number and d the wind and solar farm number. Where P_(RES,in) is the active power
produced by the wind and solar farm in the area, P_(SG,im) is the power generated by each
synchronous generator in the area.
Simulations are performed by introducing an active power disturbance of 1340MW (a simultaneous
outage of G2, G7 and G10) for each operating point. Due to uncertainties in loads and renewables, the
power output of generators varies by ±10%. Therefore, active power disturbance varies within ±10% of
1350MW. Frequency nadir following this active power disturbance is recorded on each tie-line between
NETS and NYPS area. Power transfer through line L41 is the highest; therefore, frequency nadir at L41
is discussed throughout this work It can be seen from Figure 3-K-2, penetration levels of RES vary
between 10%-30%, 18-46% and 28%-64% at the nominal, 60% and 40% loading respectively.

Figure 3-K-2 Inertia of system plotted against instantaneous penetration of renewables.

For the same installed capacity of RES, the instant percentage penetration of RES and the range of its
variation increases with the increase in disconnection of synchronous generation. It can also be
observed from Figure 3-K-2 that penetration levels of RES at 60% loading overlap penetration levels at
nominal loading and 40% loading. However, the band of variation of frequency nadir for each operating
condition is distinct.
In this analysis, the inertia of the system is constant for one operating conditions and only varies with
the operating condition of the network. The small variations in the active power disturbance result in
small variations in frequency nadir within one operating condition. The variation range in frequency nadir
within one operating condition increases as the inertia of the system reduces.

Figure 3-K-3 Inertia of system plotted against instantaneous penetration of renewables.

256
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-K-4 pdfs of frequency nadir for three studied operating conditions .
It is to be noted that the same active power disturbance results in different values of frequency nadir for
the same penetration levels at different operating conditions of the network. For 30% penetration level
of RES, frequency nadir is approximately 49.67 Hz at the nominal loading, moves to 49.46 Hz at 60%
loading. It further drops to 49.3 Hz at 40% loading. Since the number of synchronous generation varies
in three studied operating conditions, the inertia of the system is different in each operating condition.
The frequency nadir decreases as the inertia of the system reduces.
To determine the most probable values for three studied operating conditions, pdfs of frequency nadir
are plotted, shown in Figure 3-K-4. It can be seen that frequency nadir is well within the operation limits,
±200 mHz, without renewables at the nominal loading of the network. Replacement of 30% of
synchronous generation with RES moved the most probable value of frequency nadir to 49.7 Hz and
decreased the frequency nadir by 0.3 Hz. At 60% loading, disconnection of 40% synchronous generation
shifts the mode of the pdf to 49.48 Hz, further increasing the drop in frequency nadir to 0.52 Hz. At 40%
loading the most probable value of the pdf 49.38 resulting drop in frequency 0.62Hz.
It can also be observed that as the inertia of the system due to disconnection of synchronous generators
decreases the pdfs have bigger deviation. Small changes in the operating point within the same
operating condition lead to the different values of nadir.
Figure 3-K-3 and Figure 3-K-4 show that the range of variation in frequency nadir at 40% loading is
much bigger than the range at 60% load and full load. At 40% loading, frequency nadir varies between
49.45-48.8. It is the critical inertia limit for this size of disturbance as small changes in the active power
disturbance lead to a bigger (> 500mHz) variation in frequency nadir.
The statutory frequency limits for the UK system is ±500 mHz. For 60% loading and 40% loading the
most probable values are outside this limit, synchronous generation in the system is increased (SG
connected) by 10% for each operating condition. This increases the inertia of the system by 7%.
Simulations are performed with the same uncertainties. It can be seen from Figure 3-K-5 that 7%
increase in the inertia moves the most probable value of 60% loading to 49.55 Hz.

Figure 3-K-5 pdfs of frequency nadir for 60% and 40% operating.
The most probable value of 40% loading pdf with 10% increase in synchronous generation is 49.45 Hz.
Therefore, synchronous generation is further increased by 10%, this moves the mode of the pdf (thick
dashed line) within the threshold bound.

257
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-K-6 Frequency at 40% loading for different values of inertia.

Figure 3-K-6 shows frequency nadir plotted for 40% loading of the network with three values of inertia.
It could be observed that for 10% increase in synchronous generation inertia, i.e., 7% increase in the
system inertia, the frequency of the system is below 49.5Hz, but the frequency variation range is
significantly reduced. Small variations in the active power disturbance do not affect frequency nadir
significantly. For further 10% increase in synchronous generation, frequency nadir is within ±500mHz
limits for most of the penetration levels of RES. Furthermore, It can be observed that for the same inertia
when % instant penetration of RES increases frequency nadir improves because more power is supplied
by RES and generators are relatively lightly loaded, capable of providing more frequency response.
It can be concluded therefore, that when the online synchronous generation is less than 50% of total
generation, small variations in the active power can lead to big variation in frequency nadir; this is the
critical nominal penetration level of RES for this system.
c) Impact of spinning reserves
To investigate the effect of spinning reserves on frequency nadir, two more case studies are performed.
v. With 30% synchronous generation replaced by renewables, loading of the network is reduced
to 60%. Entire (40%) reduction in the network load is met by de-loading generators. This study case has
the same inertia as case study ii. However, spinning reserves are increased.
vi. With 30% synchronous generation replaced by renewables, loading of the network is reduced
to 40%. 28% reduction in the network load is balanced by disconnecting synchronous generation and
42% reduction in load by de-loading generators. This study case has the same inertia as case study iii.
However, the system has more spinning reserves.
In Figure 3-K-7 case studies ii and v have the same inertia. However, spinning reserves in case study v
are increased by 200%. Case studies iii and vi have the same inertia, but case vi has 325% more
spinning reserves due to the de-loading of the network. The inertia of the system is higher in case studies
(ii, v) than the case studies (iii, vi).
Probability distribution

case study iii

100
case study vi
case study ii
50
case study v

0
49.4 49.45 49.5 49.55 49.6 49.65 49.7 49.75 49.8
Frequency nadir 'Hz'

Figure 3-K-7 pdfs of frequency nadir for 60% loading and 40%.
It can be observed that the most probable value for the same inertia of the system has improved by
(0.06 Hz) when the spinning reserves are increased from (1500MW to 3000 (nearly 200% increase). In
pdf iii and vi the system has same inertia, spinning reserves are increased from (1000MW to 3350MW
(increased by 325%), however, the improvement in frequency nadir is 0.05Hz.
d) Impact of change in inertia
The case studies ii, iii and iv are performed without governors. It can be observed that 30% reduction in
inertia moves the most probable value from 49.48Hz to 48.5Hz resulting in 0.98 Hz increase in frequency
nadir. Further disconnection of synchronous generation at 60% loading reduces inertia by 48%. This
shifts the mode of the pdf to 47.3Hz. At 40% loading, 67% reduction in the inertia of the system shifts
the most probable value to 46.1Hz. Figure 3-K-9 shows the increase in the drop of frequency nadir with
the reduction in the system inertia that is linear.

258
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 3-K-8 pdfs of frequency nadir for three operating conditions without governors.

Figure 3-K-9 Variation in frequency nadir corresponding to variation in inertia.

This work presented an approach to perform a robust analysis of the effect of reduction in inertia and
primary response of frequency excursions in a large system. The analysis successfully established the
reduction in inertia limits and the critical nominal penetration level of RES for the grid frequency in the
test system. The results established when nominal penetration level of RES generation is higher than
the online synchronous generation; small changes in the active power disturbance can lead to much
bigger variation in frequency nadir.
The performed analysis also shows that amount of spinning reserves required to mitigate the frequency
nadir as the inertia of the system decreases.
Chapter 3-K.5 Transient stability
Chapter 3-K.5.1 Transient stability indices
The percentage of cases that each generator is exhibiting instability (i.e. the probability of instability of
each generator) is a measure of how stable each generator is. Critical generators and generator groups
can be identified in this way. The overall impact of RES on transient stability can be investigated by
observing changes in the probability of instability of the generators. The effect of added uncertainties
and of the different dynamic behaviour of RES units on system stability is identified in this way.
Transient Stability Index (TSI) is an index considering the stability of the whole system for a specific
contingency [42]. Negative value of the TSI means that the case is unstable since the difference between
rotor angles of at least two generators for the same time instance is more than 360 degrees.
360 -  max,i
TSI i  100
360 +  max,i
Equation 3-K-6
where δmax,i is the maximum rotor angle deviation between any two generators in the system for the
same time instance.
An additional measure is used to identify the impact of conventional generation disconnection on
transient stability of the system. When RES is connected, this will eventually lead to conventional
generation disconnection. The generator spare capacity SCig, defined in Equation 3-K-7, is a measure
of the conventional generator loading which is an important parameter considering transient stability.
Moreover, when spare capacity is kept low this means that in general more conventional generation will
be disconnected leading also to lower system inertia. The effect of conventional generation spare
capacity on transient stability is also investigated in this study.

PSG ,ig
SCig  1 -
S SG ,ig pf SG ,ig
Equation 3-K-7

259
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

where PSG,ig is the power produced by each generator (determined by OPF), SSG,ig is the apparent
power of each generator after considering any disconnection and pfSG,ig is the nominal power factor.
Chapter 3-K.5.2 Identifying generator groupings after a disturbance
A hierarchical clustering method is applied to determine the groups of generators exhibiting instability in
each simulated contingency. The agglomerative (bottom up) method is applied with a cut-off value of
360 degrees, since this is considered to be the transient stability limit. Euclidean distance between the
data points is used as the similarity measure and complete linkage is chosen as the linkage criterion.
This results in groupings in which generators of one group have a minimum of 360 degrees rotor angle
difference with generators of another group. Therefore, hierarchical clustering is used to identify the
unstable generators as well as the generator grouping patterns for each simulated contingency [43].
The percentage of cases that each generator is exhibiting instability (i.e. the probability of instability of
each generator) is a measure of how stable each generator is. Critical generators and generator groups
can be identified in this way.
The impact of RES on transient stability can be investigated by observing changes in the probability of
instability of the generators. The effect of added uncertainties and of the different dynamic behaviour of
RES units on system stability is identified in this way. The impact of different network topologies on
transient stability can be also identified in a similar manner.
Chapter 3-K.5.3 Effect of increased RES penetration
The effect of RES on transient stability is investigated in this clause using the probabilistic methodology
described above. In this study, all RES units are considered to have FRT capability. The control logic
followed by RES units when a disturbance happens is to reduce the active power and increase the
reactive power output to provide support to the system. In general, this operation can improve the
transient stability of the system. However, the pre-fault operating conditions are also affected by the
uncertainties introduced by RES, which can lead to either more or less favourable operating conditions
for specific generators. Furthermore, the disconnection of conventional generation to account for the
connection of RES tends to have a negative impact on transient stability, due to the reduction in inertia
and nominal capacity of the synchronous generators. Therefore, it is not straightforward to determine
the overall impact of RES on transient stability following a deterministic approach. A probabilistic
approach however, can quantify more easily the overall impact of RES on stability.
As mentioned earlier, after considering the uncertainties OPF is solved to determine the conventional
generators dispatch PSG,ig. The nominal capacity of each generator SSG,ig is then adjusted by
considering 15% spare capacity according to (3). In case the resulting SSG,ig from (3) is larger than the
initial nominal power of the generators, it is set to that initial nominal value. This means that there is no
room for conventional generation disconnection in this case. The disconnection of conventional
generation due to both load variations and RES penetration is considered in the following way: Since
the generators are considered equivalent generators, reducing the nominal power, is equivalent to a
reduction in the moment of inertia of the power plant and an increase in the generator reactance.
The following Test Cases (TCs) are initially studied: In the base case (TC1) low amount of connected
RES is considered. In TC2, all the RES units are disconnected and therefore the associated
uncertainties are also not considered. In TC3, high amount of RES is considered to be connected. TC1,
TC2 and TC3 are used to study the impact of RES on transient stability. For the given TCs the error of
the sample mean varies from 3.4% for TC3 up to 5% for TC3 for 99% confidence interval.
The TCs are described below;
TC1: System with installed capacity of RES equal to 20% of the nominal capacity of the system
TC2: System with no RES
TC3: System with installed capacity of RES equal to 60% of the nominal capacity of the system.
In Figure 3-K-10 CDFs for the TSI for TC1 (low amount of connected RES), TC2 (no RES) and TC3
(high amount of connected RES) are presented. As explained before, the spare capacity of conventional
synchronous generation is kept always constant at 15% (including for TC2), which means that there is
some generation disconnection considered even for this case due to load variations within the day.
Relatively high spare capacity combined with FRT capability of RES leads to a reduction of the total
number of instabilities in the system with RES. This is reflected by observing the probability of the TSI
to have a negative value which is 9.5% for TC3, 9.7% for TC1 and 12.9% for TC2. Moreover, TSI values
above 55 tend to appear with smaller probability as the amount of RES increases. This indicates the
occurrence of larger oscillations within the stable cases, especially for TC3.

260
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

1
TC1 20% RES
TC2 no RES
0.9
TC3 60% RES

0.8

0.7
Cumulative Probability

0.6

0.5

0.4

0.3

0.2

0.1

0
-100 -80 -60 -40 -20 0 20 40 60 80
TSI

Figure 3-K-10 CDFs of TSI for TC1 and TC3.

The critical generators of the system are also identified according to the number of times each generator
becomes unstable. The number of cases each generator is exhibiting instability is identified using the
hierarchical clustering approach presented above. Initially the grouping patterns are defined and
afterwards the instabilities for each generator are observed, i.e. the number of cases each generator
belongs to any unstable group. This means that for certain cases more than one generator might be
unstable. In Figure 3-K-11, the results for the probability of instability for each generator are shown, i.e.
the number of instabilities divided by the total number of simulated cases for each TC. G9 and G11 are
the most unstable generators in NETS and NYPS area, respectively. This can be attributed to the fact
that they are the ones having relatively smaller moment of inertia constant. In general, generators in the
NETS area exhibit more instabilities than those in the NYPS and external areas.
8
Probability of instability (%)

7
6
TC1 (20% RES)
5
4 TC2 (no RES)
3
TC3 (60% RES)
2
1
0
G1 G2 G3 G4 G5 G6 G7 G8 G9 G10 G11
Generator number
Figure 3-K-11 Probability of instability of each generator.
The probability of instability for most generators becomes smaller as the amount of connected RES is
increasing with descending order from TC2 to TC1 and TC3, for this specific system and studied
operating conditions. The probability of instability of G10, however, is very slightly increasing
(approximately 0.1%) as the amount of connected RES is increasing from TC1 to TC3. This behaviour
is further discussed below with respect to the effect of synchronous generator spare capacity. Moreover,
when comparing the same simulated contingencies (i.e. same fault location/duration and system
loading) between TCs, some of the cases become unstable while some others become stable. For

261
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

example, when comparing TC2 to TC3 there are 106 cases (out of the 6000) that change from unstable
to stable and 92 that change from stable to unstable, i.e., net increase (14 cases) in stable cases.
An additional study is performed in [31] where the amount of connected RES is increased gradually from
20% to 30%, 40%, 50% and finally 60% of the nominal capacity of the system. While the differences in
the probability of instability as the amount of connected RES are not very significant, a turning point at
around 50% can be observed after which the positive effect of RES FRT control starts to reduce due to
the high amount of conventional generation disconnection. Furthermore, the probability of the TSI to
observe relatively large positive values tends to decrease for 60% RES, indicating the existence of larger
oscillations for the stable cases (as also observed in Figure 3-K-10).
It should be mentioned that this turning point is most probably network specific. It might be affected by
network specific characteristics such as the network topology and special operating conditions and
limitations that might be present in different networks. This is even more prominent considering transient
stability where certain local conditions might cause the instability of specific generators. However, using
the probabilistic methodology described in a given network, can identify the point where deterioration of
the overall system stability starts to be observed.
To visualize the amount of conventional generation disconnection, Figure 3-K-12a is presented where,
the amount of conventional generation disconnection is plotted against the RES instantaneous
penetration level (as defined in Equation 3-K-5) for NETS area for TC1 and TC3. There is a linear trend
between the two variables, i.e. as the penetration level increases the amount of conventional generation
disconnection is also increasing. However, the values are dispersed for specific instantaneous
penetration levels within a range of approximately 15% due to different system loading level, which
varies from approximately 0.6 to 1 p.u. in this study. In Figure 3-K-12b, the CDFs of the amount of
conventional generation disconnection for different amount of connected RES (0%-60%) are presented.
As the amount of connected RES increases, the amount of conventional generation disconnection
generally increases. The amount of disconnection varies from 15% to 54% for 20% RES, from 21% to
71% for 50% RES and from 23% to 77% for 60% RES.
a)
80
Syn. Gen. Disc. NETS (%)

20% RES
60% RES
60

40

20

0
0 10 20 30 40 50 60 70
Instantaneous Penetration Level NETS (%)
b)
1
20% RES
0.8 no RES
Cumul. Prob.

30% RES
0.6 40% RES
0.4 50% RES
60% RES
0.2

0
0 10 20 30 40 50 60 70 80
Synchronous generation disconnection NETS area (%)
Figure 3-K-12 a) Variation of amount of synchronous generation disconnection with instantaneous
penetration level and b) CDF of amount of synchronous generation disconnection.
Moreover, from the CDFs it is shown that there is a 40% probability that the amount of conventional
generation disconnection is larger than 20% for the case without RES and it becomes 50% for 60%
RES. The shape of the CDFs is also changing as the amount of connected RES increases. In the
case without RES the only variable affecting generation disconnection is the system loading. From
20% to 60% connected RES, the curve is slowly changing shape, indicating the increasing impact of
RES intermittency on conventional generation disconnection.

262
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Chapter 3-K.5.4 Generator grouping patterns

The impact of IBGs on transient stability can be also investigated by observing the unstable generator
groupings that are formed after a disturbance leading to instability. In Table 3-K-2 the generator groups
for the case with no RES and 20% of the nominal capacity of the system connected RES are presented.
In general, the introduction of RES causes the generator grouping patterns to change, either by
introducing new patterns or changing the frequency that patterns appear. This difference is caused by
the uncertainties introduced by RESs as well as the dynamic behaviour of RES units.
In Table 3-K-2 the most significant identified unstable patterns are presented. The unstable groups are
shown in bold letters and the order in which they are presented is the order they lose synchronism [41].
The percentages shown refer to the total number of unstable cases for each TC. New grouping patterns
appear for different TCs and the frequency of appearance of common patterns is also changing. Pattern
1, in which G9 alone goes unstable, is the most common pattern in all cases. The frequency of
appearance, however changes between TCs. The same applies for other patterns as well, such as
pattern 2 and 3. Considering TC2, there are some patterns such as 1, 2, 8, etc. that appear less
frequently, while others such as 3, 5, 12, 13 appear more frequently. This fact along with the fact that
13 new patterns appear in TC2 (not all shown in the table due to very small frequency of appearance)
leads to the conclusion that RES units cause an overall change in the power system dynamic behaviour.
Table 3-K-2 Generator grouping patterns.
Pattern Grouping 20% RES (% of appearance) No RES (% of appearance)
1 (G9)/(G1-G8,G10-G16) 49.38 42.96
2 (G11)/(G1-G10,G12-G16) 22.12 19.91
3 (G2-G9)/(G1,G10-G16) 2.65 7.63
4 (G4-G5)/(G6-G7)/(G1-G3,G10-G16) 1.95 1.35
5 (G3)/(G1-G2,G4-G16) 3.01 4.04
6 (G4-G7,G9)/(G3)/(G1-G2,G8,G10-G16) 0.88 -
7 (G4-G7)/(G1-G3,G8-G16) 4.78 5.69
8 (G4-G5)/(G1-G3,G6-G16) 4.60 1.20
9 (G1-G9)/(G10-G16) 1.59 2.40
10 (G8)/(G1-G7,G9-G16) 2.12 1.80
11 (G5)/(G1-G4,G6-G16) 0.88 0.15
12 (G1-G10)/(G11-G16) 0.35 2.69
13 (G10)/(G1-G9,G11-G16) 0.18 2.25
14 (G1-G9)/(G10)/(G11-G16) - 2.25

Chapter 3-K.5.5 Effect of synchronous generation disconnection due to RES on transient stability
To further investigate the effect and importance of synchronous generation spare capacity, the
simulations for TC1 are repeated for fixed spare capacity of 10% and 20% for the hour of the day that
the maximum load occurs. Moreover, TC2 (no RES) with no synchronous generation disconnection (i.e.
keeping the initial SSG,ig values) for the same hour is also considered. This approach for TC2 means
that all generators are only de-loaded, which is the most favourable assumption considering transient
stability. 1000 cases are simulated for this comparison since the variation of the load within the day is
not considered. CDFs for the TSI are presented in Figure 3-K-13, showing that there is significant impact
of the spare capacity in transient stability. The probability of instability (i.e. TSI to exhibit negative value)
increases with reduction in spare capacity from approximately 4.8% to 9.7% and 14.7% for spare
capacity 20%, 15% and 10%, respectively. For the case without RES (TC2) the probability of instability
is 7% without considering any disconnection which is between the values with spare capacity 15% and
20%. Therefore, it is important to keep the spare capacity to a high value (e.g. above 15% for this specific
system) by giving priority to de-loading instead of disconnecting some of the synchronous generators to
ensure the transient stability of the system does not deteriorate.

263
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

1
no disconnection
10% additional spare capacity
0.9
15% additional spare capacity
20% additional spare capacity
0.8

0.7
Cumulative Probability

0.6

0.5

0.4

0.3

0.2

0.1

0
-100 -80 -60 -40 -20 TSI 0 20 40 60 80

Figure 3-K-13 CDFs and TSI for different spare capability.

Chapter 3-K.5.6 Model comparison using probabilistic studies

The probabilistic method presented above can be used to investigate the influence that IBG models can
have on overall system stability. The aim is to identify how important the level of modelling detail is and
to what extent the results for stability studies can be influenced by similar but slightly different modelling
approaches for IBGs. There is a level of uncertainty present in power systems which is increasing
significantly due to the introduction of RES. Probabilistic studies can shed light on the extent that IBG
models influence the results of overall system stability and therefore inform about the importance of
modelling detail level, parameters, etc.
Two generic RMS models, suitable for large scale stability studies for both Type 3 and Type 4 RES units
are investigated. The first approach (Generic model 1) is a generic approach similar to WECC and IEC
approaches and is available in the library of a commercially available software. The second approach
(Generic model 2) are models based directly on IEC 61400-27-1 [36], also available in a commercially
available software. For both modelling approaches, a Type 3 model is used to represent DFIGs and a
Type 4 model is used to represent all Full Converter Connected (FCC) units. FCC units can be either
wind generators with full converter interface or PV units.
6000 Monte Carlo simulations are conducted, considering power system uncertainties having to do with
the system loading, wind and PV production, fault location and fault duration as described before. The
same simulations are carried out using both models and the results for the TSI are presented in Figure
3-K-14. Moreover, the probability of instability of each generator is calculated for the two modelling
approaches and presented in Figure 3-K-15.

264
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

1
Generic model 1
0.9 Generic model 2 (IEC)

0.8

0.7
Cumulative Probability

0.6

0.5

0.4

0.3

0.2

0.1

0
-100 -80 -60 -40 -20 0 20 40 60
TSI
Figure 3-K-14 CDFs and TSI for two different IBG models.

6
Probability of instability (%)

5
Generic
4
model 1
3
Generic
2 model 2
(IEC)
1

0
G1 G2 G3 G4 G5 G6 G7 G8 G9 G10 G11
Generator number
Figure 3-K-15 Probability of instability of different generators for two different IBG models.
The probability of the TSI to exhibit negative value is changing from approximately 9.5% to 8.97% (32
more unstable cases) when using Generic model 1 and 2 respectively. While this is not a very significant
difference, using different models can affect the results of performed studies on a system level
considering transient stability. More specifically, the resulting probability of each specific generator to
exhibit instability is also affected. All the observed differences are lower than 1% with G4-G7 and G9
exhibiting the largest differences. In general, using Generic model 1 is increasing the number of unstable
cases for most generators. However, there are some cases when the number of unstable cases is either
not affected or the number of instabilities decreases. G11 is almost not affected and G10 exhibits a
larger number of unstable cases with Generic model 2 even if the difference is small (8 cases). This
suggests that localized effects, which are especially important for transient stability, might be revealed
or missed according to the used model. However, it is important that when comparing to the initial base
case without RES, both models reveal the same tendency on a system level.

265
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 5-A REPRESENTATION OF UNBALANCED FAULT USING RMS MODEL

[44], [45]
Chapter 5-A.1 Introduction
The introduction of the Feed-In Tariff provoked in 2012 led to significant increase in Renewable Energy
Sources (RES) in Japan and about half of the PVs are the roof-top PVs in residential area which are
connected to Low Voltage (LV) network. Central Research Institute of Electric Power Industry (CRIEPI)
has developed the root-mean square based time-domain power system analysis tools which are called
Criepi’s Power Analysis Tools (CPAT) and has been used by all Japanese utilities for dynamic studies
for over 30 years. The single-phase residential PV inverter was installed into the laboratory, CRIEPI’s
power system simulator and the single- phase residential PV model was developed, validated and
implemented to CPAT. However, the numerical PV model is unlikely to be used for the dynamic studies
in LV network and is more likely to be needed for the grid interconnection study in small-scale network
with three-phase PVs connected to MV network. Therefore, the 10 kW three-phase PV inverter was
introduced into CRIEPI’s simulator and the three-phase PV model has been developed and validated.
Various levels of the voltage dips which come from three-phase fault with various fault durations were
applied to the three-phase PV in CRIEPI’s simulator in order for the model development. Then, the
developed model was validated through comparison between the measured response obtained by
CRIEPI’s simulator and the simulated response obtained by CPAT. This paper provides the new findings
for the dynamic behaviour following faults through the laboratory test and overviews the overall structure
of the developed model with identified parameters. The technical challenge for the further modelling
work is also discussed here.
Chapter 5-A.2 Laboratory Test with Artificial Fault and New Findings
The low voltage FRT test was performed for two different three-phase PV inverters (Type A and Type
B) made by different manufacturers. Type A inverter employs P-priority for the current limiter, while Type
B inverter employs Q-priority for the current limiter. Type A inverter can often stay connected even if the
voltage of the PV terminal is below 0.2 p.u. without the valve device blocking, while Type B inverter is
disconnected with the valve device blocking during the fault when the voltage of the PV terminal is below
0.2 p.u. Other specifications of Type A inverter and Type B inverter are in Table 5-A-1.
Table 5-A-1: Specification of Equipment Under Test

Specification Type A inverter Type B inverter

Rated active power output 10 kW 10 kW
Nominal voltage 202 V 202 V
Rated current 28.6 A 28. 6 A
Maximum current 30.0 A 30. 6 A

The test system is illustrated in Figure 5-A-1. The test system is either connected to the commercial
system or isolated from the commercial system using the 200 kW variable source which can behaves
variable frequency and voltage. The commercial system is mainly used and the 200 kW variable source
is partially used especially when the disconnection of the PV inverter needs to be avoided for the test
purpose. The fault point is placed in one of the double circuit of the MV emulated transmission line. In
order to ensure that the commercial system is not affected by the artificial fault, the HV emulated
transmission line of 300 km is additionally inserted between the commercial system and the MV
emulated transmission line. The three-phase inverter is grounded at phase B and the line-to-line voltage
is used. All laboratory tests were performed in 2016.
MV Emulated HV Emulated
Step-down Transmission Line Step-down Step-up
Transmission Line
3-ph Inv. Transformer 80km Transformer Transformer
300km
(Type A) 220/1650V 1650/3300V 3300/220V
Infinite Bus
or
200 kW
Variable
Source
3-ph Inv. 60km 20km 300km
(Type B) Fault Point
(3-ph-to-ground)

Figure 5-A-1 Test system in CRIEPI’s power system simulator .

266
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Chapter 5-A.3 Low-Voltage FRT Test

Both three-phase PV inverters employ undervoltage protection as an interface protection, i.e. an external
protection. The undervoltage protection settings consist of the voltage level of 0.8 p.u. and the time-
delay, i.e. timer of 2.0 seconds. The low voltage FRT tests were performed for both three-phase PV
inverters. Because Type A inverter is tested connecting the commercial network, the voltage sag level
is hard to be adjusted and was eventually set as around 0.1 p.u. It is worth noting that the Type A inverter
continues to stay connected up to 2.0 seconds (which is the operation timing of the undervoltage
protection), once it rides through the fault (See Figure 5-A-2). On the other hand, the voltage sag level
for Type B inverter was adjusted as 0.65 p.u. due to the constraints of the 200 kW variable source (See
Figure 5-A-3). Note that the 200 kW variable source was used only for Type B inverter in order to set
the voltage sag level which is no less than 0.2 p.u. It is also worth noting that the Type B inverter
continues to stay connected up to 2.0 seconds, once it rides through the fault. It can be observed that
the current can be promptly controlled equal to or below the maximum value following faults. That means,
once the current is successfully supressed equal to or below the upper limit, the three-phase PV inverter
is unlikely to be disconnected. This result reveals that the complicated low voltage FRT characteristics
is not necessary to be represented as the low voltage FRT protection and the simple undervoltage
protection without time-delay may be applied to represent the low voltage FRT capability. Although the
low voltage FRT characteristics varies depending on TSOs and utilities, this model component generally
requires huge program due to the variable time delay in response to the voltage of the PV terminal.
Therefore, the use of the aforementioned simple undervoltage protection model enables to reduce the
program size of the three-phase PV inverter model.

Fault occurrence: 0 [s], Fault duration: 1.5 [s] Fault occurrence: 0 [s], Fault duration: 1.9 [s]
Figure 5-A-2 Example continuous operation Figure 5-A-3: Example continuous operation
(Type A) [44] (Type B) [44]

Chapter 5-A.4 Active Power and Reactive Power Derivation

According to the questionnaire survey to the representative manufactures, the active power is calculated
based on the positive-sequence or the dq0 coordinate. Type A inverter and Type B inverter employ the
dq0 coordinate for deriving the active power output. The developed model involves the two different
active power derivations. Because CPAT is a Root-Mean-Square (RMS) model, the positive-, negative-,
and zero- sequence quantity can be obtained and the positive- and negative- sequence current can be
independently injected. In order to derive the active power output based on the dq0 coordinate, the
symmetrical coordinate is transformed into the alfa-beta coordinate first, then the alfa-beta coordinate
is further transformed into the dq0 coordinate as shown in Equations (5-A-1) and (5-A-2). It is noted that
the developed model represents the voltage source converter as a current source. Because CPAT
provides the model component which injects the positive-sequence current and the negative-sequence
current, individually, the current is injected as positive- and negative-sequence current transforming from
dq0 coordinate to symmetrical coordinate using Equations (5-A-3) and (5-A-4) after the control signal of

267
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

active power and reactive power are determined/decided. It should be emphasized that the positive and
negative voltages have to be obtained from the line-to-line voltage the phase voltage must not be used
because this results in the large discrepancy between the simulated response and measured response
in case of unbalanced fault.

V̇ V̇α 1 1 0 V̇1 İα 1 1 0 İ1

V̇ 1 1 a (a2 ) ab
[ 1] = [ ] [V̇bc ] [V̇β ] = [−j j 0] [V̇2 ] [İβ ] = [−j j 0] [İ2 ]
V̇2 3 1 (a2 ) a
V̇ca V̇0 0 0 1 V̇0 İ0 0 0 1 İ0
Equation 5-A-1

V̇d cosθ sinθ 0 V̇α İd cosθ sinθ 0 İα

[V̇q ] = [−sinθ cosθ 0] [V̇β ] [İq ] = [−sinθ cosθ 0] [İβ ]
V̇0 0 0 1 V̇0 İ0 0 0 1 İ0
Equation 5-A-2
İα cosθ −sinθ 0 İd
[İβ ] = [ sinθ cosθ 0] [İq ]
İ0 0 0 1 İ0

Equation 5-A-3

İ1
1 1 j 0 İα
[İ2 ] = [1 −j 0] [İβ ]
2
İ0 0 0 2 İ0
Equation 5-A-4

Chapter 5-A.5 Outline of Developed Model

The developed model consists of the active and reactive power control model component with the
current limiter, AC protection model components including two different types of the anti-islanding
protections, the simplified DC source control model component. The electric control block diagrams with
the simplified DC source model are shown in Figures 5-A-4.
P-Priority: Idmax = Imax
Q-Priority: Idmax = （Imax2-Iqref2）
I
1) dq0 coordinate based calculation
2) positive-sequence based calculation 0: Disconnection
Simplified DC Source Controller
Idmax 1: Stay connected
Capacitor
Active Power Idref
I P Limiter (3-ph mean)
(3-ph mean)
Ini=VDC0 1) convert to positive-sequence current
I-V Panel DCVref Idini 0 2) use positive-sequence current as it is
1 Characteristics (1.0p.u.) (Initial
Value)
P [p.u.]

Pin_chopper
(MPPT) ×

DCV [p.u.] 1.2

P=a×DCV2 + b×DCV + c

P-Priority: Iqmax = （Imax2-Idref2）

Q-Priority: Iqmax = Imax
I

1) dq0 coordinate based calculation 0: Disconnection

2) positive-sequence based calculation Iqmax 1: Stay connected

Reactive Power Iqref

P limiter
(3-ph mean) (3-ph mean)

Qref 1) convert to positive-sequence current

Iqini
(Reference) -Iqmax 2) use positive-sequence current as it is
(Initial Value)

Figure 5-A-4 Electric control model with DC source [44].

Chapter 5-A.6 Model Validation

The developed model was validated comparing the measured response obtained in CRIEPI’s simulator
and the model response calculated using CPAT. It is noted that only commercial system is used for the
model validation test. In addition, all anti-islanding protections are deactivated in order to differentiate of
the dynamic behaviour between the inner inverter control and the reactive power control coming from

268
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

the anti-islanding protection. Figure 5-A-5 shows an example result of Type A inverter when the fault
duration is 100 ms and the power factor of the PV output is unity. As shown in Figure 5-A-5, there is a
good match for active power, reactive power especially between the occurrence of the fault and the
clearance of the fault.
The derived electric control model parameters for unbalanced faults are shown in Table 5-A-2. Because
5 ms is used for I element of PI controller of the reactive current control, the time step of 1 ms is usually
required for the time-domain simulation. It is noted that the same parameters derived for balanced faults
are not the same as those that are shown in Table 5-A-2.

(a) AB phase line-to-line fault (b) BC phase line-to-line fault

(c) CA phase line-to-line fault

Figure 5-A-5 Dynamic behaviour following faults (Type A) [45].
Table 5-A-2: Derived electric control parameters

P I
Active current control 5.0 0.03s
Reactive current control 5.0 0.005s

269
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 5-B PV ARRAY MODEL

Based on the characteristics of PV array, a PV array can be represented by an equivalent circuit, as
shown in Figure 5-B-1.

Rs Iarray
Iph I0 Ir

Rsh

Figure 5-B-1: Equivalent circuit of PV cell.

where, the current source Iph is obtained using a standard equation that is related with solar irradiation,
and the relation between the array terminal current and voltage is the following equation:

  q(U dc +IRs )   (U dc +IRs )

I array =I ph -I 0 exp  -1 -
  AkT   Rsh
Equation 5-B-1
where,
Iarray is the output current of PV array;
Udc is the DC voltage of PV array;
Iph is the light induced current;
I0 is the diode saturation current;
q is electronic charge, 1.602 e-19 C;
k is Boltzmann constant，1.381 e-23 J/K;
A is the diode ideality factor, and the value of A is in the range of 1 and 2;
Rs and Rsh are series and shunt resistances, respectively.

The P-V curves of the PV array are shown in Figure 5-B-2, and the output power is related to the DC
voltage that is controlled by PV inverter.
5
x 10
6
1000W m2
5
800W m2
600W m2
4
400W m2
Power (W)

0
0 100 200 300 400 500 600 700 800
Voltage (V)

Figure 5-B-2: P-V characteristics of PV array.

Furthermore, a four-parameter model of the PV array is proposed because the parameters of equivalent
circuit are not easily to be obtained and because the diode saturation current is much less than the
induced current from the sun light. The four-parameter model is expressed as Equation 5-B-2.

270
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

I array =I SC  1-  e Udc -1

U OC
 I -I UOC -U m
 =  SC m 
 I SC 
1  1+ 
=  ln  
U OC   
Equation 5-B-2
where,
Iarray is the array current output;
Udc is the DC voltage of PV array;
ISC is the short circuit current;
Im is the current of maximum power point;
UOC is the open circuit voltage;
Um is the voltage of maximum power point.

When the environment condition varies, the key parameters ISC/Im/UOC/Um could be obtained by
standard parameter at standard test condition (STC), and the equations are:

T  Tair + K  S
DT =T -Tref
DS =S -Sref
S
I SC( m ) =I SC( m )_STC   1+aDT 
Sref
U OC( m ) =U OC _STC  1-cDT   ln  e+bDS 
Equation 5-B-3
where,
T is PV array temperature;
Tair is air temperature;
Tref is PV array temperature at standard test condition (STC);
K is constant (typically value 0.03oC·m 2/W);
S is solar irradiation;
Sref (1000W/m2) is the solar irradiation at STC;
Im_STC is the current of maximum power point at (STC);
ISC_ STC is the short circuit current at STC;
Um_ STC is the voltage of maximum power point at STC;
UOC_ STC is the open circuit voltage at STC.
a is constant (typically value 0.0025/oC);
b is constant (typically value 0.0005);
c is constant (typically value 0.00288/oC);
e is the base of the natural logarithm (approximate 2.71828).

271
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 5-C EXAMPLES OF RMS-TYPE INTEGRATED PV MODELS

Two generic PV models, namely the WECC PV power plant model and the CEPRI PV power plant
model are introduced in this Sub-Chapter together with example parameters. It should be noted that
these parameter values are given for illustration purposes. They are not generic and, hence, they should
be re-used with caution.
Chapter 5-C.1 WECC PV Power Plant Model [45]
a) WECC Renewable Energy Generator/Converter Model (REGC_A – see Figure 5-C-1)
This clause describes the model used in WECC (Western Electricity Coordinating Council) for studies
of the Western Interconnection of North America. The system model used in the studies is a positive-
sequence three-phase symmetrical model of the high-voltage transmission system, therefore models of
IBG are RMS models aggregated and adjusted to a three-phase balanced system.
The model incorporates a high bandwidth current regulator that injects real and reactive components of
inverter current into the external network in response to real and reactive current commands. Current
injection includes the following capabilities:
 Reactive current management during high voltage events at the generator terminal bus
 Active current management during low voltage events to approximate the response of the inverter
controls during voltage dips. This is intended to approximate the behaviour of PLL devices at very
low voltages when the PLL cannot operate correctly.

Power logic during low voltage events to allow for a controlled response of active current during and
immediately following voltage dips. This functionality is particular to some vendors and is included for
generality.
In this model, the voltage and frequency protections are handled separately by external modules. This
model is similar to the Type-3 and Type-4 generic wind generator models documented in [47].

272
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

REGC_A Iqrmax
Iqcmd -1 Iq +
× Iolim
Upward rat e limit on Iq act ive when Qgen0 > 0 1 + sTg
-
Downward rat e limit on Iq act ive when Qgen0 < 0
Iqrmin
Vt
Vt ≤ Volim Vt > Volim
+ 0
Khv
-
0
INTERFACE
LVPL & rrpwr Volim
TO
HIGH VOLTAGE REACTIVE CURRENT MANAGEMENT
NETWORK
Ipcmd 1 Ip MODEL
1 + sTg

LOW LOW VOLTAGE

LVPL VOLTAGE ACTIVE
POWER LOGIC
gain CURRENT
MANAGEMENT
Lvplsw Lvpl1 1
0 V 1
×
1 1 + sTf lt r 0

V V
Zerox Brkpt lvpnt 0 lvpnt 1

Figure 5-C-1: Block diagram of Generator/Converter model

REGC_A Input Parameters and Sample Settings
Name Description Typical Values
Tfltr Terminal voltage filter (for LVPL) time constant (s) 0.01 to 0.02
Lvpl1 LVPL gain break-point (pu current on mbase / pu voltage) 1.1 to 1.3
Zerox LVPL zero crossing (pu voltage) 0.4
Brkpt LVPL break-point (pu voltage) 0.9
Lvplsw Enable (1) or disable (0) low voltage power logic -
rrpwr Active current up-ramp rate limit on voltage recovery (pu/s) 10.0
Tg Inverter current regulator lag time constant (s) 0.02
Volim Voltage limit for high voltage clamp logic (pu) 1.2
Iolim Current limit for high voltage clamp logic (pu on mva base) -1.0 to -1.5
Khv High voltage clamp logic acceleration factor 0.7
lvpnt0 Low voltage active current management break-point (pu) 0.4
lvpnt1 Low voltage active current management break-point (pu) 0.8
Iqrmax Maximum rate-of-change of reactive current (pu/s) 999.9
Iqrmin Minimum rate-of-change of reactive current (pu/s) -999.9
REGC_A Internal Variables
Name Description
Vt Non-filtered terminal voltage (pu, from network solution)
V Filtered terminal voltage (pu)
LVPL Active current limit from LVPL logic (pu on mva base)
Iqcmd Desired reactive current (pu on mva base)
Ipcmd Desired active current (pu on mva base)
Iq Actual reactive current (pu on mva base)

b) WECC Renewable Energy Electrical Control Model for solar PV (REEC_B - see Figure 5-C-2)
The active power control subsystem provides the active current command to the current injection model.
The active current command is subject to current limiting, with user-selectable priority between active
and reactive current. The active current command is derived from a reference active power and the
inverter terminal voltage. The reference active power is the initial active power from the solved power
flow case; or, in the case where a plant controller model (REPC_A) is included, from the plant controller.
The reactive power control subsystem provides the reactive current command to the current injection
model. The reactive current command is subject to current limiting, with user-selectable priority between
active and reactive current.

273
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

The model includes the following reactive power control modes:

 Constant power factor, based on the inverter power factor in the solved power flow case
 Constant reactive power, based either on the inverter reactive power in the solved power flow
case or, in the case where a plant controller model (REPC_A) is included, from the plant
controller.

The model has an option to process the reactive power command via a cascaded set of PI regulators
for local reactive power and terminal voltage control, or to bypass these regulators and directly derive a
reactive current command from the inverter terminal voltage. In addition, it includes a supplementary,
fast-acting reactive current response to abnormally high or low terminal voltages.

REEC_B Vt _f ilt
dbd1,dbd2 Iqh1
1 -
Vt Kqv
1 + sTrv
if (Vt < Vdip) or (Vt > + Iql1
Vup) Vref 0
Volt age_dip = 1
else iqinj
Volt age_dip = 0
Vmax
Vma Iqmax
1 Pf Flag Qmax
Pe 1 + Vf lag x+ + Iqmax
1 + sTp
× Kqp + Kqi 1 QFlag
0 s Kvp + Kvi +
- 0 s
1
Iqcmd
pf aref t an Qmin Freeze st at e if Vmi - 0
Freeze st at e if
Qgen Vmin
Volt age_dip = 1 n Volt age_dip = 1 Iqmin
Qext Vt _f ilt Iqmin
0.01
1
÷ Imax Current
1 + sTiq Freeze st at e if
Limit
Volt age_dip = 1
Logic
Pqf lag
Current Limit Logic
Q Priorit y (Pqf lag =0):
Ipmax = (Imax2- Iqcmd2)1/ 2, Ipmin = 0 Pmax & dPmax Ipmax
Iqmax = Imax, Iqmin = - Iqmax 1
P Priorit y (Pqf lag =1): Pref ÷ Ipcmd
1 + sTpord
Ipmax = Imax, Ipmin = 0
Iqmax = (Imax2- Ipcmd2)1/ 2, Iqmin = - Iqmax
Freeze st at e if Ipmin =0
Pmin & dPmin Volt age_dip = 1
Vt _f ilt
0.01

Figure 5-C-2: Block diagram of Electrical Control Model

274
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

REEC_B Input Parameters and Sample Settings

Name Description Typical Values
PFflag Constant Q (0) or PF (1) local control -
Vflag Local Q (0) or voltage control (1) -
Qflag Bypass (0) or engage (1) inner voltage regulator loop -
Pqflag Priority to reactive current (0) or active current (1) -
Trv Terminal bus voltage filter time constant (s) 0.01 to 0.02
Vdip Low voltage condition trigger voltage (pu) 0.0 to 0.9
Vup High voltage condition trigger voltage (pu) 1.1 to 1.3
Vref0 Reference voltage for reactive current injection (pu) 0.95 to 1.05
dbd1 Overvoltage dead-band for reactive current injection (pu) -0.1 to 0.0
dbd2 Undervoltage dead-band for reactive current injection (pu) 0.0 to 0.1
Kqv Reactive current injection gain (pu/pu) 0.0 to 10.0
Iqhl Maximum reactive current injection (pu on mva base) 1.0 to 1.1
Iqll Minimum reactive current injection (pu on mva base) -1.1 to -1.0
Tp Active power filter time constant (s) 0.01 to 0.02
Qmax Maximum reactive power when Vflag = 1 (pu on mva base) -
Qmin Minimum reactive power when Vflag = 1 (pu on mva base) -
Kqp Local Q regulator proportional gain (pu/pu) -
Kqi Local Q regulator integral gain (pu/pu-s) -
Vmax Maximum voltage at inverter terminal bus (pu) 1.05 to 1.15
Vmin Minimum voltage at inverter terminal bus (pu) 0.85 to 0.95
Kvp Local voltage regulator proportional gain (pu/pu) 0.1
Kvi Local voltage regulator integral gain (pu/pu-s) 40
Tiq Reactive current regulator lag time constant (s) 0.01 to 0.02
Tpord Inverter power order lag time constant (s) 0.4
Pmax Maximum active power (pu on mva base) 1.0
Pmin Minimum active power (pu on mva base) 0.0
dPmax Active power up-ramp limit (pu/s on mva base) -
dPmin Active power down-ramp limit (pu/s on mva base) -
Imax Maximum apparent current (pu on mva base) 1.0 to 1.3

REEC_B Internal Variables

Name Description
Vt Raw terminal voltage (pu, from network solution)
Vt_filt Filtered terminal voltage (pu)
Voltage_dip Low/high voltage ride-though condition (0 = normal, VRT = 1)
Pe Inverter active power (pu on mva base)
Pref Inverter active power reference (pu on mva base, from power flow solution or from plant
controller model)
Pfaref Inverter initial power factor angle (from power flow solution)
Qgen Inverter reactive power (pu on mva base)
Qext Inverter reactive power reference (pu on mva base, from power flow solution or from
plant controller model)
Iqinj Supplementary reactive current injection during VRT event (pu on mva base)
Ipmax Maximum dynamic active current (pu on mva base)
Ipmin Minimum active current (0)
Iqmax Maximum dynamic reactive current (pu on mva base)
Iqmin Minimum dynamic reactive current (pu on mva base, = -iqmax)
Ipcmd Desired active current (pu on mva base)
Iqcmd Desired reactive current (pu on mva base)

275
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

c) Renewable Energy Plant Controller Model (REPC_A – see Figure 5-C-3)

This model is used when plant-level control of active and/or reactive power is implemented. The model
incorporates the following:

REPC_A

VcompFlag Vref
Ibranch |Vreg – (Rc+jXc)· 1 1 - +
Vreg Ibranch| 0 1 + sTfltr
Qmax
+ dbd emax
Qbran + 1
Kp + Ki 1 + s Tft
ch Kc RefFlag Qext
0 s 1 + s Tfv
emin Freeze state if
- Qmin Vreg < Vfrz
1
1 + sTfltr
+
Qref

Plant_pref
Pmax Freq_flag
fema
0
+ x
Pbranch 1 - Kpg + Kig 1 Pref
1
1 + sTp 0 + s 1 + sTlag
Ddn femin
fdbd1,fdbd2 + Pmin
-
Freg
+ +
Freq_ref Dup
0

Figure 5-C-3: Block diagram of Plant Controller Model

276
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

REPC_A Input Parameters and Sample Settings

Name Description Typical Values
RefFlag Plant level reactive power (0) or voltage control (1) -
VcompFlag Reactive droop (0) or line drop compensation (1) -
Freq_flag Governor response disable (0) or enable (1) 0
Tfltr Voltage and reactive power filter time constant (s) 0.01 to 0.02
Vbus Monitored bus number -
FromBus Monitored branch “from” bus number -
ToBus Monitored branch “to” bus number -
Ckt Monitored branch circuit designation -
Rc Line drop compensation resistance (pu on mva base) -
Xc Line drop compensation reactance (pu on mva base) when VcompFlag = -
1
Kc Reactive droop (pu on mva base) when VcompFlag = 0 -
dbd Reactive power dead band (pu on mva base) when -
RefFlag = 0; Voltage dead band (pu) when RefFlag = 1
emax Maximum Volt/VAR error (pu) -
emin Minimum Volt/VAR error (pu) -
Kp Volt/VAR regulator proportional gain (pu/pu)m 18
Ki Volt/VAR regulator integral gain (pu/pu-s) 5
Qmax Maximum plant reactive power command (pu on mva base) -
Qmin Minimum plant reactive power command (pu on mva base) -
Vfrz Voltage for freezing Volt/VAR regulator integrator (pu) 0.0 to 0.9
Tft Plant controller Q output lead time constant (s) 0.2
Tfv Plant controller Q output lag time constant (s) 0.15 to 5.0
fdbd1 Over-frequency dead band for governor response (pu) 0.01
fdbd2 Under-frequency dead band for governor response (pu) -0.01
Ddn Down regulation droop (pu power/pu freq on mva base) 20.0 to 33.3
Dup Up regulation droop (pu power/pu freq on mva base) 0.0
Tp Active power filter time constant (s) 0.01 to 0.02
femax Maximum power error in droop regulator (pu on mva base) -
femin Minimum power error in droop regulator (pu on mva base) -
Kpg Droop regulator proportional gain (pu/pu) -
Kig Droop regulator integral gain (pu/pu-s) -
Pmax Maximum plant active power command (pu on mva base) 1.0
Pmin Minimum plant active power command (pu on mva base) 0.0
Tlag Plant controller P output lag time constant (s) 0.15 to 5.0

REPC_A Internal Variables

Name Description
Vreg Regulated bus voltage (pu, from network solution)
Vref Regulated bus initial voltage (pu, from power flow solution)
Ibranch Branch current for line drop compensation (pu on mva base)
Qbranch Branch reactive power flow for plant Q regulation (pu on mva base)
Qref Regulated branch initial reactive power flow (pu, from power flow solution)
Qext Reactive power command from plant controller (pu on mva base)
Pbranch Branch active power flow for plant P regulation (pu on mva base)
Plant_pref Initial branch active power flow (pu on mva base, from power flow solution)
Freq Frequency deviation (pu, from network solution)
Freq_ref Initial frequency deviation (0)
Pref Active power command from plant controller (pu on mva base)

 Closed loop voltage regulation at a user-designated bus. The voltage feedback signal has
provisions for line drop compensation, voltage droop response and a user-settable dead band on
the voltage error signal.

277
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 Closed loop reactive power flow regulation on a user-designated branch with a user-settable dead
band on the reactive power error signal.
 A plant-level primary frequency control signal derived from frequency deviation at a user-
designated bus. The frequency droop response is applied to the active power flow on a user-
designated branch. Frequency droop control is capable of being activated in both over- and under-
frequency conditions. The frequency deviation applied to the droop gain is subject to a user-settable
dead band.

This model is identical to the plant-controller for the Type-3 and Type-4 generic wind turbine models
[19], although the model parameters are not the same.
Chapter 5-C.2 A Chinese Standard PV Power Plant Model [48]
a) Structure of Photovoltaic Power Plant Model
The whole model of the China Electric Power Research Institute (CEPRI) PV power plant includes: a
PV array model, a plant level control model, an inverter model, an equivalent model of collector and
transformer, as shown in Figure 5-C-4. The PV array model corresponds to the energy source and can
emulate the changing power output during solar irradiation fluctuations. The plant level controller model
is used to simulate the dynamic response corresponding to the dispatch command received from the
TSO. The response time of the plant level controller is normally between tens of seconds to several
minutes and the plant level controller model is useful in long-term simulations. For short-term simulations,
lasting e.g. less than one minute, such as in a transient stability study after a fault, the inverter model is
the most important part and it dominates the plant dynamic characteristics. The developed model
includes a PLL representation which consists of a first-order lag element, but does not include the inner
current loop and DC source control models other than the PV array model.

S PV array
model
PFref
Pm Inverter model Equivalent model
of collecting
PPOI _ ref system and U POI
f0 Pord I p _ cmd FRT and Ip Interface U term transformer
Plant level Local
QPOI_ ref protective module to
controller controller
controller power
U POI_ ref model
Qord
module
module system
PFPOI_ ref I q_ cmd Iq I ac
P + jQ PPOI + jQPOI

U term P Q U term f
f U POIPPOI QPOI

Figure 5-C-4: Structure of Chinese Standard PV Power Plant Model

b) PV Array Model
Due to the slow change rate of temperature for RMS simulation scenario, the effect of temperature can
be neglected, while only solar irradiation is considered, as shown in Figure 5-C-5. The PV array only
gives a maximum active power output signal Pm.

S Pm
S  b 
Pm =U m_sta  I m_sta  1 + e ( S - Sref ) 
Sref

Figure 5-C-5: Chinese Standard PV Array Model

PV array model: Input Parameters and Sample Settings
Name Description Typical Values
b constant 0.0005
e Base of the natural logarithm 2.71828
Im_sta Current of maximum power point at Standard Test Condition (STC) (pu) 1
Um_sta Voltage of maximum power point at STC (pu) 1
Pm Power of maximum power point (pu) 1
S Solar irradiation (W/m2) -
Sref solar irradiation in STC (W/m2) 1000

278
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

c) Local Controller Module

The purpose of the local controller module is to generate active/reactive current commands
corresponding to the plant level reference. As shown in Figure 5-C-6, the local controller module can be
separated into three parts:
 The active power control part can emulate local MPPT mode or plant order control mode. The
MPPT mode, when chosen, is simplified to a first-order lag block. The input signal is the maximum
power Pm from PV array model which is subject to the solar irradiation. In plant order control mode,
a delay block is used to reflect the communication delays. The dynamic response of active power
reference can be realized by a closed-loop proportional integral (PI) controller or a simple open-
loop first-order transfer functions block depending on the flag P_flag. The input signal Pord comes
from the plant level controller or is set equal to the power flow result if the plant level controller is
not included.
 The reactive power control block diagram is shown in Figure 5-C-6. This control block diagram can
switch the power factor mode to the reactive power control mode using the flag PF_flag. A closed-
loop method or an open-loop method can be selected using the flag Q_flag. In the latter case, the
inverter AC voltage, Uterm is used for deriving Iq. It is noted that Q_flag=0, i.e. the open-loop
method is mainly for the long-term system studies. Besides, an additional flag Iq_flag is used to
disable the reactive power controller and keep the reactive current constant. The input signals Qord
and PFref come from the plant level controller or are set equal to the power flow result if the plant
level controller is not included.
 The current limit logic part aims at limiting the magnitude of output current. The PQ_flag sets the
priority of active power and reactive power accordingly.

The active/reactive current commands coming from the current limit logic are then sent to the FRT and
protective controller module.

1 Pmea
P
1 + sTm
Pord Delay dPord_ max
I max PQ _ flag
Active power control

(Tp_ord) ramp limit

dPord_ min - Reset if active
power recovers
Pref 1
Kp + after fault clear Current limit logic
Pm 1 dPm _ max min sTp
1 + sTmpp
ramp limit
- I max  I qmax  I max

Q priority :  I qmin  - I max
0 I pin
P _ flag
1 1  I p _ cmd
U term ÷ 1 ( PQ _ flag  0)  I p max  I max 2
- I q_2 cmd
1 + sTm 1 + sTinverter
I
 p min  0
 I p max  I max
1 I 0
÷ 1 + sTinverter I max P priority :  p min
Reactive power control

Pmea I
( PQ _ flag  1)  qmax  I max - I p _ cmd
2 2
dQord_ max 0
PFref 1 - PFref2 Q _ flag
× 0 Qref
Kq +
1  I q_ cmd
PFref  I qmin  - I max - I p _ cmd
2 2
PF _ flag 1 1
ramp limit sTq
dQord_ min
- 1 I qin I q_ cmd  lim( I qin , I qmin , I qmax )
Qord Delay I q _ flag
- I max I p _ cmd  lim( I pin , I p min , I p max )
(Tq_ord) Iq 0 0
1 Qmea
Q
1 + sTm

Figure 5-C-6: Chinese Standard Local Controller Module Model

279
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Local controller module: Input Parameters and Sample Settings

Name Description Typical Values
dPord_max Active power reference up-ramp rate limit (pu/s) 0.1
dPord_min Active power reference down-ramp rate limit (pu/s) -0.1
dPm_max Active power up-ramp rate limit when irradiation increases (pu/s) 0.1
dQord_max Reactive power reference up-ramp rate limit (pu/s) 0.2
dQord_min Reactive power reference down-ramp rate limit (pu/s) -0.2
Tm Measurement lag time constant(s) 0.01 to 0.02
Tmpp Equivalent MPPT lag time constant (s) 0.1
Tp_ord Delay time of active power order (s) 1
Tq_ord Delay time of active power order (s) 1
Kp Active power regulator proportional gain -
Tp Active power regulator integral time constant (s) -
Kq Reactive power regulator proportional gain -
Tq Reactive power regulator integral time constant (s) -
Tinverter Inverter lag time constant (s) -
P_flag Active power control mode flag -
Q_flag Reactive power control mode flag -
PF_flag Power factor control mode flag -
PQ_flag Current limit logic flag -
Iq_flag Reactive current control mode flag -
Imax Maximum apparent current (pu) 1 to 1.2

Local controller module: Internal Variables

Name Description
Pord Active power order (pu)
P Active power (pu)
Pref Active power reference (pu)
Qord Reactive power order (pu)
Q Reactive power (pu)
Qref Reactive power reference (pu)
Uterm Terminal voltage (pu)
Ipmax Maximum active current (pu)
Ipmin Minimum active current (pu)
Iqmax Maximum reactive current (pu)
Iqmin Minimum reactive current (pu)
Iq0 Initial reactive current from power flow result(pu)
Pmea Filtered active power (pu)
Qmea Filtered reactive power (pu)
Ip_in Input active current command of current limit logic part (pu)
Iq_in Input reactive current command of current limit logic part (pu)
Ip_cmd Active current command (pu)
Iq_cmd Reactive current command (pu)

d) FRT and protective controller module

The FRT and protective controller module plays a key role during abnormal operation, typically fault
periods. It represents the transient characteristics of the inverter during the voltage drop/rise and the
recovery phases. According to Chinese national grid code GB/T 19964-2012, PV power plants are
required to have zero voltage ride through as low voltage ride through (LVRT) capability and generate
reactive current during grid voltage drop. Based on the terminal voltage, the operation scenarios can be
divided into three scenarios:
 Normal operation scenario: the voltage range is from 0.9pu to 1.1pu. The FRT controller has no
effect on the inverter, and only the local controller module is in operation;
 Low voltage operation scenario: the voltage range is from 0pu to 0.9pu. Reactive current is injected
into grid in proportion to the voltage drop magnitude;

280
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

 High voltage operation scenario: the voltage range is over 1.1pu. Reactive current is drawn from
grid to decrease voltage.

As shown in Figure 5-C-7, the current control signal is switched depending on the voltage level. The
reactive current control signal can be selected among three categories, while the active current control
signal can be selected from two categories, Ip_cmd and Ip_FRT. The voltage threshold values such as
ULV and UHV can be set manually. The main aim of the FRT module is to generate reactive/active
current command corresponding to the terminal voltage level. First, the reactive current is calculated
according to the formula shown in Figure 5-C-7; then, the active current can be derived in accordance
with the current limit flag Ip_flag. After fault clearing, the active current recovers with an up-ramp rate
limit dIp_LV.
The protective controller accounts for the protective characteristics of the inverter during over/under
voltage and over/under frequency. This module is used to generate a signal to disconnect the inverter
from grid if the severity of the fault is out of the range defined by grid requirements. The PV inverter can
then trip to protect itself from damage. In this case, the output current drops to zero. The voltage
protective logic and frequency protective logics are specific to grid requirements and vendors.

Active current with up-ramp

 max( K q_ HV (U HV - U term ) + K Iq 0 _flag I q 0 + I q 0 _ HV , I qmin_HV ) (U term  U HV ) rate limit on voltage
 recovery
I q   I q_cmd (U LV  U term  U HV ) dIp_LV
 (U term  U LV )
I p _ cmd  min( K q_LV (U LV - U term ) + K Iq 0 _flag I q 0 + I q 0 _LV , I qmax_LV )
Ip
 I p _cmd (U LV  U term  U HV )
Ip  
 p_FRT
I (U term  U LV U term  U HV )
0
 min( I p _cmd , I max _ FRT - I q )
 ( I p _flag =0) Iq
I q_ cmd 
I p_FRT   min( I p _cmd , I max
2
_ FRT - I 2
q ) ( I p _flag =1)

 K p1_ FRT I p0 +K p2 _ FRTU term + I p0 _ FRT
( I p _flag =2)

High/Low Voltage detection Delay time

UHVP2 < Uterm > tHVP2

U term UHVP1 ≤ Uterm ≤ UHVP2 > tHVP1

ULVP2 ≤ Uterm ≤ULVP1 > tLVP1

Uterm < ULVP2 > tLVP2

trip_flag
&
High/Low frequency detection Delay time

fHfP2 < f > tHfP2

fHfP1 ≤ f ≤ fHfP2 > tHfP1

f
fLfP2 ≤ f ≤ fLfP1 > tLfP1

f < fLfP2 > tLfP2

Figure 5-C-7: Chinese Standard FRT Model

281
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

FRT and protective controller module: Input Parameters and Sample Settings
Name Description Typical Values
Iqmax_LV Maximum reactive current during LVRT (pu) 1.1
Kq_LV Reactive current factor during LVRT 2
Iq0_LV Initial reactive current during LVRT (pu) 0
ULV Low voltage threshold value (pu) 0.9
KIq0_flag Initial reactive current add in flag 0
Iqmin_HV Minimum reactive current during HVRT (pu) -0.5
Kq_HV Reactive current factor during HVRT 2
Iq0_HV Initial reactive current during HVRT (pu) 0
UHV High voltage threshold value (pu) 1.1
Imax_FRT Maximum apparent current during FRT (pu) 1.1
Ip_flag Active current limit flag during FRT -
Kp1_FRT Active current factor during FRT 0
Kp2_FRT Active current factor during FRT 0
Ip0_FRT Initial active current during HVRT (pu) 0.1
dIp_LV Active current up-ramp rate limit after low voltage fault clearance (pu/s) 2
FRT and protective controller module: Internal Variables
Name Description
Ip_cmd Active current command from local controller module (pu)
Iq_cmd Reactive current command from local controller module (pu)
Uterm Terminal voltage (pu)
f Frequency (Hz)
Ip_FRT Active current during FRT (pu)
Ip0 Initial active current from power flow result (pu)
Iq0 Initial reactive current from power flow result (pu)
Ip Output active current (pu)
Iq Output reactive current (pu)

e) Grid interface module

The grid interface module is represented as a controlled current source. The output current is converted
into a phasor according to:
*
 U term  I p - j U term  I q 
I ac   
 U term 
 
Equation 5-C-1
where,
Ip is the active current;
Iq is the reactive current;
Uterm is the terminal voltage phasor.
f) Plant Level Controller Model
The plant level controller model reproduces the long-term dynamics of the PV power plant in response
to the dispatch commands received from the power system operator. The model consists of an active
and a reactive power control modules, as shown in Figure 5-C-8. Two different active power control
modes can be realized, namely constant active power and frequency droop control. The frequency droop
mode is activated to limit output power in over-frequency condition only. In addition, four reactive power
control modes can be realized, namely constant power factor, constant reactive power, constant voltage
control at POI bus, and voltage droop control.

282
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Pfdn _ max

f dn K pfdn
PPOI _ flag
- f up Pfup _ max
Pmax 0
0 Pord
1 - - 1
f K pfup K p_ POI (1 + ) 1
1 + sTm sTp_ POI
-
dPref_ max
0 Pmin
Ramp rate limit
PPOI_ ref
1 dPref_ min
PPOI
1 + sTm
U POI_ ref
U POI 1
- Qmax QPOI _ flag
1 + sTm
0
0 1 Qord
K qv QVPOI _ flag K q _ POI (1 + ) 1
sTq _ POI
1
1
QPOI - Qmin
1 + sTm

1 QPOI _ref
PPOI
1 + sTm
0
PFPOI _ flag
PFPOI_ ref 1 - PFPOI_
2
ref

PFPOI_ ref
× 1

Figure 5-C-8: CEPRI Power Plant Model

Plant level controller model: Input Parameters and Sample Settings
Name Description Typical Values
Tm Measurement lag time constant(s) 0.01 to 0.02
kpfup Over-frequency droop control factor (pu/Hz) 1
kpfdn Low-frequency droop control factor (pu/Hz) 1
fup Over frequency dead band (Hz) 50.1
fdn Low frequency dead band (Hz) 49.9
△Pfup_max Regulated power limitation for over-frequency control (pu) 0.2
△Pfdn_max Regulated power limitation for low-frequency control (pu) 0.1
dPref_max Up-ramp rate limit for active power reference (pu/s) 99
dPref_min Down-ramp rate limit for active power reference (pu/s) -99
Kp_POI Active power regulator proportional gain -
Tp_POI Active power regulator integral time constant (s) -
Kq_POI Rective power regulator proportional gain -
Tq_POI Rective power regulator integral time constant (s) -
Kqv Reactive power droop control factor 0

PPOI_flag Plant level active power control enable flag -

QVPOI_flag Reactive power / voltage control mode flag -
PFPOI_flag Power factor control mode flag -
QPOI_flag Plant level reactive power control enable flag -
Plant level controller model: Internal Variables
Name Description
f Frequency (Hz)
PPOI_ref Active power reference (pu)
PPOI Active power at POI (pu)
UPOI Voltage at POI (pu)
UPOI_ref Voltage reference at POI (pu)
PFPOI_ref Power factor reference
QPOI_ref Reactive power reference (pu)
QPOI Reactive power at POI (pu)

283
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 5-D BATTERY ENERGY STORAGE MODELS

There are various battery energy storage models for different purposes. For power system dynamic
stability studies, a simplified model is generally acceptable. Such a model assumes that the battery has
the same characteristics during charging and discharging, internal resistance is constant, temperature
does not affect the battery. The model also neglects self-discharge due to losses, and other phenomena
such as memory, and Peukert effects [49]. In general, more detailed equivalent circuit dynamic battery
models [49] are able to represent the battery charge/discharge behaviour and predict the batteries
operating temperature [49]. However, such details are not pertinent in large-scale power system
simulations and are thus neglected. On the other hand, functionalities such as active and reactive power
controls are extremely important and certainly modelled for power system studies. One such energy
storage model, useful for power system studies, is the WECC battery energy storage system model
[50],[51], which is described below.
Battery Energy Storage Systems (BESS) can perform active and reactive power control, voltage support
and frequency regulation, renewable energy smoothing, peak shaving, demand reduction, support for
islanding operations, and other functions for transmission systems [52].
As all other WECC renewable energy models, the BESS model is positive sequence model, presently
implemented in several commercially used power system simulation tools. This model does not consider
the battery chemical process because it is mostly intended for solid state batteries. As in the WECC
solar PV model, details of DC dynamics are also ignored. The WECC Renewable Energy Model for
Energy Storage consists of the same modules as other models: generator/converter, electrical controls
and possibly centralized plant controller. The generator/converter model and plant control, if present,
are the same as for solar PV plants (models recg_a and repc_a) described earlier. However, due to its
charging capabilities the model for electrical controls is different from the electrical control models for
the Solar PV plants. This model was developed and validated by EPRI [51].
For a BESS, the model structure is shown in Figure 5-D-1. Here, regc_a is the renewable energy
generator/converter model; its inputs are the real (Ipcmd) and reactive (Iqcmd) current commands and
its outputs the active (Ip) and reactive (Iq) currents injected into the grid. This represents the inverter
interface for the BESS unit. The reec_c model is the renewable energy electrical controls model c; its
inputs are the real power reference (Pref), and the reactive power reference (Qref). Both of them can
be externally controlled, and feedback of the reactive power generated (Qgen). The outputs of this model
are the active (Ipcmd) and reactive (Iqcmd) current command. This represents the BESS inverter
controls and includes a basic representation of the charging/discharging dynamics.
In addition to the above models, a repc_a (renewable energy plant controller model a) model may also
be used together with this configuration to allow for voltage and frequency control at a point of common
coupling. Furthermore, the more complex plant controller currently being developed may be used to
control a BESS unit together with other devices in a complex plant.
A detailed description of the reec_c model is given in [51], and is shown in Figure 5-D-2.

Figure 5-D-1: Structure of BESS Model [49].

284
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Figure 5-D-2: Block Diagram of reec_c Model [49].

The reec_c model is very similar to the electrical control model of a solar PV plant (reec_b) with the
following differences:
1) the minimum active current, Ipmin is equal to –Ipmax to allow for charging as well as discharging.
2) It contains an additional part with a simple mechanism for charging and discharging. The latter
is shown in Figure 5-D-3.

Figure 5-D-3: Additional charging-discharging Block Diagram of the BESS Model [49].
This part of the model has the following key features.
1) A user-defined parameter SOC specifies the initial state of charge of the battery. This tells the
model how much charge the battery has prior to starting the simulation.
2) A representation of the maximum and minimum allowable state of charge (shown as SOCmax
and SOCmin). For most batteries, it is recommended that the battery not be left in a state of full-charge
or full-discharge in order to preserve the battery’s longevity and performance. The model simulates this
through the user specified values for the maximum (SOCmax) and minimum (SOCmin) allowed SOC
during operation. Many vendors recommend operating the batteries within a range of 20% to 80% state
of charge.
3) The simple integrator block, with the time constant T, represents the process of charging and
discharging via calculating the actual SOC. The level of charge in the battery is proportional to

285
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

stored/released energy which is the time integral of power. When the active power is generated, the
SOC level will decrease, while the active power is consumed, the SOC level will increase.
4) The logic block at the end of the model represents the action of collapsing the output of the
converter (i.e. forcing its active current output) to zero once the maximum or minimum state of charge
has been reached. So, for example, if the SOC is greater than the allowable SOCmax, then Ipmin is
forced to zero, meaning that the battery cannot absorb/store any more electrical energy, while the
battery can still release any electrical energy (because Ipmax is not forced to zero).
It should be noted that the WECC BESS model was initially created for transient stability studies (which
means up to 10 - 30 seconds of simulated time). Therefore, the charging block of the BESS model may
be simplified. In cases where long-term dynamic simulations are to be performed, e.g. for the analysis
of frequency control and regulation (which means simulations as long as 10 min or more), more detailed
SOC management functions, if any, should be integrated into the model.
The detailed SOC management function is important when small capacity batteries (i.e., minutes rather
than hours of capacity) are installed into an isolated grid with high penetration of IBGs. Figure 5-D-4
shows the example BESS model with SOC management function in an isolated grid. The BESS
controller is in operation in an islanded grid with PVs. Because the battery is expensive, its capacity is
often quite limited. In such a case, the SOC could easily hit its upper or lower limits due to fluctuations
of the IBG output in the isolated grid. Therefore, the SOC needs to be constantly controlled or adjusted
to its reference value. Although the control strategy depends on the manufacturer, it is likely to consist
of a first-order lag element and a limiter.
It should be noted that the SOC management could potentially prevent the BESS control from mitigating
frequency deviations, for instance when the frequency deviation mitigation block is sending the
increasing control signal, the the SOC level decreases due to the discharging of the BESS. In order to
avoid increase in the SOC level, the SOC management sends decreasing control signal. Therefore, the
SOC management control signal should not be larger than the frequency deviation mitigating control
signal.
SOC Correction

State Of Charge %→kW

(SOC) [%] - 1
2
Reference of + 1+30s
SOC [%] Scale Lag
Conversion Element

DP Input
pu→kW 100[kW] 120[kW]
+ LiC
Active 1 30s + Control
Power of PV 1.0 1000
1+0.02s 1+30s Signal
[PU] +
Output
Signal Gain Washout Scale -100[kW] -120[kW]
Transducer Element Conversion

DF Input
puHz/Hz Hz→kW 100[kW]

Frequency 1 15s 1+0.5s 1+0.5s

60 175
Deviation[PU] 1+0.02s 1+15s 1+0.37s 1+0.37s
Signal Scale Washout Lead-lag Lead-lag Gain -100[kW]
Transducer Conversion Element Element Element

Figure 5-D-4: Example of BESS model with SOC correction function in an isolated grid [53]
(Battery capacity: ±100kW - 50s)
The table below gives examples of parameters for the battery storage electrical control of the WECC
BESS model. However, these parameters may be different depending on the device.

286
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

REEC_C Input Parameters

Name Description Example Values
PFflag Constant Q (0) or PF (1) local control 0 or 1
Vflag Local Q (0) or voltage control (1) 0 or 1
Qflag Bypass (0) or engage (1) inner voltage regulator loop 0 or 1
Pqflag Priority to reactive current (0) or active current (1) 0 or 1
Trv Transducer time constant (s) 0.0
Vdip Low voltage condition activating current injection logic (pu) -99
Vup High voltage condition activating current injection logic (pu) 99
Vref0 Reference voltage for reactive current injection (pu) 0.0 (activates to
terminal voltage
dbd1 Overvoltage dead band in voltage error (pu) -0.05
dbd2 Undervoltage dead band in voltage error (pu) 0.05
Kqv Reactive current injection gain during voltage dips (pu/pu) 0.0
Iqhl Maximum limit on reactive current injection (pu on mva base) 1.05
Iqll Minimum limit reactive current injection (pu on mva base) -1.05
SOC ini Initial state of charge 0 -1
SOC max Maximum allowable state of charge 0.8 – 1.0
SOC min Minimum allowable state of charge 0.0-0.2
T Discharge time, sec 14400 - 99999
Tp Active power filter time constant (s) 0.05
Qmax Maximum reactive power wheniVflag = 1 (pu on mva base) 0.44
Qmin Minimum reactive power when Vflag = 1 (pu on mva base) -0.44 - 0
Kqp Local Q regulator proportional gain (pu/pu) 0
Kqi Local Q regulator integral gain (pu/pu-s) 0.1
Vmax Voltage control maximum limit (pu) 1.1
Vmin Voltage control minimum limit (pu) 0.9
Kvp Local voltage regulator proportional gain (pu/pu) 0.0
Kvi Local voltage regulator integral gain (pu/pu-s) 120
Tiq Reactive current regulator lag time constant (s) 0.02
Tpord Inverter power order lag time constant (s) 0.04
Pmax Maximum active power reference (pu on mva base) 1.0
Pmin Minimum active power reference (pu on mva base) 0
dPmax Up-ramp rate on power reference (pu/s on mva base) 99
dPmin Down-ramp rate on power reference (pu/s on mva base) -99
Imax Maximum allowable total current limit (pu on mva base) 1.7
Vq1 User defined voltage for VDL1 function, p.u 0
Iq1 User defined current for VDL1 function, p.u 1.45
Vq2 User defined voltage for VDL1 function, p.u 2
Iq2 User defined current for VDL1 function, p.u 1.45
Vq3 User defined voltage for VDL1 function, p.u 0
Iq3 User defined current for VDL1 function, p.u 0
Vq4 User defined voltage for VDL1 function, p.u 0
Iq4 User defined current for VDL1 function, p.u 0
Vp1 User defined voltage for VDL2 function, p.u 0
Ip1 User defined current for VDL2 function, p.u 1.15
Vp2 User defined voltage for VDL2 function, p.u 2
Ip2 User defined current for VDL2 function, p.u 1.15
Vp3 User defined voltage for VDL2 function, p.u 0
Ip3 User defined current for VDL2 function, p.u 0
Vp4 User defined voltage for VDL2 function, p.u 0
Ip4 User defined current for VDL2 function, p.u 0

287
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 6-A EXAMPLE OF TYPICAL DATA FOR OVERHEAD AND

UNDERGROUND CABLES IN PORTUGAL
Example of representative data for overhead lines and underground cables for nominal voltage, positive-
sequence line parameters (impedance and charging) and line rating are shown in Table 6-A-1 [54].
Table 6-A-1: Generator protection [54]
IS I F/S IW Iccrms S F/S Xd / Rd
UN Line Structure and Physical Constitution Rd Xd Bd Cd -QC d (0.5 s)
Line Code Summer Fall / Spring Winter Fall / Spring
kV (Cross section per phase) W / Km W / Km W -1 / Km nF / Km kvar / km 
A A A MVA pu
kA 1 pu

400 kV ACSR 595 mm 2 | Overhead simple line with twin Bundle conductors |
400 400_ACSR_B595_O_H10DG150 Suspended chains of insulators | Horizontal configuration (d=10 m) | Double 0.0294 0.3239 3.528E-06 11.23 564.45 2 306 2 532 2 680 136.1 1 754 11.0
Ground wires ACSR 150 mm 2
400 kV ACSR 595 mm 2 | Overhead Double line with twin Bundle conductors |
400 400_ACSR_DB595_O_VDG150 Suspended chains of insulators | Vertical configuration | Double Ground wires 0.0290 0.3280 3.535E-06 11.25 565.66 9 224 10 127 10 720 272.3 7 016 11.3
ACSR 150 mm 2
2
220 kV ACSR 485 mm | Overhead simple line | Suspended chains of insulators |
220 220_ACSR_485_O_FDG150 0.0770 0.4230 2.702E-06 8.60 130.80 1 002 1 086 1 141 51.0 413.8 5.49
Flag configuration | Double Ground wires ACSR 150 mm 2
220 kV ACSR 485 mm 2 | Overhead Double line | Suspended chains of insulators |
220 220_ACSR_D485_O_VDG150 0.0765 0.3933 2.891E-06 9.20 139.93 2 004 2 173 2 284 102.0 828.0 5.14
Vertical configuration | Double Ground wires ACSR 150 mm 2
60 kV ACSR 160 mm 2 | Overhead simple line | Suspended chains of insulators |
60 60_ACSR_160_O_SFG130 0.2295 0.3781 2.934E-06 9.34 10.56 225 360 450 16.1 37.4 1.65
Flag configuration | Ground wire ACSR 130 mm 2
60 kV ACSR 325 mm 2 | Overhead simple line | Suspended chains of insulators |
60 60_ACSR_325_O_SFG130 0.1181 0.3716 3.082E-06 9.81 11.09 330 545 685 31.3 56.6 3.15
Flag configuration | Ground wire ACSR 130 mm 2
60 kV ACSR D160 mm 2 | Overhead Double line | Suspended chains of insulators |
60 60_ACSR_D160_O_SVG130 0.1151 0.1981 5.689E-06 18.11 20.48 450 720 900 32.1 74.8 1.72
Vertical configuration | Ground wire ACSR 130 mm 2
60 kV ACSR D325mm 2 | Overhead Double line | Suspended chains of insulators |
60 60_ACSR_D325_O_SVG130 0.0594 0.1920 6.050E-06 19.26 21.78 660 1 090 1 370 62.5 113.3 3.23
Vertical configuration | Ground wire ACSR 130 mm 2
60 kV AL-Core 185 mm2 | Underground simple line (3 single core cables in
60 60_LXHIOLE_185_U 0.2103 0.1382 43.982E-06 140.00 158.34 328 365 365 24.7 37.9 0.66 *
touching trefoil laid on the bottom of a trench)
60 kV AL-Core 400 mm2 | Underground simple line (3 single core cables in
60 60_LXHIOLE_400_U 0.0997 0.1194 59.690E-06 190.00 214.88 486 540 540 53.5 56.1 1.20
touching trefoil laid on the bottom of a trench)
60 kV AL-Core 630 mm2 | Underground simple line (3 single core cables in
60 60_LXHIOLE_630_U 0.0601 0.1100 78.540E-06 250.00 282.74 630 700 710 65.9 72.7 1.83
touching trefoil laid on the bottom of a trench)
60 kV AL-Core 1000 mm2 | Underground simple line (3 single core cables in
60 60_LXHIOLE_1000_U 0.0373 0.1005 106.814E-06 340.00 384.53 788 875 875 133.6 90.9 2.69
touching trefoil laid on the bottom of a trench)

30 kV ACSR 50 mm 2 | Overhead simple line | Suspended chains of insulators |

30 30_ACSR_050_O_SF 0.7309 0.3940 2.897E-06 9.22 2.61 115 180 220 5.0 9.4 0.54 *
Flag configuration
30 kV ACSR 70 mm 2 | Overhead simple line | Suspended chains of insulators |
30 30_ACSR_070_O_SF 0.5208 0.3834 2.981E-06 9.49 2.68 130 205 255 7.0 10.7 0.74 *
Flag configuration
30 kV ACSR 90 mm 2 | Overhead simple line | Suspended chains of insulators |
30 30_ACSR_090_O_SF 0.4112 0.3759 3.043E-06 9.69 2.74 160 255 315 8.9 13.3 0.91 *
Flag configuration
30 kV ACSR 160 mm 2 | Overhead simple line | Suspended chains of insulators |
30 30_ACSR_160_O_SF 0.2294 0.3566 3.215E-06 10.23 2.89 225 360 450 16.1 18.7 1.55
Flag configuration
30 kV ACSR 325 mm 2 | Overhead simple line | Suspended chains of insulators |
30 30_ACSR_325_O_SF 0.1089 0.3357 3.311E-06 10.54 2.98 325 540 685 31.3 28.1 3.08
Flag configuration
30 kV ACSR D160 mm 2 | Overhead Double line | Suspended chains of insulators |
30 30_ACSR_D160_O_SV 0.1147 0.1980 4.778E-06 15.21 4.30 450 720 900 40.0 37.4 1.73
Vertical configuration
30 kV ACSR D325 mm 2 | Overhead Double line | Suspended chains of insulators |
30 30_ACSR_D325_O_SV 0.0720 0.1746 4.778E-06 15.21 4.30 560 920 1 150 47.6 47.8 2.43
Vertical configuration
30 kV AAAC 55 mm 2 | Overhead simple line | Suspended chains of insulators |
30 30_AAAC_055_O_SF 0.6507 0.3909 2.921E-06 9.30 2.63 125 190 235 6.1 9.9 0.60 *
Flag configuration
30 kV AAAC 117 mm 2 | Overhead simple line | Suspended chains of insulators |
30 30_AAAC_117_O_SF 0.3038 0.3713 3.098E-06 9.86 2.79 190 300 380 16.0 15.6 1.22
Flag configuration
30 kV AAAC 148 mm 2 | Overhead simple line | Suspended chains of insulators |
30 30_AAAC_148_O_SF 0.2415 0.3639 3.167E-06 10.08 2.85 220 350 435 20.3 18.2 1.51
Flag configuration
30 kV AL-Core 120 mm2 | Underground simple line (3 single core cables in
30 30_LXHIOV_120_U 0.3244 0.1172 53.460E-06 170.17 48.11 240 285 325 15.9 14.8 0.36 *
touching trefoil laid on the bottom of a trench)
30 kV AL-Core 240 mm2 | Underground simple line (3 single core cables in
30 30_LXHIOV_240_U 0.1603 0.1049 69.069E-06 219.85 62.16 355 420 480 32.2 21.8 0.65 *
touching trefoil laid on the bottom of a trench)
30 kV AL-Core D120 mm2 | Underground Double line (two sets of 3 single core
30 30_LXHIOV_D120_U 0.1622 0.0586 106.920E-06 340.34 96.23 480 570 650 31.8 29.6 0.36 *
cables in touching trefoil laid on the bottom of a trench)
30 kV AL-Core D240 mm2 | Underground Double line (two sets of 3 single core
30 30_LXHIOV_D240_U 0.0801 0.0458 277.887E-06 884.54 250.10 714 840 960 64.5 43.6 0.57 *
cables in touching trefoil laid on the bottom of a trench)
30 kV CU-Core 95 mm2 | Underground simple line (a three core cable with one
30 30_PHCAV_095_U 0.2271 0.1130 78.540E-06 250.00 70.69 210 245 245 14.8 12.7 0.50 *
lead sheath laid on the bottom of a trench)
30 kV CU-Core 120 mm2 | Underground simple line (a three core cable with one
30 30_PHCAV_120_U 0.1801 0.1068 87.965E-06 280.00 79.17 240 280 280 18.8 14.5 0.59 *
lead sheath laid on the bottom of a trench)
30 kV CU-Core 185 mm2 | Underground simple line (a three core cable with one
30 30_PHCAV_185_U 0.1161 0.1107 94.248E-06 300.00 84.82 350 415 470 28.7 21.6 0.95 *
lead sheath laid on the bottom of a trench)
30 kV CU-Core 240 mm2 | Underground simple line (a three core cable with one
30 30_PHCAV_240_U 0.0902 0.0929 157.079E-06 500.00 141.37 415 485 485 39.8 25.2 1.03
lead sheath laid on the bottom of a trench)

0,4 kV AL-Core 16 mm 2 |Aerial bundled cables (4 single core cables in twisted

0.4 0.4_LXS_4x16_O 2.15 0.10 --- --- --- 68 75 83 1.9 0.052 0.05 *
assembly with ph1+ph2+ph3+neutral)
0,4 kV AL-Core 25+16 mm 2 | Aerial bundled cables [5 single core cables in twisted
0.4 0.4_LXS_4x25+16_O 1.34 0.10 --- --- --- 90 100 110 3.0 0.069 0.07 *
assembly with ph1+ph2+ph3+neutral (25 mm 2)+public street lighting (16 mm 2)]
0,4 kV AL-Core 50+16 mm 2 | Aerial bundled cables [5 single core cables in twisted
0.4 0.4_LXS_4x50+16_O 0.72 0.10 --- --- --- 135 150 165 6.0 0.104 0.14 *
assembly with ph1+ph2+ph3+neutral (50 mm 2)+public street lighting (16 mm 2)]
0,4 kV AL-Core 70+16 mm 2 | Aerial bundled cables [5 single core cables in twisted
0.4 0.4_LXS_4x70+16_O 0.50 0.10 --- --- --- 171 190 209 8.4 0.132 0.20 *
assembly with ph1+ph2+ph3+neutral (70 mm 2)+public street lighting (16 mm 2)]
0,4 kV AL-Core 95+16 mm 2 | Aerial bundled cables [5 single core cables in twisted
0.4 0.4_LXS_4x95+16_O 0.36 0.10 --- --- --- 207 230 253 11.4 0.159 0.28 *
assembly with ph1+ph2+ph3+neutral (95 mm 2)+public street lighting (16 mm 2)]
0,4 kV AL-Core 16 mm2 | Underground simple line (4 core cable with
0.4 0.4_LSVAV_4x16_U 2.29 0.10 --- --- --- 81 90 99 1.9 0.062 0.04 *
ph1+ph2+ph3+neutral laid on the bottom of a trench)
0,4 kV AL-Core 35 mm2 | Underground simple line (4core cable with
0.4 0.4_LSVAV_4x35_U 1.04 0.10 --- --- --- 117 130 143 4.2 0.090 0.10 *
ph1+ph2+ph3+neutral laid on the bottom of a trench)
0,4 kV AL-Core 50 mm2 | Underground simple line (4 core cable with
0.4 0.4_LSVAV_4x50_U 0.769 0.10 --- --- --- 135 150 165 6.0 0.104 0.13 *
ph1+ph2+ph3+neutral laid on the bottom of a trench)
0,4 kV AL-Core 95 mm2 | Underground simple line (4 core cable with
0.4 0.4_LSVAV_4x95_U 0.384 0.10 --- --- --- 212 235 259 11.4 0.163 0.26 *
ph1+ph2+ph3+neutral laid on the bottom of a trench)
0,4 kV AL-Core 185 mm 2 | Underground simple line [4 core cable with
0.4 0.4_LVAV_3x185+95_U ph1+ph2+ph3 (185 mm 2)+neutral (95 mm 2 ) laid on the bottom of a trench] 0.197 0.10 --- --- --- 320 355 391 22.3 0.246 0.51 *
Note: For phase-neutral fault (Rd neutral = 0.384 W / km ; Xd neutral = 0.10 W / km)

0.23 kV AL-Core 16 mm 2 | Aerial bundled cables (2 single core cables in twisted

0.23 0.23_LXS_2x16_O 2.15 0.10 --- --- --- 77 85 94 1.9 0.020 0.05 *
assembly with phase+neutral)
0.23 kV AL-Core 16 mm2 | Underground simple line (2 core cable with
0.23 0.23_LSVAV_2x16_U 2.29 0.10 --- --- --- 86 95 105 1.9 0.022 0.04 *
phase+neutral laid on the bottom of a trench)

Note - An asterisk * is shown in the table when the ratio Xd / Rd ≤ 1 p.u. is reached.

288
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 7-A AN OPEN-SOURCE DISTRIBUTED CONTROL PLATFORM FOR HIL-

BASED TESTING AND DEMONSTRATION OF ADVANCED POWER SYSTEMS
Chapter 7-A.1 Motivation
The management of emerging, high agile power and energy systems in the context of achieving the
overall mission operations must be facilitated by a distributed control environment [55]. Applied and
basic research in distributed control is a subject actively being addressed by the naval community. The
aim of such work is to provide autonomy, intelligence, and resilience under the extreme conditions that
exist in potential mission operations, enabling a high Quality of Service (QoS) to mission loads (weapons,
sensor arrays, etc.) while maintaining other critical infrastructure needs. The difficulty of the transition
from theoretical control research, especially in a distributed context, to a physical demonstration should
not be underestimated. Simply put, the theory must be validated in a large, multi-faceted test bed
platform. Our expertise in modelling and simulation at the Florida State University (FSU) Centre for
Advanced Power Systems (CAPS) has led us to the conclusion that the unconstrained or over-simplified
models that are typically used for the purpose of advanced control research and development will often
result in what might be considered as an over-optimistic assessment of the control reachability and
functionality. This may be especially problematic during physical controller instantiation. This text
describes the existing controller hardware in the loop (CHIL) and power hardware in the loop (PHIL)
capabilities at FSU-CAPS along with recently funded infrastructure upgrades necessary to meet the
United States Department of Defence (DoD) power and energy community needs in assessing the
validity and efficacy of any particular distributed control methodology in a highly flexible and realistic
environment that is appropriate for the specific applications.
Chapter 7-A.2 Existing Facilities
The CAPS PHIL laboratory has been designed to test electrical and mechanical hardware under test
(HUT) with power capability up to 5 MW, as depicted in Figure 7-A-1, while coupled with a rest-of-system,
real-time simulation. CAPS’ real-time modelling and simulation capabilities are currently based on the
two most prevalent commercially available real-time simulators (RTDS from RTDS Technologies, Inc.
and eMEGAsim/RT-LAB from Opal-RT Technologies, Inc.). CHIL experiments are achieved through
either dedicated wiring using analogue and digital signals or standard communication network protocols
such as Distributed Network Protocol (DNP3), and IEC 61850 (communication networks and systems
in substations). In both cases—PHIL and CHIL—the HUT is part of a closed-loop simulation and it allows
evaluation of dynamic responses, while the remaining part is modelled and simulated using the real-
time environments.
CAPS has utilized real-time simulation capabilities successfully in numerous projects, such as dynamic
testing of a prototype 5 MW high-temperature superconducting motor [56], a high-speed generator [57],
superconducting fault current limiters [58], and also successfully conducted co-simulation of electro-
mechanical and thermal systems [59], [60]. CAPS’ founding role in the Electric Ship Research and
Development Consortium (ESRDC), its research and development contributions in this community, and
its success with PHIL and CHIL testing, have established a credible foundation with the Navy and the
Navy ship systems supplier community [61]. Based on this established competency, developmental
testing and characterization of candidate future ship technologies is an on-going process at CAPS.
In addition to the aforementioned facilities, two key test beds have been developed that are focused on
distributed controls. The low-power Distributed Controls test bed is shown in Figure 7-A-2. The 400 V/10
kVA system in Figure 7-A-2a consists of 1) controllable power electronics-based source/sinks (Rk1 or
Rk2) that are fully integrated into the test bed for emulation of a variety of devices allowing for AC/DC
or hybrid systems to be developed, 2) AC inverter-fed (INV+AC) loads to emulate ship propulsion, 3)
controllable storage (Bat) to emulate batteries, ultra-capacitors, or flywheels, and 4) a pulse power load
(DC) structure to emulate EMRG or radar. The test bed has distributed control platform composed of 18
national instruments devices (15 myRIO and 3 sbRIO) operating with LabVIEW environment (Figure 7-
A-2c). The control units are connected via dedicated Ethernet and to test bed equipment via CAN,
RS232, Ethernet, and serial communication. This test bed can cooperate with RTDS (Figure 7-A-2d) via
analogue and digital I/O (Figure 7-A-2e) to do the CHIL or PHIL experiments.

289
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

MVDC Lab 5 MW High Bay

Gear Gear
CU3 CU1 Box Box DYNO
12000 1800 225
24000 3600 450 VSD
MMC 3 MMC 1
+ (PU3) - + (PU1) - Variable Voltage Source Converter
Controller Controller
0...4.16 kV 0...1.1 kV
0...0.4/0.8 kA AC DC 0...2.5 kA
Each Module: +/- 0...6 kV DC, VVS VVS
45...120 Hz
+/- 0...210 A DC

MVDC Parallel: 6 kV, 0.8 kA

MVDC 0...480 V
T2 Series: 24 kV, 0.21 kA T1 1.5 kA
CU4 CU2 T9.2 T9.1
T1
MMC 4 MMC 2 4.16 kV
+ (PU 4) - + (PU 2) - Rest of System Simulation

Feedback and
Control Signals

Utility 12.47 kV
RTDS Simulator Lab

Figure 7-A-1: 5 MW PHIL facility at FSU-CAPS.

Figure 7-A-2: Low-power Distributed Controls Test Bed: a) architectural layout, b) control room, c)
distributed control devices, d) simulator room, e) analogue and digital IO interface to RTDS.

Additionally, CAPS has developed a CHIL test bed composed of 6 Mamba computational platforms. The
platforms are networked together via Ethernet and connected via fibre to the RTDS system. This test
bed has been mainly used with simulated power sources and sinks, with up to 4 control nodes per board.
The maximum number of control nodes is based on the complexity of the devices and the required
control cycle-time. Most importantly, controller-controller communication is networked through an
OPNET real-time platform which simulates the salient features of the controls network infrastructure
between the Mamba nodes. This unique laboratory setup, referred to as the Cyber-Physical Hardware-
in-the-Loop Test Bed (CP-HIL-TB) allows the researcher to conduct a realistic CHIL experiment utilizing
distributed controls. Given high fidelity sources, loads and systems in the RTDS, a distributed control

290
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

environment of limited size can be created in order to conduct “high risk” experimentation without the
added complexity of the utilization of actual equipment [62]. This is very advantageous given that many
power electronic-based energy conversion devices that are currently under research may only exist in
the modelling and simulation environment and have yet to be physically built. Figure 7-A-3 illustrates
the CP-HIL-TB.

Figure 7-A-3: Cyber-physical Hardware-in-the-loop Test Bed.

Chapter 7-A.3 Towards a fully integrated CHIL/PHIL facility

In September 2016, CAPS received a substantial Defence University Research Instrumentation
Program (DURIP) grant from the US Office of Naval Research (ONR) to establish a fully integrated
CHIL/PHIL facility of unprecedented level in system size and flexibility. Most important will be the
complete integration of all the PHIL and CHIL resources to allow experimenting with up to approximately
100 distributed control nodes simultaneously. This will provide to researchers the ability to test and
evaluate control algorithms for emerging next generation terrestrial and shipboard power systems in a
realistic system environment to meet Technology Readiness Level 6 (TRL6) requirements at reasonable
system scale. Figure 8 is a systems level overview of the open-source distributed control concept that
envisioned after full implementation of this DURIP award. The core HIL engines of RTDS and OPAL-
RT, in the lower left, provide the ability to simulate large-scale systems and provide emulation
characteristics and commands to physical equipment. The physical systems near the top of the diagram
are connected via fibre optics to the real-time simulators. At each of the physical devices is an array of
distributed control nodes corresponding to physical controller devices. In addition to the physical
systems, virtual loads and sources are present in the simulation. In the same way as the physical devices,
the virtual devices are connected to physical controller devices. Each of the distributed control platforms
are connected via Ethernet to create a true networked distributed control layer. An emulated network
layer is also added, which via emulation software such as OPNET and OMNeT can more accurately
represent the complexity of a physical distributed network.

Low Voltage 10kVA

5 MW MMC-based Other
5 MW VVS Distributed Control
Dynamometer 5MW MVDC Hardware
Test Bed
Controller Controller Controller(s) Controller(s) Controller(s)

Network

Controller(s) Controller(s)
Simulated Simulated
Sources Loads

Real-Time Computational Network Emulation

Systems (RTDS, OPAL-RT) (Opnet, Omnet)

Figure 7-A-4: CAPS Open-Source Distributed Control System Layout.

291
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

APPENDIX 7-B TYPE OF STUDIES AND TESTS FOR MODEL VALIDATION

It is recommended that the following studies and tests in Table 7-B-1 should be considered for model
validation. The measurements required are P, Q, Vrms, and Irms at the output terminals of the inverter or
at the grid connection point for several inverters. The voltage and current at the DC side of the inverter
are also useful and should be measured in some tests.
 System dynamic response to control signal (small variation in control signals)
 System dynamic response to small disturbances in grid voltage
 Long-term voltage stability
 Short-term voltage stability
 FRT (Low/High/Zero voltage ride through)
 Short-circuit current contribution
 Rotor angle stability
 Solar radiation variation study
 Frequency instability study
 Unintentional islanding detection studies

It should be pointed out that not all the tests in Table 7-B-1 should be carried out in model validation. As
a minimum, the model validation should be exercised for the studies it is designed for. For example, if a
model of a PV converter is designed to represent the PV converter’s anti-islanding protection, the model
must be validated against the AI tests following the system requirements or according to the relevant
standard such as IEEE 1547 in the US.
The IEC 61400-27-2 which describes the validation of the RMS model for power system dynamic studies
can be referred to. It should be pointed out that this document is currently in voting stage.
Although Table 7-B-1 provides the detailed information for the required test, what the inverter can
perform during testing very much depends on the control design of the inverter. That means, different
inverters have different capabilities and performance. Such difference can be even significant. With the
PV as primary energy source, this issue become even more system dependent. Therefore, it is noted
that the figures in Table 7-B-1 come from limited experience and they could change.

292
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

Table 7-B-1: Type of studies and tests for model validation

Study Names Tests to be Performed

System dynamic response to Set the inverter operating in supporting grid mode, at an operation point,
control signal adjust the command of P within a small range, e.g. (-10%, +10%)
(small variation in control monitor the time-trend of the defined measurements for comparison
signals) This test can be repeated a number of times at different operation points
over the entire operation range
System dynamic response to Set the AC Grid at its nominal voltage, the inverter at a fixed operational
small disturbances in grid mode (P/Q or P/V) and operate at its rated power,
voltage emulate a step change in the grid voltage in the range of (-10, +10%) of the
nominal value
monitor the time-trend of the defined measurements for comparison
This test can be repeated a number of times with the inverter initially running
at 0.2, 0.5, 0.8 and 1.0 p.u.
Long-term Small voltage disturbance test:
voltage stability * Select the inverter at a fixed operational mode (P/Q or P/V etc.), Test the
system response to the grid voltage variation in the range of (-10, +10%) of
the nominal value.
Start the test with the inverter operating at its rated power level.
Change the inverter operation point to a different value and repeat the test
The test should cover the entire operation range of the inverter
1) Short-term voltage stability Large voltage disturbance tests
2) FRT (Low/High/Zero voltage Testing the voltage ride through capability
ride through) Low voltage ride through test: with a fixed P, Q output, simulate the grid
3) Short-circuit current voltage dips for a specified duration. The depth (80% to 25% remaining
contribution voltage) and duration should be selected according to the standards that are
applicable. In general, the duration is inversely proportional to the depth of
the voltage dip, i.e. the lower the voltage dip, the shorter time the dip
duration.
Zero voltage ride through test: with a fixed P, Q output, simulating grid
voltage dips down to very low level (15% to 5% remaining voltage) for a fixed
duration, say 150ms.
High voltage ride through test: with a fixed P, Q output, simulating the grid
voltage rises for a specified duration. The magnitude and duration should be
selected according to the standards that are applicable.
Rotor angle stability Large frequency and voltage disturbance test with synchronous generators
and long transmission lines:
Do this at different operating points, simulate the Out Of Step (OOS) with
acceleration of the synchronous generators. First swing OOS and multi-swing
OOS should be represented.
Solar radiation variation study Solar Irradiance disturbance test: start with operation at rated 1.0 pu power
output, change the solar irradiance power input (PV cell output)**.
Measurements: P, Q and V, I at the Grid connection point
Frequency instability study Set the DC system operating at its 1.0 p.u. rated power level,
Conduct a generator tripping or line tripping test which causes system
separation:
Do this at different operating point, simulating the frequency drop.
Unintentional islanding Change the loading balance conditions.
detection studies Change the type of loads, such as static load and induction motor load
*: Power flow based analysis for long-term voltage stability is out of scope.
**: It depends on geographic location, installation, PV panel and time and season variation.

293
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

LINKS AND REFERENCES OF APPENDIX C

[1] J. V. Milanović, K. Yamashita, S. Martínez Villanueva, S. Ž. Djokic and L. M. Korunović,

"International Industry Practice on Power System Load Modeling," IEEE Trans. on Power
Systems, pp. 3038-3046, August 2013.
[2] NERC Standard TPL-001-4, "Transmission System Planning Performance Requirements,"
January 2016, URL:
http://www.nerc.com/_layouts/PrintStandard.aspx?standardnumber=TPL-001-
4&title=TransmissionSystem/Planning/Performance/Requirements&jurisdiction=United/States.
[3] Western Electricity Coordinating Council (WECC) Renewable Energy Modeling Task Force,
URL: https://www.wecc.biz/PCC/Pages/MVWG.aspx.
[4] International Electrotechnical Commission (IEC) TC88 WG27, "Wind turbines - Electrical
simulation models for wind power generation," URL:
http://www.iec.ch/dyn/www/f?p=103:14:0::::FSP_ORG_ID,FSP_LANG_ID:5613,25.
[5] R. Elliott, A. Ellis, P. Pourbeik, J. J. Sanchez-Gasca, J. Senthil and J. Weber, "Generic
Photovoltaic System Models for WECC – A Status Report," in 2015 IEEE Power & Energy
Society General Meeting, Denver, July 2015. pp. 1-5.
[6] G. Lammert, L. D. Pabón Ospina, P. Pourbeik, D. Fetzer and M. Braun, "Implementation and
Validation of WECC Generic Photovoltaic System Models in DIgSILENT PowerFactory," in 2016
IEEE Power & Energy Society General Meeting, Boston, July 2016. pp. 1-5.
[7] P. Pourbeik, „Proposal for the DER_A model, “Presentation at WECC meeting, November 2016.
[8] G. Lammert, K. Yamashita, L. D. Pabón Ospina, H. Renner, S. Martínez Villanueva, P. Pourbeik,
F.-E. Ciausiu and M. Braun, "Modelling and Dynamic Performance of Inverter Based Generation
in Power System Studies: An International Questionnaire Survey," in CIRED conference,
Glasgow, 2017. (to be published)
[9] P. Kundur, Power System Stability and Control, Mc-Graw Hill, 1994.
[10] O. Ruhle and F. Balasin, "Simulations of Power System Dynamic Phenomena," in 2009 IEEE
PowerTech, Bucharest, June-July 2009. pp. 1-6.
[11] H. W. Dommel, "Digital Computer Solution of Electromagnetic Transients in Single- and
Multiphase Networks," IEEE Trans. on Power Apparatus and Systems, pp. 388-399, April 1969.
[12] Western Electricity Coordinating Council (WECC) Renewable Energy Modeling Task Force,
"WECC solar PV dynamic model specification," September 2012, URL:
https://www.wecc.biz/Reliability/WECC/Solar/PV/Dynamic/Model/Specification/-
/September/2012.pdf.
[13] G. Lammert, K. Yamashita, L. D. Pab´on Ospina, H. Renner, S. Martinez Villanueva, P.
Pourbeik, F.-E. Ciausiu, and M. Braun, “International Industry Practice on Modelling and
Dynamic Performance of Inverter Based Generation in Power System Studies,” CIGRE Science
& Engineering Vol. 8, June 2017.
[14] Western Electricity Coordinating Council (WECC) Renewable Energy Modeling Task Force,
August 2010, “WECC PV plant power flow modeling guidelines,” pp. 1-15.
[15] S. Soni, G. G. Karady, M. Morjaria and V. Chadliev, April 2014, “Comparison of full and reduced
scale solar PV plant models in multi-machine power systems,” 2014 IEEE PES T&D Conference
and Exposition, Chicago, pp. 1–5.
[16] P. Pourbeik, November 2016, “Proposal for the DER_A model,” Presentation at WECC meeting.
[17] K. Yamashita, S. Uemura, H. Kameda, K. Yokoi, S. Fubo and A. Takeuchi, “Verification of model
accuracy of islanding protection relay for synchronous generator,” in proc. of IEEJ annual
meeting, Vol. 6, No. 191, Japan, 2005 (in Japanese).
[18] IEEE C.37.2-2008, "Electrical Power System Device Function Numbers, Acronyms, and
Contact Designations," Oct. 2008.
[19] NERC Standard BAL-003-1 “Frequency Response & Frequency Bias Setting,” April 2015, URL:
http://www.nerc.com/_layouts/PrintStandard.aspx?standardnumber=BAL-003-
0.1b&title=Frequency Response and Bias&jurisdiction=United States
[20] K. Misak, Blackout im Übertragungsnetz – den Ernstfall beherrschen, 2. Fachtagung Blackout -
Auswirkungen, Maßnahmen und Konsequenzen auf allen Spannungsebenen, 6.- 7.5. 2014,
Hannover (in German).

294
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

[21] WECC, “WECC Guide for Representation of Photovoltaic Systems In Large-Scale Power flow
Simulations. WECC Modelling and Validation Work Group. Jan. 2011, URL:
https://www.wecc.biz/_layouts/15/WopiFrame.aspx?sourcedoc=/Reliability/WECC/PV/Plant/P
ower/Flow/Modeling/Guidelines/-/August/2010.pdf&action=default&DefaultItemOpen=1
[22] EPRI Technical Update – Generic Models and Model Validation for Wind Turbine Generators
and Photovoltaic Generation, URL:
http://www.epri.com/abstracts/Pages/ProductAbstract.aspx?ProductId=000000003002001002
[23] I. Green, “CAISO Experience with Impact of High Penetration of Renewable Resources on
Short-term Voltage Stability,” IEEE PES GM 2015, Panel - Challenges of Voltage and Reactive
Power Control from Renewable Resources, July 2015.
[24] Chen Wei, “Reactive power capability of SG2500HV_SG3000HV Inverter,” Aug. 2016.
[25] N.W. Miller, "Keeping It Together: Transient Stability in a World of Wind and Solar
Generation," in IEEE Power and Energy Magazine, vol.13, no.6, pp.31-39, Nov.-Dec. 2015.
[26] R. Preece and J. V Milanović, “Assessing the Applicability of Uncertainty Importance
Measures for Power System Studies,” IEEE Trans. Power Syst., 2015, available online.
[27] P. Lund, C. L. Bak, P. Thogersen, Z. Chen, C. Liu, K. Sun, and Z. H. Rather, “A Systematic
Approach for Dynamic Security Assessment and the Corresponding Preventive Control
Scheme Based on Decision Trees,” IEEE Trans. Power Syst., vol. 29, no. 2, pp. 717–730,
2014.
[28] L. Shi, S. Sun, L. Yao, Y. Ni, and M. Bazargan, “Effects of wind generation intermittency and
volatility on power system transient stability,” IET Renew. Power Gener., vol. 8, no. 5, pp.
509–521, 2014.
[29] M. Fan, V. Vittal, G. T. Heydt, R. Ayyanar, “Probabilistic power flow analysis with generation
dispatch including photovoltaic resources,” IEEE Trans. Power Syst., vol. 28, no. 2, pp. 1797–
1805, 2013.
[30] S. Eftekharnejad, G. T. Heydt, L. Fellow, V. Vittal, “Optimal Generation Dispatch With High
Penetration of Photovoltaic Generation,” in IEEE Power & Energy Society General Meeting,
26-30 July 2015.
[31] P. N. Papadopoulos, A. Adrees, J. V. Milanovic, “Probabilistic Assessment of Transient
Stability in Reduced Inertia Systems,” PES General Meeting 2016, Boston, July 2016.
[32] A. Adrees, P. N. Papadopoulos, J. V. Milanovic, “Framework to Assess the Effect of Reduction
in Inertia on System Frequency Response,” PES General Meeting 2016, Boston, July 2016.
[33] R. Preece, J. V. Milanovic, “Efficient Estimation of the Probability of Small-Disturbance
Instability of Large Uncertain Power Systems,” IEEE Trans. Power Syst., vol. 31, no. 2, pp.
1063–1072, 2015.
[34] G. Rogers, Power System Oscillations. Kluwer Academic, 2000.
[35] WECC Wind Power Plant Dynamic Modeling Guide, WECC Renewable Energy Modeling
Task Force, January 2014.
[36] Wind turbines - Part 27-1: Electrical simulation models - Wind turbines, IEC 61400-27-1, 2015.
[37] WECC PV Power Plant Dynamic Modeling Guide, WECC Renewable Energy Modeling Task
Force, May 2014.
[38] T. Weissbach and E. Welfonder, "High-frequency deviations within the European Power
System: Origins and proposals for improvement," in IEEE Power Systems Conference and
Exposition, 2009.
[39] J. W. Ingleson and D. M. Ellis, "Tracking the Eastern interconnection frequency governing
characteristic," in IEEE PES General Meeting, 2005.
[40] I. Erlich, K. Rensch, and F. Shewarega, "Impact of large wind power generation on frequency
stability," in IEEE PES General Meeting, 2006.
[41] P. Pal and B. Chauduri, Robust Control in Power Systems. New York: Springer Science &
Business Media, 2005.
[42] “DSA Tools TSAT User Manual,” Powertech Labs Inc, 2011.
[43] T. Guo and J. V Milanović, “Online Identification of Power System Dynamic Signature Using
PMU Measurements and Data Mining,” IEEE Trans. Power Syst., 2015, available online.
[44] K. Yamashita and H. Satoh, “Small-scale Three-Phase Photovoltaic Inverter Model for Grid
Interconnection Studies,” in Proc. of CIGRE AORC 2017, Auckland, New Zealand. Sep. 2017.

295
 MODELLING OF INVERTER-BASED GENERATION FOR POWER SYSTEM DYNAMIC STUDIES

[45] H. Sato and K. Yamashita, “Three-Phase Photovoltaic Inverter Model for Unbalanced Fault,” in
Proc. of IEEJ Annual Conference of Power and Energy Society, Sep. 2018 (in Japanese).
[46] A. Ellis, M. Behnke, and R. Elliott, “Generic Solar Photovoltaic System Dynamic Simulation
Model Specification,” Sandia National Laboratories, 2013.
[47] A. Ellis, Y. Kazachkov, E. Muljadi, P. Pourbeik, and J. J. Sanchez-Gasca, “Description and
technical specifications for generic WTG models - A status report,” in Proceedings of 2011 IEEE
PES Power Systems Conference and Exposition (PSCE 2011), 2011.
[48] L. Ge, L. Zhu, L. Qu, L. Zhang, L. Zhao and N. Chen, “Model study of photovoltaic inverter for
bulk power system studies based on comprehensive tests,” proceedings of IET renewable
power generation 2015.
[49] EPRI, “Model User Guide for Generic Renewable Energy System Models,” NB 3002006525,
2015.
[50] P. Pourbeik, S. E. Williams, J. Weber, J. Sanchez-Gasca, J. Senthil, S. Huang, and K. Bolton,
“Modeling and Dynamic Behavior of Battery Energy Storage: A Simple Model for Large-Scale
Time-Domain Stability Studies,” IEEE Electrif. Mag., vol. 3, no. 3, pp. 47–51, Sep. 2015.
[51] P. Pourbeik, “Simple Model Specification for Battery Energy Storage System,” WECC REMTF,
MVWG and EPRI, NB 173.003, 2015.
[52] E. Sasano, K. Shinya, M. Matsumoto, and S. Horiuchi, “Demonstration projects for providing
ancillary services using different three types of large-scale battery systems,” CIGRE Paris
Session 2018 (to be published).
[53] K. Yamashita, O. Sakamoto, Y. Kitauchi, T. Nanahara, T. Inoue, T. Arakaki, and H. Fukuda, “A
frequency-stabilizing scheme for integrating photovoltaics into a small island grid,” in 2011 2nd
IEEE PES International Conference and Exhibition on Innovative Smart Grid Technologies,
2011, pp. 1–7.
[54] José Carvalho Martins. “Electrical Characteristics of Overhead and Underground Lines VHV,
HV, MV and LV”. Work Report to Support the Planning of Electrical Networks. EDP Distribuição.
Lisbon. Feb. 2017.
[55] A. Seman, “Autonomy Is a Must for Ship Systems,” in Future Force, Spring Edition, pp 17-19,
2014.
[56] M. Steurer, S. Woodruff, T. Baldwin, H. Boenig, F. Bogdan, T. Fikse, M. Sloderbeck, and G.
Snitchler, “Hardware-in-the-Loop Investigation of Rotor Heating in a 5 MW HTS Propulsion
Motor,” IEEE Trans. on Applied Superconductivity, Vol. 17, No. 2, Part 2, pp. 1595-1598, 2007.
[57] James Langston, Michael Steurer, Karl Schoder, John Hauer, Ferenc Bogdan, Isaac Leonard,
Tim Chiocchio, Michael Sloderbeck, Andrew Farrell, Jay Vaidya, and Kevin Yost, “Megawatt
Scale Hardware-in-the-Loop Testing of a High Speed Generator,” in Proc. ASNE Day, May 9-
10, 2012, Arlington, VA.
[58] C. Schacherer, J. Langston, M. Steurer, M. Noe, “Power Hardware-in-the-Loop Testing of a
YBCO Coated Conductor Fault Current Limiting Module,” IEEE Trans. on Applied
Superconductivity, Vol. 19, No. 3, Part 2, June 2009 pp. 1801 – 1805.
[59] T. Chiocchio, R. Leonard, Y. Work, R. Fang, M. Steurer, A. Monti, J. Khan, J. Ordonez, M.
Sloderbeck, and S. L. Woodruff, “A Co-Simulation Approach for Real-Time Transient Analysis
of Electro-Thermal System Interactions on Board of Future All-Electric Ships,” in Proc. of
Summer Computer Simulation Conference, July, 2007.
[60] M. O. Faruque, M. Sloderbeck, M. Steurer, and V. Dinavahi, “Thermo-electric co-simulation on
geographically distributed real-time simulators,” in Proc. of the IEEE Power & Energy Society
General Meeting, Calgary, AB, Canada, July, 2009.
[61] James Langston, Mike Sloderbeck, Mischa Steurer, Don Dalessandro, Tom Fikse, “Role of
Hardware-in-the-Loop Simulation Testing in Transitioning New Technology to the Ship,” in Proc.
of IEEE Electric Ship Technologies Symposium (ESTS), April 2013.
[62] K. Schoder, Z. Cai, S. Sundararajan, M. Yu, M. Sloderbeck, I. Leonard, M. Steurer, “Real-Time
Simulation of Communications and Power Systems for Testing Distributed Embedded
Controls,” Engineering & Technology Reference, 10.1049/etr.2015.0022, ISSN 2056-4007,
www.ietdl.org

296