Modeling Inverter-Based Resources in Stability Studies

Songzhe Zhu, David Piper, Deepak Ramasubramanian, Ryan Quint, Andrew Isaacs, Rich Bauer

Abstract—The NERC Inverter-Based Resource Performance

Task Force (IRPTF) is developing recommended performance
characteristics for inverter-based resources connected to the bulk
power system. This paper provides modeling considerations and
simulations results from studies performed by the NERC IRPTF.
A primary focus is on modeling capability and impacts of
momentary cessation on grid stability.

Index Terms— frequency response, inverter-based resource,

momentary cessation

I. INTRODUCTION
Several transmission system faults in the Southern
Figure 1: Impact to BPS Bus Voltages for a 500 kV Fault in the Southern
California region of the Western Electricity Coordinating California Region [3]
Council (WECC) footprint have resulted in the loss of solar
photovoltaic (PV) resources connected to the bulk power bolted fault at a key bus location in the Southern California
system (BPS). The first significant event was a phase-to-phase region. This illustrates the widespread impact that extra high
fault that occurred on August 16, 2016 near Lugo substation voltage (EHV) faults can have. Thus, the importance of
and resulted in the widespread loss of 1200 MW of BPS- inverter-based resources to have robust ride-through
connected solar PV plants [1]. This “Blue Cut Fire” disturbance capabilities and controls that support grid stability and
initiated a NERC-WECC joint task force to explore the reasons reliability.
for the solar PV tripping. The majority of these resources (700
MW) tripped in response to an erred frequency measurement NERC initiated the Inverter-Based Resource Performance
during the fault. The task force also discovered that the vast Task Force (IRPTF) to explore these disturbances and develop
majority of BPS-connected solar PV resources may use recommended performance specifications for inverter-based
momentary cessation (MC), which is the temporary ceasing to resources connected to the BPS in North America. The NERC
energize (zero AC current injection) the inverter controls while IRPTF formed a modeling and simulations sub-group of
connected to the grid in response to a disturbance in terminal industry experts and Transmission Planning engineers in the
voltage [2]. WECC. This group is identifying any modeling deficiencies,
and assessing reliability of the Western Interconnection under
Another event on October 9, 2017, coined the “Canyon 2 high penetrations of inverter-based resources under different
Fire” disturbance, resulted in the widespread reduction of 900 control paradigms.
MW of BPS-connected solar PV generation after faults
occurred on 230 kV and 500 kV transmission lines. In this case, II. CHARACTERISTICS OF MOMENTARY CESSATION AND
many inverters momentarily ceased the production of AC TRIPPING EVENTS
currents during the fault, followed by other inverters that
tripped due to a transient overvoltage that occurred after the The Blue Cut Fire, Canyon 2 Fire, and other fault events
faults cleared. Similar to the Blue Cut Fire event, these faults have led to the identification of common performance
cleared normally in under 3 cycles, and no solar resources that characteristics that inverters at PV plants can exhibit during and
tripped were directly connected to the transmission element(s) after transmission faults. Inverter tripping and momentary
that experienced the fault. The impacted area of solar loss cessation are the two most disruptive responses that have been
reached upwards of 2500 km2 around the fault location. Figure observed.
1 shows the deviation in grid voltage for a 500 kV, 3-phase, High speed power system monitoring devices such as
Songzhe Zhu is with California Independent System Operator. digital fault recorders (DFR) or phasor measurement units
David Piper is with Southern California Edison (PMU) can be used to analyze fast system events such as
Deepak Ramasubramanian is with Electric Power Research Institute. momentary cessation. Figure 2 was developed using a DFR that
Ryan Quint is with North American Electric Reliability Corporation. was measuring a plant’s generation tie line during the Blue Cut
Andrew Isaacs is with Electranix. Fire event. This figure shows that a significant portion of the
Rich Bauer is with North American Electric Reliability Corporation.
plant entered into momentary cessation and the total plant

978-1-5386-7703-2/18/$31.00 ©2018 IEEE

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on February 21,2022 at 21:31:39 UTC from IEEE Xplore. Restrictions apply.
 TABLE I
MOMENTARY CESSATION PARAMETERS

Symbol QUANTITY Unit

Vmc Momentary cessation voltage threshold Voltage (pu)

Delay between voltage recovery and the start
Δtsr Time (s)
of power recovery
Δtrr Total time of recovery Time (s)
The % of a plant that enters into momentary
%MC %
cessation
RR Ramp rate Per unit/s

Voltage
Figure 2: Characteristics of Momentary Cessation Vmc

output was significantly reduced. A small portion of the plant

rode through the event and continued to inject power. The plant
reactive power was also affected. The plant had been
Δtsr Δtrr
consuming reactive power prior to the event. After the majority

Power
of the plant’s inverters shut down, the plant’s VAR losses
RR
significantly reduced. Plant-side fixed capacitors remained %MC
in-service and the net VAR flow reversed direction with the
plant injecting VARs into the system. As the plant output began
to recover, the VAR flow returned to the pre-fault value. Time
Figure 4: Characteristics of Momentary Cessation
Figure 3 shows the characteristic response of a PV plant
during an event that caused a significant portion of the plant to III. INVERTER CONTROLS AND POSITIVE SEQUENCE
trip offline. PV inverter tripping events can be identified using MODELS
slower recording devices such as SCADA systems because the
power output remains reduced for several minutes. New technology variable energy resources (e.g., Type 3 and
4 wind, solar PV) are interfaced to the BPS through power
The IRPTF has defined a set of standardized metrics to electronics that use control algorithms to define their
quantify the performance of PV inverters during momentary performance characteristics. Solid state switches in the
cessation or tripping events. Table I shows the parameters that converters enable fast and programmable control, unlocking
can be used to quantify the performance of PV plants during new capabilities as well as new challenges. Inverter-based
momentary cessation events. The symbols in Table I resources use power electronic controls to change active and
correspond to Figure 4. The aggregate reduction in energy reactive current injection, and subsequently active and reactive
[MW*s] that momentary cessation and inverter tripping causes power output. For these reasons, inverter-based resources have
has a direct and negative impact on system frequency. As a the capability of responding to disturbances nearly
result, accurately modeling these performance characteristics in instantaneously.
power system studies is critical to ensure system reliability and
resiliency to power system faults [4]. Several key assumptions and simplifications are made in
commercially available positive sequence simulation tools that
are used to model the BPS in planning and operating studies.
Voltages and currents are assumed to be balanced and therefore
only the positive sequence impedance network is required to
adequately represent the system. Historically, stability models
have been developed around the premise that the system is
dominated by large rotating machines. As a result, the models
were developed assuming voltage and frequency swings on the
order of 0-10 Hz, but typically 0-5 Hz, which may be changing
with higher penetrations of inverter-based resources [5]. Lastly,
positive sequence transient stability simulations typically use
an integration time step around 4 ms, which does not capture
many of the faster dynamics of inverter-based control systems.
As converters do not have the same inherent physical
characteristics as synchronous machines, each manufacturer
can apply their own control algorithms to make their product
Figure 3: Characteristics of Tripping competitive with other vendors while still meeting relevant

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on February 21,2022 at 21:31:39 UTC from IEEE Xplore. Restrictions apply.
distribution- and transmission-level performance requirements Iplim
In-Run
[6,7]. The unique aspects of different inverter models led to the Vt
EPCL Iqlim
creation of customized transient stability models to accurately
capture the unique control and protection features. The Iplim
proliferation of custom models led to the development of Ipcmd Pgen
Electrical
generic converter models for use in interconnection-level BPS Controls Iqlim Converter
planning studies [5,8]. However, the construction and Block
Model
Iqcmd Qgen
assumptions used in these models lead to limitations as
discussed in [9]. Of particular interest in this paper are the Figure 6: Simplified Block Diagram of In-Run EPCL
limitations of the generic positive sequence models to model
momentary cessation and the recovery from momentary (PV1G/PV1E, WTG4, etc.) to REEC_A was considered, but
cessation characteristic. To accurately model momentary due to the lack of an accurate mapping between the models, the
cessation, five elements are required: in-run EPCL method was preferred.
1. Both active and reactive currents go to zero if the The in-run EPCL model was designed to interact with the
voltage is outside a threshold band electrical control blocks (REEC_A, REEC_B, PV1E, WT4E)
2. A definite non-negative time delay between the instant and effectively supervise the maximum real and reactive
of voltage recovery to within the threshold band and current command that can be sent to the converter models
the start of recovery of current injection (REGC_A, PV1G, WT4G) as shown in Figure 6. The inverter
3. Ramp rate limits on active current terminal voltage magnitude is used in conjunction with logic for
4. Control of reactive current during recovery the five elements of momentary cessation to determine the
5. Priority between active and reactive current injection maximum real and reactive current command. The advantage
of using such a user-defined model is that it is flexible and can
Among the various available generic models that are be customized as required. Additionally, it does not require a
currently available, REEC_A has the best capability to change in the values of the parameters provided by the
represent the five elements (see Figure 5) [10]. In REEC_A, the Generator Owner upon model submission. However, due to the
VDL1 and VDL2 voltage dependent current limit characteristic nature of the model, it is only a temporary workaround for the
can be used to model momentary cessation to a certain limitations of the generic models and can thus cannot be used
rudimentary extent. REEC_A can also model the time delay in in regular planning/operation studies. The next section
recovery of active current using the Voltage_dip and thld2 describes simulation results applying this model to the case.
parameters, but cannot model a delay in reactive current
recovery. While the generic REEC_A model is the most IV. STUDY RESULTS
flexible for modeling momentary cessation, the REEC_B
model has typically been used to represent large BPS-connected The NERC IRPTF is performing interconnection-wide
solar PV plants. However, the first two elements above cannot reliability studies for the Western Interconnection to evaluate
be modeled in the generic REEC_B. the impacts of inverter momentary cessation. Operating
conditions representing high renewable penetration and low
In order to analyze the impacts momentary cessation may inertia were established based on historical load, wind, and
have on BPS reliability, the IRPTF chose to develop a custom solar output data. The initial analyses focus on frequency
script (“in-run EPCL”) that is run during the course of the response impacts from fault-induced widespread inverter
simulation, with the capability to represent the five elements of momentary cessation. Simulations were carried out in GE
momentary cessation [10]. This in-run EPCL was applied to all PSLF™ transient stability simulation program [10].
existing models (REEC_A, REEC_B, PV1E, WT4E, etc.).
Conversion of REEC_B and first generation generic models A. Momentary Cessation Impacts on Frequency Response
When a 3-phase fault occurs, a widespread region of the
BPS experiences a voltage depression during the fault (Figure
1). Many inverter manufacturers have stated that existing
inverters use a momentary cessation voltage low threshold
around 0.9 pu. Initial studies show that almost 10 GW of BPS-
connected inverter-based resources could see a drop in voltage
below 0.9 pu during a severe fault in the Southern California
area, and therefore could potentially case energization. Figures
7-10 compare system-wide bus frequencies with and without
momentary cessation modeled. In both simulations, a
temporary 3-phase fault is applied at t=0 and then cleared after
4 cycles. Key takeaways from these simulations include:
• Inverters that use momentary cessation but do not
include a delay in recovery and a very fast ramp rate
(e.g., 10 pu/sec) could cause a significant frequency
deviation but do not necessary pose a risk to BPS
Figure 5: REEC_A Model [10] reliability at these penetration levels.

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on February 21,2022 at 21:31:39 UTC from IEEE Xplore. Restrictions apply.
 • Increasing delay in momentary cessation results in a interactions or tripping of the plant to voltage and frequency
longer power imbalance in the system, resulting in disturbances within the PRC-024-2 ride through curves [7,11].
larger frequency deviations.
• Ramp rate and/or delays in recovery from momentary
cessation that exceed ~0.5 seconds may result in
unacceptable grid performance such as triggering the
first stage of underfrequency load shedding (UFLS) or
causing pump loads to lose stability.

B. Inverter Performance Comparison Figure 7: Bus Frequencies – without MC

Momentary cessation caused by a fault is typically
categorized into:
• Undervoltage Momentary Cessation: inverters cease
energization for terminal voltage less than some
threshold (e.g., 0.9 pu)
• Overvoltage Momentary Cessation: inverters cease
energization for terminal voltage greater than some Figure 8: Bus Frequencies – MC w/ no recovery delay
threshold (e.g., 1.1 pu)
Tuning of inverter gains and time constants is critical to
ensure a stable and effective response for each resource
connected to the BPS. All generators, synchronous and non-
synchronous, reduce active power output during a fault due to
the low terminal voltage. A synchronous machine will
accelerate when electrical power output decreases without an
appreciable change in mechanical power input due to the power Figure 9: Bus Frequencies – MC w/ 0.5 sec recovery delay
imbalance and injected energy into the rotor. On the other hand,
an inverter does not have this type of stability problem. In
addition, an inverter-based resources uses closed-loop controls
that can provide fast response in active and reactive power;
however, use of momentary cessation should be coordinated
with these controls to avoid any adverse impacts on the plant’s
response to changes in terminal voltage and frequency
conditions [11]. Figure 10: Bus Frequencies – MC w/ 0.5 sec delay (first 1.2 sec)
In cases where controls are not tuned properly, a number of
stability issues may arise. Figure 11 shows a fault simulation V. CONCLUSIONS AND RECOMMENDATIONS
where the inverter-based resource continues to oscillations with
controls stability issues. Figure 12 shows an example It is important to accurately model momentary cessation
simulation where active current priority is set. The inverter control philosophy for existing resources. This is particularly
delivers a ramped active current injection after fault clearing, important since many inverter manufacturers have stated that
which drives the voltage lower since there’s insufficient legacy equipment and controls may not be easily re-
reactive current injection. Upon voltage hitting the MC low programmed to eliminate its use. Clear guidance is needed for
voltage threshold, the inverter re-enters MC and the cycling generator owners to provide accurate PV plant models that
continues. Figure 13 shows the simulated response of an reflect momentary cessation, as applicable, and for
inverter to a fault where voltage does not enter the MC transmission planners to be able to assess its impact to the BPS.
threshold. However, the inverter provides significant reactive Existing generic models for inverter-based resources used
current injection which drives the voltage high upon fault in positive sequence simulation software have limitations for
clearing. The inverter then enters high voltage MC with a modeling momentary cessation. Particularly in capturing the
delayed recovery affecting grid frequency. delayed recovery of momentary cessation observed in some
In general, inverter control priority should be given to inverters’ response to actual grid events. The generic models
reactive current to help stabilize local grid voltage. However, should be updated to create an ability to accurately represent
active current injection should also be held to pre-contingency momentary cessation for existing resources.
values and any reduction in active current during fault
conditions should be restored very quickly upon fault clearing
to assure frequency stability. The overall response of active and
reactive current should ensure a stable, well damped control
that returns to within acceptable limits without any control

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on February 21,2022 at 21:31:39 UTC from IEEE Xplore. Restrictions apply.
 Figure 11: Poor Inverter Control Response (P,Q,v,f) Figure 13: High Voltage MC Response (P,Q,v,f) – 0.5sec Recovery
Delay

REFERENCES
[1] North American Electric Reliability Corporation, “1,200 MW Fault
Induced Solar Photovoltaic Resource Interruption Disturbance Report:
Southern California 8/16/2016 Event,” NERC, Atlanta, GA, June 2017.
[2] IEEE Standard for Interconnecting Distributed Resources with Electric
Power Systems, IEEE Std. IEEE P1547 (full revision), Oct. 2017.
[3] ABB Ventyx, 2017.
[4] D. Piper and D. Donaldson, “Response of Grid Interconnected Solar PV
to Transmission System Faults,” unpublished, IEEE PVSC-44,
Washington, DC, June 2017.
[5] Western Electricity Coordinating Council, “WECC Solar Plant Dynamic
Modeling Guidelines,” WECC, Salt Lake City, UT, May 2014.
[6] Standard for Inverters, Converters, Controllers and Interconnection
System Equipment for Use with Distributed Energy Resources, UL 1741-
SA, June 2016.
[7] Generator Frequency and Voltage Protective Relay Settings, NERC
PRC-024-2, NERC, Atlanta, GA, May 2015.
[8] Western Electricity Coordinating Council, “WECC Second Generation
Wind Turbine Models,” WECC, Salt Lake City, UT, January 2014.
[Online]. Available:
https://www.wecc.biz/Reliability/WECC-Second-Generation-Wind-
Turbine-Models-012314.pdf
[9] P. Pourbeik et al., "Generic Dynamic Models for Modeling Wind Power
Plants and Other Renewable Technologies in Large-Scale Power System
Studies," in IEEE Transactions on Energy Conversion, vol. 32, no. 3, pp.
1108-1116, Sept. 2017.
[10] "PSLF," GE Energy Consulting. [Online]. Available:
http://www.geenergyconsulting.com/
Figure 12: Active Power Priority Control Response (P,Q,v,f) [11] North American Electric Reliability Corporation, “Integrating Inverter-
Based Resources to Low Short Circuit Strength Systems,” NERC,
Atlanta, GA, Dec 2017, in press.

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on February 21,2022 at 21:31:39 UTC from IEEE Xplore. Restrictions apply.