1

Validation of Power System Models

Eric Allen, Senior Member, IEEE, Dmitry Kosterev, Senior Member, IEEE, and Pouyan Pourbeik, Fellow, IEEE

Abstract—Since models form the basis for most power 2) Generating Units: In today’s power systems this
system studies, power system model validation is an includes not only conventional hydro and steam
essential procedure for maintaining system security and and gas turbines but also combined cycle power
reliability. The procedure may be viewed as a “top-
down” approach to model verification; comparisons with plants and wind turbine generators. There is an
measured data indicate the quality of the overall model. imminent need for modeling distributed genera-
Analysis of the differences demonstrates which subsystem tion (e.g. solar, micro-turbines, fuel-cells etc.).
component models need to be revalidated. Numerous ex- 3) Load: Representing the electrical load in the
amples are presented to illustrate the use and importance system, which range from simple light-bulbs to
of system model validation.
large industrial facilities.
A steady-state model for each of the above cate-
I. I NTRODUCTION gories of components (transmission elements, gener-
Models are the foundation of virtually all power ators, and load) is developed. The individual models
system studies. Calculation of operating limits, plan- are then combined into a complete system model
ning studies for assessment of new generation and load for representing steady-state behavior of an entire
growth, performance assessments of system integrity interconnection. This model is known as the “power
protection schemes (SIPS) – all of these studies de- flow” model. For some studies, remote parts of a
pend on an approximate mathematical representation large interconnection far from the locations of interest
of the transmission, generation, and load. If a particular are represented using reduced-size models known as
model does not represent observed phenomena on the “equivalents.”
power system with reasonable accuracy, then how can Models that represents the dynamics of the active
one have confidence in the studies derived from that devices are also developed for each of the categories
model? of equipment listed above. By active devices here
The purpose of this paper is to call attention to the we mean any power system component that would
importance of model validation. The paper begins with have a time-varying characteristic for the time-domain
an overview of power system models in section II. The simulation of events. For stability studies, the charac-
importance of system model validation is examined teristics of concern typically range in the few tens of
with several practical examples in section III. The milliseconds to many seconds time frame. Thus, in this
process of system model validation is described in sec- context, active devices would be power plants and their
tion IV. Improvements to the process are discussed in controls, certain components of loads, power electronic
section V. Finally, conclusions and recommendations transmission devices (i.e. FACTS and HVDC), and,
are summarized in section VI. for some studies, on-load tap changers, Programmable
Logic Controller (PLC) controls on shunt devices,
remedial action schemes, and other similar control
II. OVERVIEW OF P OWER S YSTEM M ODELS
devices. The components in the power flow model
In modeling a large power system, such as the need to be matched with their corresponding dynamics
western or eastern interconnection in North America, models.
there are several categories of models that need to be For transmission lines, transformers, and shunt ca-
developed: pacitors/reactors, model development means making an
1) Transmission System: This includes not only accurate calculation of the line parameters and obtain-
transmission lines, power transformers and me- ing accurate name plate information (for transformers
chanically switched shunt capacitors and reac- etc.) to be incorporated into the steady-state network
tors, but also many other transmission devices model. For FACTS and HVDC, it means working
such as phase-shifting transformers, flexible ac closely with the equipment vendors upon commission-
transmission systems (FACTS), and several high- ing to obtain a validated model based on comparisons
voltage dc (HVDC) transmission systems. with detailed 3-phase models or field/factory tests. For
generation, one should ideally obtain models that have
E. Allen is with the North American Electric Reliability Corpo-
ration (NERC), Princeton, NJ, 08540 USA. been validated through tests at commissioning (for
D. Kosterev is with the Bonneville Power Administration (BPA), new plants) including in particular validated generator
Vancouver, WA 98666 USA reactive capability. For load models, an aggregated
P. Pourbeik is with the Electric Power Research Institute (EPRI),
Knoxville, TN 37932 USA static/dynamic model from a component based ap-
Manuscript received January 29, 2010. proach using data on customer types and categories at

978-1-4244-6551-4/10/$26.00 ©2010 IEEE

 2

substations obtains a first estimate of load composition.

Such work would have to be done at multiple times of
the year, e.g. summer peak, winter peak, fall and light 4600 Observed COI Power (Dittmer Control Center)
load. 4400

4200

III. I MPORTANCE OF M ODEL VALIDATION 4000

Grid planning and operating decisions are based on 4600 Simulated COI Power (initial WSCC base case)

the results of power system simulations. Simulations 4400

rely on power system models to predict system per-

4200

formance during anticipated disturbance events. Both

4000
technical and commercial segments of the industry
0 10 20 30 40 50 60 70 80 90
must be confident that the dynamic simulation models Time in Seconds

and database are accurate and up to date. Optimistic

models can result in grid under-investment or unsafe Fig. 1. Comparison of observed system response during August 10,
operating conditions and ultimately wide-spread power 1996 disturbance with the response using the models then in use.
outages, such as occurred in the summer of 1996 in
the Western Interconnection. On the other hand, pes-
simistic models and assumptions can result in overly MW, and transfer capability of Pacific HVDC Intertie
conservative grid operation and under-utilization of was de-rated from 3,100 MW to 2,000 MW. Successful
network capacity. Pessimistic models can also lead to completion of model validation studies was one of
unnecessary capital investment, thereby increasing the the key requirements for the restoration of the intertie
cost of electric power. Therefore, having realistic mod- ratings.
els is very important to ensure reliable and economic The August 10, 1996 event was a complex distur-
power system operation. bance that included voltage collapse, growing power
Model validation needs to be done on a regular basis. oscillations and loss of generation. The lessons learned
There will always be evolving changes in the char- from the validation studies had a monumental impact
acteristics of the power system over time, particular on how the studies are done :
with respect to loads. Unforeseen interactions can also 1) A major revision of the generator model dy-
occur when new control strategies are implemented namic database was required. Following the out-
through the addition of novel devices and technologies. age, the Western Systems Coordinating Council
The models must therefore be checked periodically to (WSCC)1 required that all generators larger than
ensure that trends in the power system which can affect 10 MW be tested for the purpose of model data
reliability are captured in system studies. verification. The benefits achieved by the West-
To predict power system performance, one must ern model validation program are indisputable
be able to model correctly all aspects of the system [2]:
performance. In order to model system performance, • The first benefit was from the simple fact of
one must understand the system’s behavior and as- having experienced staff visit power plants
sumptions that go into model development. To under- to review the correlation between dynamic
stand system behavior, one must observe, measure and models used to represent them and the ac-
analyze its actual behavior. Observation, measurement tual equipment. There were cases where
and analysis of multiple system events provide the best inappropriate models were used to represent
opportunity for model validation. generating controls, e.g. a static excitation
system was represented by a rotating DC
A. 1996 Western Interconnection Outages exciter model. There were cases when pieces
Two major disturbances occurred in the Western of equipment were modeled that were out of
Interconnection in summer of 1996, on July 2 and service, such as Power System Stabilizers.
• Model data was improved significantly in
August 10. When planning engineers attempted to re-
produce both events in simulations, there was no match many cases, as demonstrated by a number
between the simulations and the actual disturbance of before- and after-test examples. Excitation
recordings [1], as shown in Figure 1. The failure of system models were the most improved,
the simulations to correlate, even remotely, with the particularly for older analog controls.
• Improved generator control tuning was a
actual disturbance data was alarming for utilities and
grid operators in the West, since many of the critical positive “side effect” of the testing program.
paths are stability limited, and the models are used Having an experienced engineer with ap-
for setting their operating transfer limits. Following propriate instrumentation on site presented
the outages, operating transfer capability of California- 1 This regional council was superceded by the Western Electricity
Oregon Intertie was de-rated from 4,800 MW to 3,200 Coordinating Council (WECC) in 2002.
 3

a great opportunity to correct poorly tuned implementation, induction motor models were added
controls or detect and repair malfunctioning uniformly to the high side buses across the Western
equipment. Interconnection. The percentage of the motors was
2) Correct representation of generator reactive adjusted to get a good match for the North-South
power capabilities was identified as one of the oscillations for August 10, 1996 and August 4, 2000
key modeling issues. All generators larger than events. The model was recognized as “interim” and
10 MVA were required to test and demonstrate was used to address critical operation issues related
their reactive capabilities. An effort was made to damping of power oscillations on the California-
to test and document settings of generator over- Oregon Intertie [4]. It was also recognized that a
excitation limiters. Effects of plant-level controls more comprehensive model is needed, and the Western
on generator reactive power output need to be Electricity Coordinating Council (WECC) accordingly
recognized in the post-transient time frame. formed the Load Modeling Task Force to develop and
3) Modeled governor response was found to be implement a composite load model.
optimistic in the validation studies. A significant
portion of thermal generation had to be repre- C. WECC Frequency Response Reserve
sented as lacking frequency response in order
By the early 2000s, WECC had a large record of
to match the under-frequency profiles between
under-frequency events in the Western Interconnection.
simulations and reality. A governor “blocking”
Two observations were obvious:
recommendation was put in place for operational
• the simulated frequency dip was not as large as
and planning studies. Blocking thermal governor
response also affected damping of the inter-area the one observed in reality for the same amount
oscillations. (See also section III-C.) of generation lost
• the simulated power pick-up on California-Oregon
4) Controls of the Pacific HVDC Intertie (PDCI)
were a major contributing factor to voltage col- Intertie was not as large as one observed in reality
lapse and growing oscillations. A true multi- Since voltage stability of several transmission paths is
terminal HVDC system model was developed, affected by the distribution and amount of governor
and detailed control models were developed for response, the discrepancy between simulated and actual
PDCI. Following the event analysis, control mod- governor responses again raised concerns about safety
ifications were implemented to improve DC per- of operating limits on major transmission paths.
formance with respect to the grid [3]. A sequence of system tests was conducted in May
5) Automatic Generation Controls played a negative 2001 as a part of WECC efforts to establish a Fre-
role in the 1996 disturbance by replacing lost quency Responsive Reserve (FRR) Standard. Auto-
generation in the places further away from the matic Generation Control was disabled throughout the
intertie thereby increasing the system stress and entire Western Interconnection so that pure governor
eventually resulting in the instability. Effects of response could be recorded. The tests included planned
AGC action need to be acknowledged in mid- generation drops at Grand Coulee and Hoover dams.
term dynamic time frame. The tests revealed that a large portion of thermal
6) Finally, the study recognized the need for rep- generation in Southern part of the Western Intercon-
resenting the dynamic component of the load. nection is operated “baseloaded”, i.e. non-responsive
Motor load had to be added to the case to to under-frequency events. The analysis also revealed
get better agreement in oscillation damping and a wide use of load controllers that allow initial response
voltage collapse. to under-frequency but quickly withdraw the response
It took nearly six months of modeling work be- and return back to their set-point.
fore the validation studies could match the actual WECC used historic synchrophasor and SCADA
disturbance well. Similar modeling changes were later data to determine which units normally operate under
applied to the July 2, 1996 validation case. governor control, which units normally are on load
control, and which units are normally baseloaded.
These operating practices were reflected in the study
B. August 4, 2000 Western Interconnection oscillation cases. Several validation studies were performed for
event large generation outages that occurred in 2002 to
The Western Interconnection experienced a poorly validate the over-all system performance, specifically
damped oscillation on August 4, 2000. The event system frequency, power pick-up on major paths, and
presented a great opportunity for model validation. damping of inter-area oscillations.
Several modeling changes had to be put in place; the The governor modeling recommendation had a huge
most notable was revision of Power System Stabilizer impact on the studied transfer capability of major
data. The event also presented an opportunity to test paths in the Western Interconnection. Results of this
the motor load modeling assumptions that were derived modeling work fueled the development of Frequency
from the 1996 disturbance studies. For simplicity of Responsive Reserve standards by WECC.
 4

D. August 14, 2003 Eastern U.S./Canada Blackout simulated events. A hybrid modeling approach
The blackout that affected the northeastern United was developed by SCE and FPL engineers. A
States and Ontario on August 14, 2003 was comparable three-phase motor model is used to represent
to the legendary 1965 blackout. Fifty million people the running state. When the voltage drops below
and 61,800 MW of load were affected. A major part a designated stalling voltage (generally around
of the analysis of the blackout depended on the con- 60%), the motor model is replaced with a con-
struction of a power flow model of system conditions stant impedance representing a stalled motor [8].
preceding the disturbance. The effort required to build This approach was later enhanced by the Electric
this model was extensive (thousands of hours). The Power Research Institute (EPRI) working with
model was used to reconstruct the sequence of events Arizona Public Service (APS) [12].
and determine the causes and impacts of the individual • It is necessary to model a distribution system
events that took place. Construction of the model also equivalent to capture the voltage drop and reac-
revealed that the system models then in use underesti- tive power losses in the distribution system. The
mated load reactive power. The recommendations from WECC Load Modeling Task Force is completing
the studies of the blackout included model validation a multi-year effort on developing and implement-
and benchmarking [5]. ing a composite load model [13]. The model was
shown to reproduce in principle historic FIDVR
E. June 14, 2004 Westwing Event events and is now intended for use in voltage
stability assessments.
A major disturbance occurred on June 14, 2004,
where several protection relay failures resulted in a IV. M ODEL VALIDATION P ROCESS
fault lasting longer than 30 seconds and the trip of three The need for power system models has been dis-
Palo Verde nuclear units. The simulations of the event cussed in the previous sections with several illustrative
reproduced the system frequency and power pick-up on examples. Assuming that we have established a model
major paths reasonably well. However, the validation of the entire power system to be studied, the next
study could not reproduce reactive power output from logical question is “do we have a valid model?” The
Palo Verde generators during the depressed voltages. answer is not necessarily as obvious as it may seem.
The model was showing nearly 50% higher reactive It is often simple to identify a model that is not valid,
power response than the actual. Independently, BPA as for example depicted in Figure 1. The question of
compared measured and simulated reactive capabilities model validation is multifaceted. The eventual goal
for large hydro-power plants. Almost uniformly, the is to have a total system model that can reasonably
model was showing much greater reactive power output predict the outcome of an event (e.g. a model that
for a given set of boundary conditions (stator voltage, would show the growing oscillations in Figure 1, even
field current and active power). Such misrepresenta- if it did not match exactly); however, to achieve this,
tions have a major impact on studies related to fault- one needs to have individual constituents of the model
induced delayed voltage recovery (discussed next), to also be valid. The process of model validation and
when machines can be forced to their over-excitation the eventual “validity” of the model requires sound
limits. The issue was attributed to the treatment of “engineering judgment” rather than being based on a
machine saturation by the simulation programs. simple pass/fail of the model determined by a rigorous
criteria. This is because any modeling activity necessi-
F. Fault-Induced Delayed Voltage Recovery tates certain assumptions and compromises, which can
Fault-Induced Delayed Voltage Recovery (FIDVR) only be determined by a thorough understanding of the
is the phenomenon of transmission voltages remaining process being modeled and the purpose for which the
at significantly depressed levels (70 to 85% of pre- model is to be used.
disturbance) for several seconds after a fault is cleared. The current approach to model validation is to some
Multiple FIDVR occurrences have been observed in extent formulaic as it pertains to individual power
Southern California, Florida, and the southeast United system components – e.g. individual power plants.
States since the late 1980s. The phenomenon is related However, presently the entire interconnected power
to stalling of motor loads, especially residential single- system model is not taken through a systematic model
phase air-conditioners, in the area close to the fault. A validation process. What is presented here is a logical
severe FIDVR event can result in fast voltage collapse. progression from building up a power system model
The simulations of these events show instantaneous and applying best efforts to validate the individual
post-fault voltage recovery when using the “interim” components, and then extending the effort to “system
load model [6]. Studies conducted by Southern Califor- wide model validation.”
nia Edison (SCE) and Florida Power and Light (FPL)
concluded that: A. Steps in Component Model Validation
• Three-phase motor models did not reproduce the Presently, in North America, much effort is spent
air-conditioner stalling phenomenon during the in individual component model validation – one clear
 5

responses of system wide frequency excursions and

voltage and power fluctuations at major substations and
Line impedances,
transformers, etc. transmission lines. Parts of the power system model
Collect basic
that may require further refinement are thus identified.
Generation Collect basic Network
Dispatch & Load system data Topology
name plate data for all
active devices Adjusting the conditions in the power system model
to match the observed conditions of the actual power
Build Power Flow
Case
For all power plants
(and other active
devices, e.g. FACTS)
system is a major task but essential for model val-
do individual model
validation thorough
staged-testing or on-
idation. The model needs to represent the observed
Collect Steady- line disturbance
State
Measurements of
key tie-line flows
Compare Power Flow
Model with Measured
monitoring: Individual
Component Model
Valdiation
power system conditions (i.e. the initial steady-state
and Transmission
System EHV
Substation
Voltages and
System Steady-State
Condition:
Model Validation
conditions, which are constituted by the initial power
Angles
(PMUs)

Build Complete Dynamic

flow and nodal voltages etc.) for the moment in time
Model Database
Comparison is
Good
No (iterate)
just prior to the event that is used for the validation.
Yes
Various aspects of a power system model that need to
Combined Power flow
and Dynamic Model
Database
be matched to a specific event include:
1) network topology (lines in and out, general net-
Collect Time-
Synchronized
Measurements of key
tie-line flows and
Simulate the Event
and compare
work connections and impedances),
Transmission System Measured and time-
EHV Substation
Voltages and Angles
during a major
domain simulated
results:
Model Validation
Refine Models
2) generation dispatch,
disturbance
(PMUs)
3) load profile,
Comparison is
Good 4) load dynamics and composition, and
5) power plant dynamics.
DONE A relatively new enabling technology for system
wide model validation is phasor measurement units
Fig. 2. Process of system model validation.
(PMUs). Through using PMUs we can capture both
steady-state and time-domain varying response of the
power system for periods when the power system is
example is the efforts in the WECC for generator at an equilibrium point and when the power system is
model validation and certification [2], [15]. Compo- under a disturbance condition. Data can be captured
nent level testing can be done either through staged on a time synchronized basis including major nodal
tests [16]–[19] or on-line disturbance based model voltages and angles, and major tie-line real and reactive
validation [20]. Reference [21] presents an example of power flows. Then by comparing these measurements
model validation based on disturbance monitoring for to those predicted by our system wide models, we can
a FACTS device (SVC), while references [22], [23] assess the validity and adequacy of our models. Other
discuss much work done in developing better load sources of data for dynamics model validation include
models and component level model validation for load digital fault recorders (DFRs) and digital relay data.
models performed in the laboratory. Such component
level testing is of course a crucial step and necessary C. Making Refinements to the Model
part of system wide model validation.
Model refinement during the model validation pro-
For load models, data from distribution feeders can
cess for individual component modeling can to some
be used for validation. This data can then be used
extent be automated (see for example [20]). For system
for comparison with the aggregated models developed
wide model validation this is unlikely to be viable or
through the component based approaches to offer a
practical. In such efforts the focus should be on using
means of refining them and/or helping to identify
engineering judgment and a thorough understanding
ranges of potential variation in load composition and
of the system under study. An example of using dis-
parameters for sensitivity studies in planning analysis.
turbance monitoring to refine the entire power system
Research is still being done in this area.
model might be the case for a loss of a large generator
in the southern part of a system. In this case one
B. Steps in System-Wide Model Validation might see from recorded disturbance monitor data that
In contrast, systematic system wide model valida- power flow over a major north-south inter-tie ends up
tion, which we believe is the next logical step, is not being greater from the north to the south than what is
done routinely at present. Figure 2 shows the process predicted by simulations, providing a hint that perhaps
diagram for such system wide model validation. the spinning reserve and primary governing response
Power flow simulation results are compared with of units in the north are greater than actually modeled.
time-synchronized recordings of nodal voltages, an- Thus, it would help to identify models that need to be
gles, and key inter-tie flows during equilibrium con- refined. At this stage we may have to go back and look
ditions to ensure the adequacy of system wide power at a group of generating units so identified, one by
flow models. Also, the simulated dynamic response to one, to recognize the modeling refinements required.
a measured event is compared to dynamic recorded Again, if appropriate digital recording of individual
 6

plant response is available for the same event (e.g. used as measures of success. WECC is pursuing
as discussed in [20]) we would be able to identify an automated tool for generator model validation
the specific unit(s) under question by looking at a that can scan the dynamic performance of each
comparison of individual unit response to simulated Balancing Authoritys generating fleet following
individual unit response for the event. This iterative a grid disturbance. Another recently developed
refinement process is shown by the dotted lines in approach for individual power plant model val-
Figure 2. idation is the use of on-line disturbance monitor-
ing [20]. Once individual model components are
V. I MPROVEMENTS TO M ODEL VALIDATION validated, the system model validation can focus
on the remaining elements. System model adjust-
Looking forward, the following are areas for im- ments require engineering judgment to determine
provement of the model validation process: which system model components may need to
• Development of the validation power flow base be adjusted. Sensitivity studies are necessary to
cases. A power flow case is used as a starting determine the effect of system model changes.
point for dynamic simulations. Experience shows
that the base case must be reasonably close to the VI. C ONCLUSIONS AND R ECOMMENDATIONS
actual system conditions. In the past, a manual
Model construction and validation are important
process was used for creating the validation base
tasks that form the foundation of all power system
cases. Such process is prone to errors and very
studies. These tasks need to be performed regularly in
time consuming, e.g. development of validation
order to keep up with ongoing changes and additions
case for June 14, 2004 event took almost nine
to the power system.
months. Therefore, model validation studies are
not done as often as they would be if valida- Disturbances present great opportunities for model
tion cases were readily available. WECC success- verification and identification of the model improve-
fully implemented a West-Wide System Model ment needs. Over the past 12 years, the quality of
(WSM) project that develops a WECC-wide pow- the dynamic models and databases have improved
erflow snapshot of the system conditions based immensely in the Western Interconnection, in large part
on the WECC-wide state-estimator. Mapping the due to the commitments made to the model validation
dynamic database onto WSM base cases needs to efforts.
be completed to make the cases usable for model The emerging best practice for model maintenance
validation studies. These efforts are under way. is as follows:
• Availability of West-wide synchronized distur- 1) Modeling of the individual components of the
bance recordings. The Western Interconnection system (i.e. transmission lines, FACTS, HVDC,
has reasonably good synchro-phasor coverage. An power plants and aggregated load models at
issue is that many PMUs are not networked, and substations) using the best available information.
manual work is required to bring all the data to a 2) Routine validation of the generation models.
central location for analysis and to align multiple Such revalidation can readily be done through
data files. The Western Interconnection Synchro- disturbance monitoring [20] (that would record
Phasor Program will increase PMU networking internal plant variables such as field quantities
and bring real-time synchronized phasor data to on the generator).
a central location. Synchronization is also critical 3) For load models, utilization of data from power
for the sequence of event reconstruction, partic- quality meters of recorded transmission distur-
ularly for complex events where many switching bances as seen at distribution feeders to obtain
events occur in a short time frame. load model parameters and composition, where
• Tools for model analysis. It is very common for practical and possible.
an initial validation model simulation to not match 4) Routine system wide model validation.
actual disturbance recordings. The causes of mis- The key to this process is that monitoring and/or testing
match include different powerflow conditions, to validate at the component level (e.g. each large
incorrect sequence of events, un-modeled con- power plant, key substations etc.) as well as monitoring
trollers and protection elements, mis-represented to validate the entire model are both needed. Just one
operational practices, or bad model data. The or the other will not suffice. Also, a means must be
first step, therefore, is to validate as many model established to allow the sharing of information across
components as practical. Playback functions are the interconnections to allow for system wide model
currently available in the grid simulators. The validation.
function allows playback of recorded boundary Power system modeling and model validation is a
conditions (voltage and frequency) in the dynamic critical task in transmission investment planning and
simulation to validate the component model re- operational planning. Periodic system model validation
sponse. Recorder active and reactive power are is necessary to ensure that the power system models
 7

are accurate and up to date. Without model validation, [16] IEEE Task Force on Generator Model Validation Testing,
power system studies become little more than academic Guidelines for Generator Stability Model Validation Testing,
Proceedings of the IEEE PES General Meeting, Tampa, FL,
exercises. June 2007.
eha [17] L. N. Hannett and J. W. Feltes, Derivation of Generator,
January 29, 2010 Excitation System and Turbine Governor Parameters from
Tests, presented at the CIGRÉ Colloquium on Power System
Dynamic Performance, Florianpolis, Brazil, 1993.
ACKNOWLEDGMENT [18] L. M. Hajagos and G. R. Berube, Utility Experience with Gas
Turbine Testing and Modeling, Proceedings of the IEEE PES
The authors would like to thank their respective Winter Power Meeting, January 2001.
[19] P. Pourbeik and F. Modau, Model Development and Field
companies for their support of this work. The authors Testing of a Heavy-Duty Gas-Turbine Generator, IEEE Trans.
also acknowledge the efforts and activities of the West- PWRS, May, 2008.
ern Electricity Coordinating Council, Inc. (WECC) [20] P. Pourbeik, Automated Parameter Derivation for Power Plant
Models From System Disturbance Data, Proceedings of the
Modeling and Validation Working Group under the IEEE PES General Meeting, Calgary, Canada, July 2009.
WECC . [21] P. Pourbeik, A. P. Bostrom, E. John and M. Basu, Operational
Experience with SVCs for Local and Remote Disturbances,
Proceedings of the IEEE Power Systems Conference and
R EFERENCES Exposition, Atlanta, GA, October 29th November 1st, 2006.
[22] D. Kosterev, A. Meklin, J. Undrill, B. Lesieutre, W. Price,
[1] D.N.Kosterev, C.W.Taylor, W.A.Mittelstadt, Model Validation D. Chassin, R. Bravo, and S. Yang, Load modeling in power
for the August 10, 1996 WSCC System Outage, IEEE Trans- system studies: WECC progress update Proceedings of the
actions on Power Systems, vol.14, no.3, pp.967-979, August IEEE PES General Meeting, Pittsburg, July 2008.
1999 [23] A. M. Gaikwad, R. J. Bravo, D. Kosterev, S. Yang, A. Maitra,
[2] John Undrill, Les Pereira, Dmitry Kosterev, Shawn Patterson, P. Pourbeik, B. Agrawal, R. Yinger, and D. Brooks, Results of
Donald Davies, Robert Cummings, Baj Agrawal, Steve Yang, Residential Air Conditioner Testing in WECC, Proceedings of
Generator Model Validation in WECC, presented at 2009 the IEEE PES General Meeting, Pittsburg, July 2008.
Power Engineering General Meeting, Calgary, AB, July 2009.
[3] R.H.Bunch, D.N.Kosterev, Design and Implementation of AC
Voltage Dependent Current Order Limiter at Pacific HVDC
Intertie, IEEE Transactions on Power Delivery, vol.15, no.1,
pp.293-299, January 2000.
[4] Les Pereira, Dmitry Kosterev, Peter Mackin, Donald Davies,
John Undrill, Wenchun Zhu, An Interim Dynamic Induction
Motor Model for Stability Studies in WSCC, IEEE Transac-
tions on Power Systems, vol.17, no.4, pp1108-1115, 2002.
[5] U.S.-Canada Power System Outage Task Force, (2004) Fi-
nal Report of the August 14, 2003 Blackout in the United
States and Canada: Causes and Recommendations.[Online].
http://www.nerc.com
[6] Les Pereira, John Undrill, Dmitry Kosterev, Donald Davies
and Shawn Patterson, A New Thermal Governor Modeling in
WECC, IEEE Transactions on Power Systems, vol.18, no.2,
pp.819-829, May 2003.
[7] Baj Agrawal, Dmitry Kosterev, Model Validation for a Distur-
bance Event that Occurred on June 14 2004 in the Western
Interconnection, presented at 2007 Power Engineering General
Meeting, Tampa, FL, June 2008.
[8] Bradley Williams, Wayne Schmus, Douglas Dawson, Trans-
mission Voltage Recovery Delayed by Stalled Air Conditioner
Compressors, IEEE Transactions on Power Systems, vol.7,
no.3, pp.1173-1181, August 1992.
[9] John Shaffer, Air Conditioner Response to Transmission Faults,
IEEE Transactions on Power Systems, vol.12, no.2, May 1997,
pp.614-621.
[10] Garry Chinn, Modeling Stalled Induction Motors, presented
at the IEEE Transmission and Distribution Conference, Dallas
TX, May 2006.
[11] NERC, White Paper on Fault-Induced Delayed Voltage Recov-
ery
[12] P. Pourbeik and B. Agrawal, “Hybrid Model for Representing
Air-Conditioner Compressor Motor Behavior in Power System
Studies,” Proceedings of the IEEE PES General Meeting, to
be published, July 2008.
[13] Dmitry Kosterev, Anatoliy Meklin , John Undrill, Bernard
Lesieutre, William Price, David Chassin, Richard Bravo, Steve
Yang, Load Modeling in Power System Studies: WECC
Progress Update, presented at 2008 Power Engineering General
Meeting, Pittsburgh, PA, July 2008.
[14] P. Pourbeik, P. S. Kundur and C. W. Taylor, The Anatomy
of a Power Grid Blackout, IEEE Power & Energy Magazine,
September/October 2006, pp.22-29.
[15] WSCC Control Work Group and Modeling & Validation Work
Group, Test Guidelines for Synchronous Unit Dynamic Testing
and Model Validation, February 1997. (www.wecc.biz)