Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Power Plant Model Validation for Achieving

Reliability Standard Requirements Based on


Recorded On-Line Disturbance Data
Pouyan Pourbeik, Christopher Pink and Ron Bisbee

electrical generator itself). MOD-027-1 deals with turbine-


governor response. The Western Electricity Coordinating
Abstract—This paper describes a novel approach for
automated synchronous generator parameter derivation using
Council (WECC) has had mandated generator model
disturbance data. Instead of traditional off-line testing, this validation requirements since 1997 [1]. With the imminent
process supports validation of dynamic models for generators NERC standards, the volume of such model validation will
and their governor, excitation, and stabilizer systems using increase tremendously. There is, therefore, a need to
data gathered while the units are on-line. The approach is consider new ways to address model validation, which was
based on using digitally recorded unit responses to system the impetus behind the research presented in this paper.
disturbances. Continued model validation is necessary to The aim was to achieve model validation by utilizing useful
ensure generating units are simulated correctly. Model disturbance recordings from unplanned system wide events.
validation is imminently to be mandated by the North Useful events must result in significant perturbations of the
American Electric Reliability Corporation (NERC) through
its modeling, data, and analysis standards. The benefits of the
generating unit under study, without causing the unit to trip
model validation method presented here are explained when or go unstable. Such disturbances include, for example,
compared to traditional techniques, which typically include remote transmission faults, loss of generation elsewhere in
collecting data from a series of mostly off-line staged tests. the interconnected system, sudden changes in system
The comprehensive process of model validation and a review voltage (e.g. switching of large shunt devices or loss of
of the key-limitations of this process are also discussed. load etc.) and other similar type events.
Lastly, the paper presents the successful application of this
technique to three large thermal units for the purpose of There are well established methods to perform synchronous
WECC generator model validation and certification. Key generator power plant model validation using staged tests.
results of the case studies are reported together with fruitful Generally, these methods mainly involve off-line tests or at
insight into the limitations of some of the models available for least some form of planned maneuvering of the unit, during
power plant modeling in planning studies. which the unit is removed from normal operation [2 – 8].
Index Terms—Generator Model Validation, Power Plant Such methods involve giving access to the power plant to
Modeling, Power Plant Model Validation testing staff, allowing them to connect their testing
equipment/monitoring devices to the generator controls,
I. INTRODUCTION and then maneuvering the unit through a series of off-line
The practice of using power system simulation models for and on-line tests to extract the necessary data. This process
performing system wide planning and operational studies is prudent initially or upon commissioning of new
has been well established since the 1970’s. System planners equipment in order to establish a baseline model. However,
and operators rely on power system models of varying for revalidation of models this process clearly takes time
complexity for analyzing the power system. While there and expense to achieve. The costs to larger thermal power
have been large amounts of new generation technologies plants can be particularly significant when the opportunity
integrated into the power system, generating facilities loss for sale of power is included. Furthermore, many
utilizing conventional synchronous generators still eastern-US utilities are still concerned with the risk of
dominate most regions. damage to the unit during such tests, particularly for large
thermal and nuclear power plants. It should be noted that
Forthcoming NERC Modeling, Data, and Analysis (MOD) this type of staged testing work has been ongoing in the
standards MOD-026-1 and MOD-027-1 will require routine WECC for over a decade, with hundreds of units tested,
validation of models used in planning studies and to the knowledge of the authors, there has been no
(www.nerc.com). The first standard, MOD-026-1, deals reported damage to units as a consequence of tests – thus,
with the generator excitation system (and implicitly the the risk of damage is very low when tests are conducted by
properly trained staff.
Corresponding author: Pouyan Pourbeik
EPRI, Knoxville TN, USA The aim of this research was to identify a practical
Ph: (919) 794 7204 methodology to validate the models associated with power
Email: ppourbeik@epri.com or plants, while reducing some of the cost and risks associated
pouyan@ieee.org
with repeated staged testing. One important caveat is that a

978-1-61284-788-7/11/$26.00 ©2011 IEEE


baseline set of tests (whether during commissioning, or
scheduled maintenance outages) in most cases will still be
required. However, the methodology and work presented
enables the routine revalidation of the models, without
service interruptions, in order to meet the imminent NERC
(and existing WECC) reliability standards. Thus, this
reduces the burden of cost and some of the potential risks
associated with routine retesting through staged tests.
Another tangible benefit of model validation using
disturbance monitoring is that the models are validated
against “actual” system events. In the end this is exactly
what the models are used for. That is, system planners use
these models to attempt to predict plant response to
potential system events in order to plan the system for
possible future scenarios. Thus, through this method the Figure 1: The model for parameter fitting.
models are being validated against real life events rather
than being tested under light load conditions and “staged”
tests that do not necessarily reflect actual events.
Section II provides an outline of the methodology. Section
III details the type of data collection required. Section IV
presents results from a few case studies, and Section V
concludes with a summary and recommended further work.
A stand alone executable tool was developed (using
MATLAB® and the MATLAB® Compiler), with a graphic
user interface that can be easily used to achieve model
validation and parameter estimation from the recorded
disturbance data. The work presented here is follow-on
work from a previous research project [9, 15].
II. THE ON-LINE DISTURBANCE MONITORING BASED
METHODOLOGY
The general concept is as follows:
1. The terminal voltage, current, real and reactive
power, speed, and field quantities (field voltage Figure 2: Connection diagram for recording equipment.
and current) of a synchronous generator are
recorded by a digital fault recorder (DFR) during a
system event.
2. Following the event the data is extracted from the
recording device.
3. Using an automated iterative simulation tool, the
power plant is appropriately modeled and the
generator, governor, and excitation system model
parameters are optimized to match the recorded
response.
Figure 1 shows the block diagram structure of the models
used. Figure 2 shows the typical connection diagram for
the recording device. Figure 3 depicts the flow of the
process.
A. Modeling
The modeling of the plant is done using standard methods
for modeling synchronous generators, excitation systems, Figure 3: Flow-chart of the parameter estimation process.
and turbine-governors [10 – 14]. The differential and B. The Validation and Optimization Process
algebraic equations defining each device were developed
The model validation and optimization is done in three
and coded in MATLAB®.
steps:
Step 1 – The “base-line data” for the unit is collected. This The DFRs are typically configured to trigger on voltage
includes the known equipment control structure, equipment and/or frequency deviations. The triggers used for the case
ratings, and the originally provided manufacturer model study were:
structures and parameters.
• Frequency: - 80 mHz, or + 60 mHz .
Step 2 – If necessary, convert the captured data into ASSCI
or COMTRADE format and per unitize the data into RMS, • Vrms: ± 5% over one cycle.
positive sequence quantities. Once an event is captured, the data must be extracted in
Step 3 – Port the data into and run the software tool. The COMTRADE or ASCII format and then per unitized.
models are then validated through an optimization routine. The DFRs were set up to record all 3-phase stator voltages,
The details of the actual optimization method and algorithm all 3-phase stator currents and the field voltage and current.
are described in [15]. Here for completeness a quick Internally the DFR calculated 3-phase unit megawatts and
summary is provided. megavars and also terminal voltage frequency. This was all
calibrated and checked. This data is then captured at two
The power plant can be divided into four subsystems for sampling rates when an event occurs (i) a fast kilo-hertz
validation separately. As shown in Figure 1, the rate (which is not used for the analysis here) and (ii) in
subsystems include: the generator (block 1), the excitation RMS format at a sampling rate of 60 samples per second
system (block 2), the unit shaft dynamics, for verifying for many tens of seconds. The DFR can also calculate the
inertia (block 3) and the turbine-governor (block 4). positive and negative sequence components of voltage for
The goal for model validation is to confirm that the models unbalance events. The DFRs are capable of storing up to
for a power plant adequately represent the actual plant 1000 such events and the events can be extracted remotely
dynamics, and if not, what parameters appear to be through the internet. Figure 4 shows a picture of one of
inappropriate and to optimize those parameters so that it these DFRs.
does. It is not feasible to attempt to optimize all the plant It should be noted that the sampling rate of the recording
model parameters simultaneously. Also, if a parameter device should be adequate to preventing aliasing,
appears to be significantly out of range, further particularly for the field quantities.
investigation at the plant may be warranted.
For example, to validate the generator subsystem model,
the disturbance recordings of field voltage, stator current
(Id, Iq) and speed are fed into the model. The resulting
simulated unit field current and stator voltage are then
generated by the software tool and are compared to the
measured quantities. If the fit is acceptable, the process
ends and the model can be considered validated. If the fit
is not within the set tolerances, it can still be optimized
within user defined bounds through a least-square
algorithm. Prudent judgment is used to decide which
parameters to optimization around. Typically, only those
parameters that have the greatest uncertainty are optimized
(e.g. field time constant which can vary as a function of DFR
rotor temperature). The optimization tool can then be used
to derive values for the selected parameters so that model
performance closely matches the DFR measured
performance.
For more details on the algorithm and optimization method, Figure 4: One of the DFRs in the plant.
please refer to [15].
IV. CASE STUDY – THERMAL POWER PLANT
The tool is called the Power Plant Parameter Derivation
tool [16]. This work was done for three large steam-turbine
generating units rated at 496 MVA each.
III. THE DATA COLLECTION
Baseline data was available for all three units based on
Data is collected at the generator terminals and on the detailed original equipment manufacturer datasheets and
generator field (for a brushless unit one would collect data data from a recent commissioning test several years ago for
from the exciter field) using a dedicated Digital Fault one of the units where the excitation system was retrofitted.
Recorder (DFR) installed in the plant. For the case study Thus, two of the units are nominally identical, while the
discussed in the next section two dedicated DFRs exist in third unit has a modern digital excitation system that was
the plant, one collects data for two of the generator units installed in 2005.
and the third unit has its own dedicated DFR.
Data was capture over a period of seventeen (17) months – the bounds on the others appropriately. For example, if one
from May 2008 to September 2009). During this period a knows (such as in this case) the gain setting of an exciter
total of seven events of adequate magnitude were captured (e.g. from the digital controls etc.) then it should be used as
for model validation. This is shown in Table 1 below. the initial estimate and fixed or tightly bound around that
value. The key here is the automated algorithm simplifies
Event data through 2008 was also available in the historic
the parameter fitting exercise and saves considerable
database, however, prior to the commencement of this work
engineering time.
the DFRs were never used for model validation. Thus,
field voltage and field current were not previously It is not possible to show all of the results here. Model
monitored by the DFRs. With the commencement of this validation was conducted for all three units over the
work, those additional quantities were added to the multiple events captured. Thus, we were able to show that
recording channels of the DFRs. The field voltage and the same models and model parameters for the generator,
current measurements were added to unit 1 & 2 in exciter and PSS could be used to simulate the various
November 2008 and to unit 3 in August 2009. system events captured. Here only a few samples of the
Table 1: Event recording description.
results are presented.

Event Event Date Event Description Figure 5 below shows a case were the inertia of the unit has
Number been reconfirmed through the observed speed transient of
1 9/20/2009 System wide event; loss of the unit to nearby faults.
generation in WECC,
followed by a nearby fault
2 8/13/2009 Nearby transmission fault
3 6/10/09 System wide event; loss of
generation in WECC
4 11/7/2008 System wide event; loss of
generation in WECC
5 9/16/2008 System wide event; loss of
generation in WECC
6 6/11/2008 System wide event; loss of
generation in WECC
7 5/20/2008 System wide event; loss of
generation in WECC

All three units are round rotor 3600 rpm generators. Two
of the units have static excitation system that can be
Figure 5: Comparison of measured and simulate speed response of a
represented by the IEEE ST3A model, while the third unit generating unit to nearby faults.
was retrofitted with a modern digital static excitation
system that can be represented by the IEEE ST4B model. As shown in Figure 6, the fitted and measured responses
The units fitted with the old static exciters both have match quite well for the electrical response of the units. In
frequency input power system stabilizers (PSS), while the all cases the fitted response was achieve with parameter
third unit is fitted with a modern dual-input PSS2A. very close to the OEM data – thus we kept the OEM data as
the validated models and model parameters.
The PSS parameters were fixed (i.e. not allowed to vary in
the optimization process) since these are known tuned field Figure 7 and 8 show similarly good fits achieved for
settings1. The generator electrical parameters were also unbalanced fault events – in these cases the fit was done
fixed, since good manufacturer data was available and there against the positive sequence component of unit terminal
had been no material change made to the unit since the last voltage.
field test. A few pertinent remarks about the curve fitting are:
Parameter optimization and validation was performed on 1. For the unbalanced fault events we could still achieve a
the excitation system parameters for three events. One item reasonable fit by extracting the positive sequence
needs to be emphasized. An optimization process will find component of stator voltage and fitting to it.
multiple solutions to the problem. The key to identifying
the “right” solution – that is a meaningful solution that 2. In all cases, there was a steady-state error between
properly represents the physical equipment – is to know measured and simulated field current. This was true for all
what the correct model structure is (e.g. static excitation three units. This discrepancy could be due to residual flux,
system versus brushless, etc.), what the physically measurement error (e.g. biases on these signals in
meaningful range of each parameter is, fixing (or tightly transducers measuring field current/voltage or in the DFR
bounding) those parameters one already knows and setting device) or errors in the estimated unit saturation curve (i.e.
Ifdbase or S10/S12). The relative error was in the order of
1
A PSS should be tuned with well known and established parameters 4% and thus well within the realm of measurement errors.
for the purpose of providing damping.
a) Unit 1 Figure 7: Comparison of measured and simulated response for an
unbalanced fault (Event 1).

b) Unit 2
Figure 8: Comparison of measured and simulated response for an
unbalanced fault (Event 2).

The units all draw their mechanical power from triple


pressure steam-turbines. There was no prior model
provided for the turbines. Based research of the OEM data,
the power fractions for the turbines were identified and so
we could establish an initial reasonable model using the
IEEEG1 model an outer loop load-controller (see Figure
11). Based on events 5, 6 and 7 reasonable fits were
achieved – Figure 9 shows some examples.
These units are typically either base-loaded or run on outer-
loop MW control, as can be seen from the various event
recordings.
c) Unit 3
There is some significant difference in a few of the
parameters between events – namely the droop and load
controller gain. The reason for this is the simplicity of our
models as compared to the complexity of the turbine
controls; e.g. the initial MW is quite different between
cases, which would easily lead to different initial steam
pressure/temperature and our models do not represent this.
Thus an average model was recommended.

d) System frequency
Figure 6: Comparison of measured and simulated response for a WECC
system wide (Event 1).
inception of the event is a 0.4 Hz oscillation showing
clearly the participation of this unit in the WECC inter-area
electromechanical mode of rotor oscillation system wide.

a) Event 5

a) Unit 2, Event 6: MW response

b) Event 6

b) Event 6, Frequency
Figure 10: The MW response of Unit 2 during Event 6.

V. CONCLUSIONS, DISCUSSION AND FURTHER WORK


A new and novel automated methodology for fitting
parameters to, and thus validating the models of, power
plant equipment based on recorded on-line system
disturbances has been presented. The method has been
successfully applied to three steam-turbine generators for
the purpose of WECC generator model validation and
c) Event 7
certification.
There are a few key conclusions and insights from this
work:
1. Power plant model validation based on
disturbance recording, while the unit is on-line, is
feasible and valuable.
2. The power plant models can be validated by
recording the response of the unit at its terminals.
With the approach presented here, there is no need
for knowing either what the disturbance was or the
condition of the power system. The recorder
d) Frequency for Events 5, 6 & 7. terminal response of the generating unit is
sufficient data for validating the models, since
Figure 9: Turbine-governor response for unit 3 for multiple events.
some of the recorder variables (i.e. real and
Figure 10 shows clearly the MW capability of these units. reactive power) as uses as inputs to our simulation
As shown, when the unit is at 456 MW, even for a major tool and the others (i.e. voltage and field
frequency event there is no more MW capability left in the quantities) are use as outputs for comparison and
unit to increase its output. The interesting observation in confirmation of model validity.
this case is that the perturbations in electrical power at the
3. In previous work [15], it was assumed that the 5. For large thermal power plants, there is
recorded disturbance needs to be a balanced event considerable lost-opportunity cost for the duration
to facilitate a convincing exercise of model of the staged tests for model re-validation (i.e. the
validation. However, by extracting the positive tests take four to six hours typically, that is time
sequence component of terminal voltage it has the units could have been selling power but did
been demonstrated here that unbalanced events not). Also, some utilities believe that there may
(such as an unbalanced fault) can also be used for be some (thought very low) risk of damage to the
model validation. unit due to undue stress during tests such as load
rejection tests. Such costs and risks are totally
4. It is important for this type of model validation to
avoided by using on-line monitoring.
have dedicated recording of the generator terminal
voltage, current and field voltage and field current, Finally, with the imminent NERC requirements for
at a minimum. Speed can be reasonably generator model validation through the modeling, data and
calculated from the frequency of the voltage analysis (MOD) standards, a pragmatic approach for testing
waveform, and real and reactive power can be and validation in North America (on a routine basis) will
calculated from the voltage and current be:
measurements. These calculations can be done
1. To test a unit thoroughly once (on commissioning
internally with most modern DFRs. Sampling
for new units) using staged tests or a similar
intervals of 20 ms or lower are adequate (these
approach.
higher sampling rates are particularly important
for validating the excitation system response). At 2. The above step will allow one to set the models,
least twenty to thirty seconds of data should be model structures and parameters based on actual
captured for each event, especially for the purpose observed equipment settings and equipment
of validating the turbine-governor response. For performance. This will form the “baseline model”.
validating the turbine-governor response,
3. Now moving forward collect disturbance data (e.g.
sampling as slow as 1 sample per second may be
adequate. using digital fault recorders as illustrated here) on
an on-going basis.
Perhaps the most important benefits of this approach are:
4. Use the DFR collected data, and an automated tool
1. The ability to validate models against actual such as the one described here to re-validate the
events that are often looked at in planning studies, models periodically. If one sees a huge divergence
rather than against staged tests, which do not between the model and field recorded response,
necessarily represent the actual behavior of control flag and pursue further investigation of the plant –
systems when the unit is on-line (e.g. the influence there may have been a retrofit or retuning of
of load or megawatt controllers on the turbine- controls, which may require more detailed testing
governor can only truly be observed for a system (step 1) or simply to identify the changes and work
disturbance). those changes into the model structures and
established thought the disturbance based method
2. There is effort and cost involved in collecting the
that the new models are valid.
initial baseline data and setting up the DFRs, to
collect the data event data. However, this is a one- For nuclear units, and other large thermal units, there is, in
time process for a given unit. Thus, the process of some regions of North America more than others, still
continual revalidation becomes markedly less concern with respect to field testing the units for model
costly and time consuming as compared to staged validation. For these units in particular, the approach
testing in which case routine re-testing requires presented here offers a welcome approach for routine
unit down-time, connecting monitoring equipment model revalidation. However, the two key factors
and hiring outside consultants. emphasized throughout this paper should not be forgotten:
(i) that the generating unit must have an adequate DFR (or
3. The DFRs also provide other benefits for
similar) device in place to capture the necessary data
monitoring and diagnostic efforts for the
during system disturbances, and (ii) that adequate baseline
generating unit.
data must be available (e.g. detailed manufacturer supplied
4. For the Western US power system, to which the models and data sheets that represent the actual field
units studied here are connected, we observed equipment or such data backed by field tests during
during the course of this work an event frequency commissioning). If adequate recoding is not available at
of roughly one system event per 2.5 months. the plant, a DFR can perhaps be installed during a schedule
Thus, re-validation is practical on a relatively maintenance outage.
frequent basis2.
One final comment is that the method proposed here may
not be appropriate for peaking units or units that are in-
2
At the time of submitting this paper for publication several recent
system events were captured and used to reconfirm the plant models. The excitation system models through the software tool was roughly 30
time taken from downloading the data from the DFR to validating the minutes.
service for limited time during the year thereby reducing [9] P. Pourbeik, “Automated Parameter Derivation for Power Plant
Models Based on Staged Tests”, Proceedings of the IEEE PSCE,
the probability of capturing system events. In this case, Seattle, WA, March, 2009.
however, on could for example inject a small voltage [10] P. Kundur, Power System Stability and Control, McGraw-Hill, 1994.
reference step into the unit’s voltage regulator while the [11] IEEE 421.5-2005, Recommended Practice for Excitation System
unit is on-line and capture such an “emulated event” for Models for Power System Stability Studies, April 2006.
[12] L. Pereira, J. Undrill, D. Kosterev, D. Davies, and S. Patterson, “A
model validation. New Thermal Governor Modeling Approach in the WECC”, IEEE
VI. ACKNOWLEDGEMENTS Trans. PWRS, May 2003, pp 819-829.
[13] L. Pereira, D. Kosterev, D. Davies and S. Patterson, “New thermal
The primary author wishes to thank Tri-State Generation governor model selection and validation in the WECC”, IEEE
Transactions on Power Systems, Volume 19, Issue 1, Feb. 2004
and Transmission Associate Inc. co-sponsored this research Page(s):517 – 523.
work and volunteered their power plant for the purpose of [14] CIGRE Technical Brochure 238, Modeling of Gas Turbines and
this research project. Steam Turbines in Combined-Cycle Power Plants, Prepared by
CIGRE TF C4.02.25, December 2003.
We are grateful to Mr. Gary Preslan who played a key role [15] P. Pourbeik, “Automated Parameter Derivation for Power Plant
in the initial stages of extracting data and helping to set-up Models From System Disturbance Data”, Proceedings of the IEEE
the triggers for the in-plant digital fault recorders; and to PES General Meeting, Calgary, Canada, July 2009.
[16] Power Plant Parameter Derivation (PPPD) Software User’s Manual:
Mr. Mark Graham and Mr. Sam Reed for the many fruitful Version 2.0, Software Product ID 1017803, Software Manual, May
discussion and support during this work. 2009 (wwww.epri.com).

We are also grateful to all utility members of the Grid, Pouyan Pourbeik (M’1993, SM’2002, F’2010) received his BE and PhD
Operations and Planning Program 40 and the Steam in Electrical Engineering from the University of Adelaide, Australia in
1993 and 1997, respectively. From 1997 to 2000 he was with GE Power
Turbine, Generators, and Balance-of-Plant Program 65, Systems. From 2000 to 2006 he was with ABB Inc. In June 2006 he
who funded the development of the Power Plant Parameter joined EPRI Solutions Inc. In 2007 EPRI Solutions Inc. became part of
Derivation tool developed by EPRI and used in this work. EPRI. Throughout his career he has been involved in and led studies
related to many aspects of power systems modeling, dynamics and control.
VII. REFERENCES: He has also conducted field testing on a total of more than sixty turbine-
generator units. He is presently Chairman of the IEEE PES Power System
[1] WSCC Control Work Group and Modeling & Validation Work Stability Subcommittee and Secretary of CIGRE Study Committee C4 –
Group, “Test Guidelines for Synchronous Unit Dynamic Testing and System Technical Performance. He has authored/co-authored over fifty
Model Validation”, February 1997. (www.wecc.biz) technical publications on power systems modeling and analysis. He is a
[2] IEEE Task Force on Generator Model Validation Testing, registered professional engineer in the state of North Carolina, USA.
“Guidelines for Generator Stability Model Validation Testing”,
Proceedings of the IEEE PES General Meeting, Tampa, FL, June Christopher Pink (S’02–M’03) received the B.S. and M.S. degrees in
2007. electrical engineering from the Colorado School of Mines, Golden, CO.
[3] L. N. Hannett and J. W. Feltes, “Derivation of Generator, Excitation He is currently a Senior Planning Engineer at Tri-State Generation and
System and Turbine Governor Parameters from Tests,” presented at Transmission Association where he is involved with dynamic modeling
the CIGRÉ Colloquium on Power System Dynamic Performance, and validation of generator facilities. Previous occupations have included
Florianópolis, Brazil, 1993. Research Engineer at the National Renewable Energy Laboratory, Golden,
[4] L. M. Hajagos and G. R. Berube, “Utility Experience with Gas CO, where he was involved in research and modeling of advanced systems
Turbine Testing and Modeling”, Proceedings of the IEEE PES to interconnect both renewable and conventional distributed generation
Winter Power Meeting, January 2001. systems to the utility grid. He has also been engaged in designing,
[5] J. Undrill, “Testing of Generating Unit Dynamic Behavior – Notes installing, and commissioning protective relay, controls, and distribution
on Proposed NERC Testing Requirements”, Presented at SERC systems.
Generator Testing Workshop, Atlanta, GA, December 2000.
[6] J. Undrill and A. Murdoch, “Power Plant Dynamic Performance Ron Bisbee (S’00–M’01) received the B.S. degree in electrical
Issues in Relation to Grid Interconnection Codes”, CIGRE Session engineering from the South Dakota School of Mines and Technology,
2002, paper 39-206. Rapid City, SD. He is employed by Tri-State Generation and
[7] P. Pourbeik and F. Modau, “Model Development and Field Testing Transmission Association, Westminster, CO. He is currently a Senior
of a Heavy-Duty Gas-Turbine Generator”, IEEE Trans. PWRS, May, Generation Engineer and is responsible for supporting maintenance,
2008. reliability improvements and testing for Tri-State’s generation facilities.
[8] P. L. Dandeno, H. C. Karmaker, C. Azuaje, M. Glinkowski, I. Prior to this position he was a Senior Engineer Power Plant also for Tri-
Kamwa, S. J. Salon, R. M. Saunders and S. Umans, “Experience State G&T where he was involved in the operation and maintenance of
with Standstill Frequency Response (SSFR) Testing and Analysis of Tri-State’s Craig Station power plant. He is a member of IEEE PES.
Salient Pole Synchronous Machines”, IEEE Trans. on Energy
Conversion, vol. 14, no. 4, pp. 1209-1217, December 1999.
speed

+ 1.0
Irmax
Prefo
Kp fb
emax
+
+
+ Irmax +
Pmwset
db
_
Ki
s 1
1 + sTpelec
-Irmax

-Irmax Pe

+ + +
Pm
+ + +

K1 K3 K5 K7
Uo
Pmax
+
dbd1 _
+ K (1 + sT2) 1 1 1 1 1 1
speed
1 + sT1 T3 s 1 + sT4 1 + sT5 1 + sT6 1 + sT7
_ _

Pmin
1.0 Uc

Figure 11: Steam-Turbine model. This is a model developed by combining the standard IEEEG1 steam-turbine model with an outer-loop MW-controller (the
lcfb1 model in GE PLSF®)

You might also like