II. Load Modeling - Western Electricity Coordinating Council

advertisement
Western Electricity Coordinating Council
Modeling and Validation Work Group
Progress Report to TSS
August 2014
The WECC Modeling and Validation Work Group (M&VWG) meeting was held in Salt
Lake City, UT on June 17-19, 2014. Renewable Energy Modeling Task Force (REMTF)
hosted the renewable energy workshop on June 17, and M&VWG was held on June 1819.
The following topics were discussed at the M&VWG meeting:
I. Next Meetings and Workshops
II. Load Modeling
III. Renewable Energy Modeling
IV. System Model Validation
V. Generator Modeling, Testing, and Model Validation
VI. HVDC Modeling
VII. RAS and Relay Modeling
VIII. NERC Updates
IX. WECC Project Updates
X. Approved Dynamic Model Library and Status Update of Models
XI. Program Updates: GE PSLF, PSS®E, and PowerWorld
Meeting materials and presentations are posted on the WECC MVWG Meeting Website
for the M&VWG and REMTF meetings.
Page 1 of 13
I. NEXT MEETINGS AND WORKSHOPS
The next MVWG meeting is planned to be at Albuquerque NM, hosted by Sandia
National Labs, on November 18-20, 2014. The REMTF will be held November 18th
afternoon and 19th morning, the Load Modeling Task Force (LMTF) will be held
November 19th morning, and M&VWG will be held November 19th afternoon and 20th all
day.
A webinar of the PI Data Frequency Response Tool Demo is planned to be 9-11 am PST
on July 22nd.
In addition, a workshop on how to use PSLF v19 to create a Model Validation Case from
WSM data will be held after GE officially releases v19 in about two months. More details
to follow.
II. LOAD MODELING
A) Composite Load Model Phase 1 Update
Dmitry Kosterev (BPA) informed the team that he has seen more PSSE users starting to
use the composite load model phase 1, for example, American Electric Power, Virginia
Power, and Southern Company etc. He indicated that the TF will continue the outreach
to PSSE user community outside of WECC, as the TF has done in the WECC.
B) Composite Load Model Phase 2 Update
Dmitry Kosterev (BPA) indicated that the phase 2 results may be pessimistic. He also
informed the team of John Undrill’s work regarding to the sensitivities to the point on
wave and good analysis there.
SCE has installed PQube data on the distribution circuit and shared the FIDVR data. BPA
also installed microPMU on the distribution circuit. Dmitry encourages utilities to
volunteer and do the same practice so that we can have more data for FIDVR analysis.
The PQube monitoring devices are free from Lawrence Berkeley National Lab, the
participant just needs to deploy the devices and agree to share collected data.
Dmitry also mentioned that there is a renewed interest in the load composition of the
residential and commercial buildings. DOE will have the FIDVR website, where the data
will be posted and made available for entities to perform the analysis.
Page 2 of 13
C) Composite Load Model – Model Specifications and Test Case
Stephanie Lu (SCL) will distribute the “WECC Composite Load Model Specifications”
drafted by Bill Price (GE) for group review. This includes the new additions including the
distributed generation component. Jay Senthil (Siemens) also suggested having test
cases for the composite load model so that all the software vendors are consistent in
model implementation. Dmitry Kosterev (BPA) will send the test case to the vendors for
benchmarking across platforms to ensure consistent implementation of the composite
load model.
D) Composite Load Model Monitoring Improvements in Software Implementation
When performing the study with the composite load model in PSLF, it is hard to track
how much load was lost per zone/area/owner with the existing LSMON model, so
multiple users have expressed the desire to have the load trip monitoring capability in
the software. Jun Wen (SCE) agreed to draft the user desired specifications, circulate
among the TF for comments, and send to GE for implementation.
There are two other suggestions from the PSLF users to GE:
1) Dmitry Kosterev (BPA) brought up the issue that the user has to update the swt file
that is used for dropping load when load is modeled with motorw to open the
distribution transformer equivalent when load is modeled with cmpldw. This is due to
the fact that the load is moved to the distribution bus equivalent with cmpldw and so
the swt that was developed for motorw is no longer able to find the load in the case
that is using cmpldw.
2) Jun Wen (SCE) brought up the issue that the bus that is treated as the load bus when
using motorw will be treated as the non-load bus when use cmpldw due to the fact
that the load is now moved down to the lower distribution bus. Since we may want to
still treat the bus where the load is modeled in the power flow as the load bus and
there are different transient voltage and frequency criteria for the load bus and the
non-load bus, it is better to have ways to differentiate those buses with different
variable names for WCA analysis.
Jun Wen (SCE) will write up the user expectation and work with GE for potential
modifications.
E) WECC-0100 Project Update and Transient Voltage Criteria
Chuck Matthews (BPA), chair of the WECC-0100 project, provided an update on the
WECC-0100 project. The project team is continuing to discuss various alternatives for
the transient voltage performance, voltage stability criteria, steady state voltage limits
and post contingency voltage deviations. Chuck indicated that the project may need
input from M&VWG on the transient voltage performance criteria. There are also some
discussions around whether the transient criteria should be a regional business practice
or a guideline. The majority of the project members would prefer to have this as a
Page 3 of 13
guideline. If there is any additional comment from M&VWG, Stephanie Lu (SCL) will
provide the feedback to Chuck.
III. RENEWABLE ENERGY MODELING
Pouyan Pourbeik (EPRI) provided an update of the Renewable Energy Workshop that
was held on June 17, 2014. There were about 30 people in person and 10 on the
webinar. The workshop covered standards, modeling guideline, dynamic model
specification, user experience, and model validation tool. The recording is posted on
MVWG > Documents > Seminars > Renewable Energy Modeling Workshop 2014-06-17.
A) Transition Plan for Removing Wind Phase 1 Generic Models
Eric Bakie (Idaho Power) was tasked to come up with the transition plan for removing
Wind Phase 1 Generic Models from SRWG when the 2nd generation wind models and PV
models were approved. This is also to support the last item on the MVWG
implementation plan of these models. Eric presented a proposed transition plan. The
proposal is to start the transition from the next base case request, which is July 18,
2014, and remove the 1st generation wind models on June 5, 2015. To follow up,
PPMDTF will then review the data again with a targeted completion date in December
2015.
In preparing the proposed time line, Eric reviewed the current data in the MDF and
conducted a statistics report for the dynamic model records that have been used for
wind and solar. He pointed out the issues of missing models and the use of wrong
models for renewable modeling.
REMTF recommended removing wt1p and wt2p from the WECC Approved Model List
immediately. Pouyan Pourbeik (EPRI) indicated that no pitch control should be used for
the passive stall units, and wt1p_b should be used for active stall units. REMTF will take
the action to inspect the list and give the recommendation of either deleting or
replacing the wt1p and wt2p with wt1p_b. The model wt1p_b should have also been
added to the WECC Approved Dynamic Model List, and will be updated to reflect this.
B) Energy Storage Modeling
Pouyan Pourbeik (EPRI) indicated that the Task Force will have the proposed structure
by the next meeting. Jun Wen (SCE) commented that the three PTOs in California
already received energy storage project interconnection requests and would like to see
the modeling being expedited and asked if REMTF can provide an interim modeling
guide. Jun also mentioned that most of the energy storage projects are battery, and
there are some hybrids, either combination of PV and battery, or gas turbine and
Page 4 of 13
battery. Pouyan asked to have manufacturer representatives engaged early in the
model development which can help expedite this effort.
C) Plant Level Control
Pouyan Pourbeik (EPRI) indicated that the Task Force will provide a proposal at the next
meeting. The goal is to expand the plant controller to control multiple elements,
including different types of wind turbines, SVCs or STATCOMs, and MSCs.
He also indicated a potential new item. ABB has a configuration of using VSC and TSC as
a hybrid SVC. He will investigate whether SVSMO3 can model this hybrid SVC and
update the team in the next meeting.
IV. SYSTEM MODEL VALIDATION
A) System Model Validation Efforts
Dmitry Kosterev (BPA) showed a list of large disturbance events in 2013 and 2014.
Dmitry presented two ways of performing system model validation studies:
1) Use Planning case and planning dyd
Page 5 of 13
2) Use WSM case and WSM dyd
Based on Slaven Kincic (Peak RC), 95% of generation MW in WECC is mapped between
the planning case and the WSM case, where the majority of the discrepancies are from
the mapping of the wind generation.
Dmitry Kosterev (BPA) presented the study results on the model validation of the May
16, 2014 event. The frequency and voltage have good matches, while power pickup has
some offset. Overall, the simulation had a good match to the actual event. Further
investigation is suggested for Kemano trip and Springerville trip. Dmitry will complete
the model validation report for the May 16, 2014 event and send to the group for
review. The report is expected to be getting M&VWG approval prior to the next
meeting.
Dmitry mentioned that the model validation study for the May 26, 2014 event is in
progress.
Dmitry indicated that at the time of the May 16th event, a lot of wind generation is on,
which is where the modeling/mapping lacks most. There are some missing wind models
in SCE’s TRTP area, so are some other areas in the WECC. The issue of the missing and
bad dynamic models for wind/solar in the MDF was extensively discussed. The PPMDTF
Chair Kent Bolton (WECC) will take the action to create a list of wind/solar generators
that have either missing models or bad data, based on the information from: 1) Dmitry’s
findings from the model validation studies; 2) The list that identifies model and turbine
type discrepancies found in the MDF done by Eric Bakie (Idaho Power); and 3) The
TEPPC spreadsheet that has information on the type of generation. PPDMTF will then
Page 6 of 13
identify the owner of the data and request TP for the generator information including
manufacturer, type, and vintage. In the short-term, REMTF can help to provide default
data for those generators based on the collected information, for the system model
validation studies, which will be better than load netting those generation. For the long
run, the generator list will be provided to SRWG to request data submitters to correct or
submit the required data in the MDF.
B) WBRTF Update
Slaven Kincic (Peak RC) provided update on WBRTF efforts. 95% of generation MW has
been mapped in the WECC. He indicated that he is still waiting for some responses, and
hope the work to be completed by end of the year. He will also contact Doug Tucker
(WECC) to decide what to do with the retired units.
C) Node-Breaker Model Development
Brian Thomas (GE) presented an update on the efforts to provide the node-breaker
capabilities in PSLF. The PSLF v19 will have the capability to take WSM case and WSM
dynamic file which is mapped from the planning dynamic file to run the validation study.
The base case preparation time has been reduced from two weeks to one day. The beta
version is available for testing for the interested working group members. The PSLF v19
is estimated to be officially released this July or August. A workshop to have hands-on
training on the program is being discussed to be held after the release of the program.
D) Demo of Frequency Response Tool using PI Data
Eric Bakie (Idaho Power) provided a presentation of the frequency response tool that he
has enhanced based on the tool that was developed by BPA. The tool included batch
processing capability for model validation screening of multiple units and the ability to
easily perform model validation screening for multiple events.
Since the March meeting, he has completed the user guide. The tool along with the
validation package will be posted in the WECC MVWG > Documents > Tools folder. A
webinar is planned on July 22nd for Eric to provide hands on training on how to use this
tool.
V. GENERATOR MODELING, TESTING, AND MODEL VALIDATION
A) Power Plant Model Data Task Force
Kent Bolton (WECC) reviewed the dynamics error list. There hasn’t been much progress
since last June, and M&VWG discussed how to move things along in this group. Kent will
take the action to compile the data errors that have been identified by Irina Green
Page 7 of 13
(CAISO), Slaven Kincic (Peak RC), and Ran Xu (BPA) in their studies, as well as the
wind/solar modeling issues. The PPMDTF will work with SRWG to enhance how to
display the dynamic data errors. Kent will set up a meeting soon to kick-off the efforts
and another meeting prior to the next M&VWG meeting to check the progress.
B) Excitation System Models
Eric Bakie (Idaho Power) shared his observation on the impacts of not modeling UEL
where the generation exceeds Qmin and goes out of step. His simulations used a userwritten model and is requesting for the UEL models to be implemented in the programs.
Shawn Patterson (USBR) mentioned this has also been identified as needed for FIDVR
studies, and he requested to have UEL1 and UEL2 implementation at higher priority. Jay
Senthil (Siemens) will send to Juan Sanchez (GE) what was implemented in PSS@E which
was based on the 2005 version of IEEE 421.5 so that the implementation across the
software platform will be the same. The UEL1 and UEL2 models specified in the 2005
version will not be changed with the new revision of the standard. The IEEE 421.5 is
close to going out for ballot.
C) Cross Current Compensation Model
John Undrill provided an update to the ccomp model, which is for a pair of units that
have cross current compensation. The model was checked against the Dalles and John
Day (ABB equipment). Pouyan Pourbeik (EPRI) asked if this will work with all equipment
vendors. Pouyan will check this against a Duke case. The ccomp model is now available
in PSLF dll. It is expected that Dmitry Kosterev (BPA) and Shawn Patterson (USBR) will
bring the ccomp model for approval in the next meeting.
D) Hydro Turbine and Governor Models
John Undrill provided an update to the h6b model. The model has been previously
approved. John identified that there is a coding error in PSLF that PSLF will fix.
PowerWorld already has John’s fix and has the model available. John will also provide
the code to Jay with the errors fixed.
E) Update to Typical Machine Data Document
Stephanie Lu (SCL) updated the Typical Machine Data document. The changes include
the following:
 Replacement of the 1st generation wind models with the 2nd generation wind
models, added the PV models, and added references to the modeling guideline
documents in the Wind and PV sections.
Page 8 of 13

Minor improvements to the hydro model section, which includes adding the
accel parameter in gentpj model, and changing the ks2 parameter in pss2a
model to be Tw2/(2H), same as in the thermal units.
The Typical Machine Data document was approved unanimously by M&VWG members.
VI. HVDC MODELING
Pouyan Pourbeik (EPRI) provided an update on HVDC modeling.
Accomplishments to date include:
 Multi-terminal conventional HVDC powerflow model – had been there
 Multi-terminal VSC HVDC powerflow model (Multi-terminal in PSLF, twoterminal in PSS@E and PowerWorld) - now in beta version and being
tested/benchmarked
 Conventional HVDC dynamic model - in progress
 VSC HVDC dynamic model - in progress
The 2014 Goals are as follows:
 Complete the testing of the VSC powerflow model and get it out in the tools
 Complete the specification of the dynamic model for conventional HVDC and
testing and hopefully have the software vendors start implementation by early
2015
Irina Green (CAISO) asked if TransBay will be asked to use the generic model once the
VSC model approved. Pouyan Pourbeik (EPRI) and Ronnie Lau (PG&E) commented that
the generic model is available to use, but it is up to TransBay to decide whether they will
continue the user-written model or switch to the generic model.
VII. RAS AND RELAY MODELING
Stephanie Lu (SCL) provided an update on the MSRATF activities, including relays, RAS,
contingency definitions, node breaker topology, and sequence component network
modeling.
Amos Ang (SCE) presented the current differential relay specification, and motioned to
approve the current differential relay model specification. Some members approved,
and some abstained. The motion to approve the current differential relay model
specification was approved by M&VWG. Though, many members abstained.
Pouyan Pourbeik (EPRI) who abstained motioned to not approve the wide-use of the
current differential relay model for full-scale system studies yet until more studies are
Page 9 of 13
performed. Dmitry Kosterev (BPA) seconded. Many agreed. Both of them as well as
most members believe that there is no or little value of deploying the current
differential relay and distance relay throughout the system. The current way of
modeling fault clearing explicitly is adequate to address the immediate reliability
concerns. Having the wide use of the detailed relay modeling could be a huge burden on
the TPs and the ability to have a clean relay model database is also questioned
considering the status of the current dynamic database. It is good to make these models
available in the software packages; however the need, cost, value, and increased
database management, etc. for adding data collection requirements for all the new
relay models to the WECC Data Preparation Manual have not fully been identified.
Dmitry and Pouyan as well as some others believe that the RAS and out-of-step
protection should have the highest modeling priority, and request Stephanie Lu (SCL) to
bring the feedback to TSS.
There was extensive discussion on the philosophy of modeling primary and secondary
protection relays. Baj Agrawal (APS) commented that planners should model the relay
that trips first and the breaker failure condition, regardless of the primary and
secondary terminology. Also to make sure that there is value before deploying relays
throughout the system. Eric Bakie (Idaho Power) mentioned to the team that similar
discussion occurred in SRWG and the conclusion was to model primary and secondary.
Pouyan Pourbeik (EPRI) again emphasized his point of only seeing value of modeling
out-of-step and RAS system wide, as other protection is local and can be captured in the
swt file. Mahendra Patel (EPRI) echoed on the comment, and indicated that we should
understand the value before putting into action. He also mentioned that EPRI is working
on a whitepaper of the value of modeling relay and asked if anyone interested and have
the relay modeled in their system to contact him. Dmitry Kosterev (BPA) indicated that
he has used a generic model for distance relay across the WECC system and has not
seen it operate. He has agreed that only out-of-step and RAS are necessary. Eric Allen
(NERC) commented to model the generic characteristics and only some specific ones in
detail. Hassan Ghoudjehbaklou (SDG&E) commented that it is also important to model
overcurrent relay. Overall, the team agreed to have the model available, but there are
different opinions as to whether to deploy them system wide. The team did agree that
RAS and out-of-step relays are important and should be deployed system wide.
Amos Ang (SCE) brought up another issue of how to model relays for generation
protection, if gp1/gp2 is adequate, if we need to modify gp1/gp2, or if we develop a
group of generation protection models. John Undrill commented that gp1/gp2 was
designed to represent the relay performance behavior not to replicate the relay physical
action. It is adequate to use gp1/gp2 for monitoring and serve as indication of generator
protection action.
Following the meeting, Dmitry provided a written justification for his abstain vote to the
differential relay model specifications. A summary of the group discussion is provided
above, earlier in this section, and his more detailed response is provided below:
Page 10 of 13
While detailed models of relays are certainly welcomed in grid simulators, the wide use
of such models in full-scale system studies needs further discussion.
There seem to be two drivers for representing relays in system studies:
a) Relays can potentially operate during power swings or events of Fault-Induced
Delayed Voltage Recovery, and study engineers need to be aware of such risks
b) Relays models clear faults automatically in dynamic simulations
Certainly (a) looks like more immediate and greater reliability concern. Default distance
relay models are available in GE PSLF, and we used these models over the past two years
in monitoring mode. The primary concern was operation of distance relays during FIDVR
events characterized by low voltages and high reactive power flows. Up to now, we have
not seen distance relay operation for FIDVR events or severe marginally stable power
swings. Analysis of actual events (incl. September 8) analysis suggests that distance
relays are very unlikely to operate unless the system is very deep into collapse, i.e. the
place you are not supposed to be.
Item (b) is a long-term effort. It is also “nice to have”, as the current way of modeling
fault clearing explicitly is adequate to meet the Reliability Standards. In my opinion, item
(b) over-reaches the original intent of September 8 outage recommendations. I am also
very concerned that resource requirements to support such effort are not clear, and
there are very wide differences in estimated effort among various TPs. A pilot project by
one utility would be very instructive before imposing such burden on the interconnection.
There is also risk that implementing (b) will create excessive reliance on database
management rather that engineering and understanding of the protection systems.
There are also questions about data management, as relay database will be significantly
larger than the current dynamic database. Today, many generators are consistently
have off status in base case cases because their associated dynamic models are unstable
or absent. Imagine an operating study engineer who needs to complete the study having
to deal with randomly operating relays because of bad data (just look at messages given
by gp1 , g2 and oel1 models). The first thing I do today when I get WECC case is
comment all OEL models for my studies. I am very skeptical in our abilities to maintain a
pristine quality relay database.
Response-based RAS has been identified as the high modeling priority, along with overcurrent relay models. May 16 event reminded once again why dynamic models of
response-based RAS are high priority. May 26 and June 12 Alberta separation events
reminded why out-of-step protection needs to be modeled.
I see the modeling priorities as follows (10-high, 1-low):
- (10) Model response-based RAS schemes (e.g. Colstrip ATR, Kemano RAS, Chief
Joseph brake, etc ) and out-of-step protection (e.g. MATL RAS, ….)
- (9) Model over-current relays and RAS – there are many schemes in WECC similar to
Imperial Valley RAS
- (6) Review adequacy of generic relay models, and used them in studies in monitoring
mode for potential interactions
Page 11 of 13
-
(2) Put detail relay modeling on back-burner, add the models in the programs as
lower priority, and then have a pilot project to determine the effort and to do costbenefits analysis over the current practices. It is also important to benchmark
against what PCs and TPs are doing in the Eastern Interconnection
With this in mind, I abstain from approving differential relay models as
a. differential relays are much lower priority compared to response-based RAS and outof-step relay modeling,
b. the existing approaches are adequate to address immediate reliability concerns,
c. costs with inter-connection wide use of relay models for clearing faults automatically
are not known,
d. ability to have a clean relay model database is questioned.
VIII. NERC UPDATES
Eric Allen (NERC) provided an update on the NERC SAMS/MWG group activities. The
details can be found at
http://www.nerc.com/comm/PC/Agenda%20Highlights%20and%20Minutes%202013/Pl
anning%20Committee%20Meeting%20Presentations%20-%20June%201011,%202014%20(Orlando,%20FL).pdf
The activities include:
 Benefit analysis for node-breaker representation in off-line and real-time study
models
 Model validation field trial
 SW outage recommendations 5, 6, 7, & NERC2 on Seasonal Assessment and sub100 kV elements
 Standardized component models
IX. WECC PROJECT UPDATES
A) WECC-0101 - SAR Generator Validation Policy Conversion
Shawn Patterson (USBR), chair of the WECC-0101 project team, provided an update to
the regional variance that the drafting team has developed. The last draft included more
stringent requirement than what was required by NERC MOD-026 and MOD-027,
including baseline testing requirements, using the lower threshold of 10 MVA on unit
and 20 MVA on plant, and having a 10 year re-testing period unless it was an analog
exciter which would require 5 years.
There were total of 4 comments back: 1 agree, 1 disagree, and 2 half-half agree. One
comment, which is expected to be incorporated, is to limit the requirements to BES
Page 12 of 13
generators (20 MVA units, 75 MVA plants). Another comment is that there are not
many analog systems in the WECC system and therefore the 5 year applicability should
not be included. Shawn indicated that he has done research and found that about 3040% exciters are analog, and more than 50% governors are analog. He also noted that
there is no governor that can be completely digital. Another draft is expected to come
out in the near future.
B) WECC-0107 - Power System Stabilizer Design and Performance
Shawn Patterson (USBR) briefed the team on the WECC-0107 project. The drafting team
is taking most of the existing PSS requirements and putting it into a standard. The
applicable size of the generator is changed from 30 MVA and connected above 60 kV in
the old WECC policy to align with the BES definition of generators greater than 20 MVA
and connected above 100 kV. There were discussions among the drafting team
regarding the measurement and notification requirements of when the PSS is in/out of
service. The draft is expected to be posted soon.
X. APPROVED DYNAMIC MODEL LIBRARY AND STATUS UPDATE OF MODELS
It was identified that the wt1p_b model should have also been added to the WECC
Approved Dynamic Model Library. The document will be updated to reflect this.
Stephanie Lu (SCL) updated the Status Update of WECC Models document with the
differential relay that just got approved, as well as adding a note that there is a RAS
common format, which the MATL and Bridger RAS models are expected to be able to
use.
The MVWG approved documents are posted on the MVWG Documents area on the
WECC website in the MVWG Approved Model Specifications folder and the MVWG
Approved Documents folder.
XI. PROGRAM UPDATES
GE PSLF, PTI PSS®E, and PowerWorld provided program updates. Presentations are
posted on the meeting website.
Jeni Mistry (SRP) talked about the need of expanding lsdt2 model to accommodate their
5 stages of UVLS. PSLF and PowerWorld have expanded it from 3 to 6 stages. PSS @E
currently has 3 stages, and will also expand to 6 stages.
Page 13 of 13
Download