UCTE 2002 Sample Loadflow Case
Summary
This study presents a user study case received from an InterPSS user. It discusses AC Loadflow and DC Loadflow studies of the UCTE network, representing the continental European countries, using InterPSS. Also, performance testing data are presented regarding the usage of the IEEE ODM model to transfom the original case data file in the IEEE Common Data Format into the OMD model.
Please Note : You can run the UCTE 2000 sample case in in InterPSS 2.0: Summer Case, Winter OffPeak Case, Winter Peak Case
Case Description
Hi, InterPSS Support
There is an Approximate Model of European Interconnected System:
Maybe you want to use it as a sample file in InterPSS? I think it could help other TSOs to convince that InterPSS is capable
of calculating real networks in the size of the UCTE.
I tried to import this file into InterPSS but it faild with a NullPointerException.
However, this sample network doesn't include R, B and G values for branches (only X is given). So it is only suitable for DC
loadflow. (With PSAT, AC loadflow doesn't converge.) Or maybe it would be possible to simply estimate R from X?
Further links:
(1) Qiong Zhou and Janusz W. Bialek, "Approximate Model of European Interconnected System as a Benchmark System
to Study Effects of Cross-Border Trades", IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 20, NO. 2, MAY 2005
(2) Janusz W. Bialek Publications List
Best Regards
Mario
Österreichische Elektrizitätswirtschafts-AG
VERBUND-Austrian Power Grid AG (APG)
Marktmanagement (UMM)
UCTE System
An approximate model of UCTE system containing 1254 buses and 378 generators.
AC Loadflow Study
The IEEE CDF sample files (*.CF) are used for the stusy.
Please Note: InterPSS uses IEEE ODM model as the intermediate step. IEEE ODM model is an XML document. All id in a XML document needs to start with a character. Therefore all bus numbers are prefixed with "No". You will see Bus 1 in IEEE CDF file is turned to bus id "No1" in InterPSS representation.
IEEE CDF Format
InterPSS first uses the IEEE ODM Adapter for IEEE CDF format, developed by InterPSS development team, to translate data in IEEE CFD format. The main concept is outlined in the following:
IEEE CDF -> IEEE ODM Model -> InterPSS Object Model
We found some information, such as Year, Month and branch circuit number are missing in the sample file. The information was feedback to the IEEE ODM Adapter development team. The adapter has been updated to handle the situation gracefully. After the modification, the sample data files could be transfered to IEEE ODM model. See more discussion below.
Non-divergence Loadflow
Three sample cases (2002 Summer, 2002 Winter Offpeak, 2002 Winter Peak) have been added to InterPSS sample workspace under the ieee_format project folder.
We found that three Loadflow cases are very challenging cases in term of convergence. Since, we do not have any business domain knowledge, we cannot explain the reason why normal Newton-Raphson (NR) or Fast-Decoupled (PQ) won't produce converged Loadflow results. We have to use the advanced InterPSS non-divergence Loadflow approach. The following are the Loadflow calculation results:
2002 Summer Case
Loadflow converged in 6 iterations with the following mismatch information:
Max Power Mismatches
Bus dPmax Bus dQmax
---------------------------------------------------
No209 0.000006 No211 0.000005 (pu)
0.572009 0.527872 (kva)
2002 Winter Offpeak Case
Loadflow converged in 6 iterations with the following mismatch information:
Max Power Mismatches
Bus dPmax Bus dQmax
---------------------------------------------------
No204 0.000000 No206 0.000000 (pu)
0.012215 0.011278 (kva)
2002 Winter Peak Case
Loadflow stopped after 6 iterations with the following mismatch information:
Max Power Mismatches
Bus dPmax Bus dQmax
---------------------------------------------------
No1229 0.356976 No1227 0.137612 (pu)
35697.627874 13761.239239 (kva)
InterPSS stopped after 6 iterations, since no better solution could be found. Please note, this may be a case with Loadflow solution. However, InterPSS's non-divergence Loadflow algorithm could not produce better results.
Contingency Analysis Using Grid Computing
Research team at South China University of Technology, Guangzhou, China, lead by Prof Yao Zhang 张尧), Prof Zhigang Wu (武志刚) and Stephen Hau (侯冠基) created a 10 CPU computing grid using 2 notebooks, 6 desktop computers and a 100M router, details as follows:
NoteBook(x1) NoteBook(x1) Desktop(x3) Desktop(x3)
CPU 2 x 1.86GHz 2 x 1.72GHz 2.66GHz 2.4GHz
Memory 1 GB 2GB 760MB 512MB
Grid Node 1 maser 2 remote 3 remote 3 remote
2 remote one/machine one/machine
They installed and configured InterPSS on one notebook as the master grid node and Gridgain agents on all computers as the remote grid nodes. 2500 contingencies were created and simulated by running 2500 AC Loadflow. The computation time using one CPU and 10 CPU is shown in the following table:
1 CPU 10 CPU Acceleration
Time(min) Time(min) Ratio
250 26 9.5
Also, the contingency analysis 2500 Loadflow runs were repeated by gradually increase the number of CPU.
The investigation demonstrated:
- Linear scalability - InterPSS Grid Computing solution has linear scalability, 10 CPU allows you to achieve almost an order (10 times) of computation speed increase.
- Practical Solution - The research team used their existing computers to build the computation grid. They completed the installation and experiment almost independently, with minimum support from the InterPSS development team.
More information about InterPSS Grid Computing Solution can be found Here
DC Loadflow Study
InterPSS has a Power Market Solution, which includes DC Loadflow and sensitivity analysis. InterPSS DC Loadflow and sensitivity analysis are applied to the sample system.
2002 Summer Case
DC Loadflow
DC Loadflow Results
Bud Id VoltAng(deg)
=================================
No1 0.0000
No2 -8.21075
No3 6.2732
No4 10.87944
No5 0.93345
No6 -9.79714
No7 -5.10159
No8 -2.22967
No9 -8.95315
....
FromId->ToId Power Flow(pu)
======================================
No1->No2(1) 0.19279
No1->No2(2) 0.19254
No1->No79(1) -0.19435
No6->No2(1) -0.02432
No6->No2(2) -0.02543
No4->No3(1) 0.24915
No5->No3(1) -0.46461
No4->No8(1) 1.35745
No8->No4(2) -1.33608
No5->No7(1) 0.18624
...
Power Transfer Distribution Factor
Inject BusId : No497
Withdraw BusId : No514
Branch Id PTDF
========================================
No499->No497 -0.55667
No498->No514 0.21884
Assume 1.0 pu power is injected at generator bus 497 and withdrew from load bus 514, PTDF (Power Transfer Distribution Factor) tells the amount of power of the 1.0pu flowing on a particular branch.
Please Note: since none of us knows the actual UCTE network, no attempt had been made to interpret the simulation results.
The following InterPSS Xml Scripts are used for the calculation.
<ipss:InterPSS xmlns:ipss="http://www.interpss.org/schema"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<ipss:runStudyCase>
<ipss:analysisRunType>RunDclf</ipss:analysisRunType>
<ipss:dclfStudyCaseList>
<ipss:dclfStudyCase>
<ipss:recId>PDTF_STUDY</ipss:recId>
<ipss:pTransferDistFactor>
<ipss:senType>PAngle</ipss:senType>
<ipss:injectBusId>No497</ipss:injectBusId>
<ipss:withdrawBusId>No514</ipss:withdrawBusId>
<ipss:branch>
<ipss:recId>Branch499_479</ipss:recId>
<ipss:fromBusId>No499</ipss:fromBusId>
<ipss:toBusId>No497</ipss:toBusId>
</ipss:branch>
<ipss:branch>
<ipss:recId>Branch498_514</ipss:recId>
<ipss:fromBusId>No498</ipss:fromBusId>
<ipss:toBusId>No514</ipss:toBusId>
</ipss:branch>
</ipss:pTransferDistFactor>
</ipss:dclfStudyCase>
</ipss:dclfStudyCaseList>
</ipss:runStudyCase>
</ipss:InterPSS>
More information about InterPSS Xml Scripts could be found at InterPSS Power Market Solution
An Open Model for Exchanging Power System Simulation Data, ODM - Open Data Model for short, is currently under development. The model is currently sponsored by the IEEE PES OSS (Open Source Software) Task Force, chaired by Prof Federico Milano of Univ. Castilla - La Mancha.
The model is designed to solve real world power system simulation problems. These problems, like the one discussed in this study case, are often quite complex and large in scale. Performance is usually the first concern when selecting a new approach. Some performance tests have been conducted using the Adapter Runtime included in the ODM Model Refererence Implementation.
Please Note: Currently, the Adapter Runtime is an open-source implementation, contributed by InterPSS and PSAT, maintained by InterPSS. You can download from InterPSS SourceForge Download Site.
File Size
Filename Format Size
UCTE_2002_Summer.CF IEEE CDF 426K
UCTE_2002_Summer.xml IEEE ODM 1592K
The ODM Xml document is about 4 times large than the original IEEE CDF file.
2 P-2 1201 0 1.0000 -13.21 868.10 0.00 0.00 0.00 380.00 0.0000 0.000 0.000 0.0000 0.0000 0 2
The above line, a line of bus data, in the IEEE CDF file is converted into the following Xml statements
<pss:bus>
<pss:name>P-2</pss:name>
<pss:area>1</pss:area>
<pss:zone>201</pss:zone>
<pss:id>No2</pss:id>
<pss:baseVoltage>
<pss:voltage>380.0</pss:voltage>
<pss:unit>KV</pss:unit>
</pss:baseVoltage>
<pss:loadflowBusData>
<pss:voltage>
<pss:voltage>1.0</pss:voltage>
<pss:unit>PU</pss:unit>
</pss:voltage>
<pss:angle>
<pss:angle>-13.21</pss:angle>
<pss:unit>DEG</pss:unit>
</pss:angle>
<pss:loadData>
<pss:code>CONST_P</pss:code>
<pss:load>
<pss:p>868.1</pss:p>
<pss:q>0.0</pss:q>
<pss:unit>MVA</pss:unit>
</pss:load>
</pss:loadData>
</pss:loadflowBusData>
</pss:bus>
As we all know, to solve engineering problems, "No pain, no gain". What have been added are Xml tags, something called metadata, or data about data. With the additional information, anyone with power system analysis background and some Xml syntax training can read the Xml document with almost no ambiguity. The original IEEE CDF format is not really human-readable. The major advantage of the Xml representation is that the Xml document could be easily and very efficiently processed by Xml standard confirming tools. There are lot of open-source, free Xml processing tools available from such reputable companies, like IBM, SUN and Microsoft. All tools used in the ODM reference implementation are open-source solutions.
CUP Time
Using a Ienovo IBM ThinkPad T60 laptop, CUP for loading the UCTE_2002_Summer.CF and converting to an Xml DOM object confirming to the ODM Schema was about 1.6 sec.
Conclusion
Our conclusion is that the ODM model and accompanying reference implementation, although still under development, are ready for solving real world power system simulation problems. More important is that Xml technology and open-source Xml processing tools are very mature now. We should take advantage of Xml and make extensive use of the technology in power system simulation.