Incorporating FACTS Set Point Optimization In Day-Ahead Generation .

Incorporating FACTS Set Point Optimizationin Day-Ahead Generation SchedulingConfidential. Not to be copied, distributed, or reproduced without prior approval.Kwok W. Cheung, Ph.D., PE, PMP, FIEEEJun WuDirector, Global Market Management SolutionsGE Grid Software SolutionsSenor Software EngineerGE Grid Software Solutions

Outline BackgroundProblem and Mathematical FormulationSolution AlgorithmsPreliminary Numerical ResultsConclusions

Background

Introduction Security and reliability functions of grid operators: Day-ahead generation scheduling is a typical business process of grid operators formarkets and non-markets

Introduction (cont’d) RTOs/TSOs are reliant on wholesale market mechanism to optimally dispatch energy andancillary services [3]Indust rialResident ialCommercialAggregat orDISCOTSOMarket erRTO / ISOGENCOkWhRETAILCOBrokerMarket erinfocontract

Wholesale Power Market Platform In the beginning of this century, FERC pushed for a common market design framework calledStandard Market Design or the so-called Wholesale Power Market Platform. Variation of such amodel (LMP-based two-settlement system) has been adopted by all RTOs in the United States. As energy-only markets approached maturity, RTOs one after another enhanced their energymarkets to incorporate clearing of ancillary services. A co-optimized approach of clearing energy and ancillary services simultaneously has beenextensively accepted by all restructured electricity markets in USA.RTO/ISONetworkModel spatchPointsDualSolutionsof SCEDBid DataFERC Technical Conference June 2013 - P 6MarketOperatorMarketPrices

Growth of Renewable Power and High Ramp Net Load

Smart Dispatch to Cope with Uncertainty of Renewable PowerRTM Day-Ahead Market (DAM) process provides functions for day-ahead bid datasubmission, market clearing, and marketsolution publishing. Reliability Unit Commitment (RUC) process- provides system operators a set of tools torevise the day-ahead unit commitmentschedule as necessary in order to ensurethat the forecasted load and operatingreserve requirements will be met and thetransmission system is reliable andsecured. Look-Ahead Commitment and Dispatch(LACD) process – provides a forwardlooking view of system operating conditionsand recommend start-up/shut-downrecommendation of fast-start resources tooperators. Real-Time Market (RTM) process –provides market-clearing functions tobalance generation and load, and meetreserve requirements based on actual realtime system operating conditions. The RTMprocess computes ex-ante pricing andprovides the dispatch signals either MW orprice back to the Market Participants.Electronic DispatchLACDRUCFERC Technical Conference June 2013 - P 8

Problem and Mathematical Formulation

Transmission Congestion The increase in renewable generation connected to the grid has exacerbated the problemof transmission congestion considerablyo Copper-plate assumptiono Loop flow problem

Transmission Congestion Leads to High Production Cost15:3016:0016:30Source: PJM Due to transmission constraints, the economic merit order dispatch is not feasible Some low-cost units have to decrease their production, while some high-cost units have toincrease their generation Production costs increase by the order of billions of dollars annually due to transmissioncongestion in USA

Solutions to Transmission Congestion Market-based congestion managemento Locational pricing and dynamic pricingo Transmission constraints need to be considered under the framework of SCUC/SCED Dynamic line ratings Optimal transmission switching FACTS controlo Transmission including FACTS are traditionally treated as non-dispatchable asset in thenetworko Co-optimizing transmission and generation dispatch has the potential to further maximize themarket surplus

Basic Unit Commitment & Economic Dispatch Model Alternative Transmission flow model(Power Transfer Distribution FactorPTDF model) Location marginal price (LMP) Promoted by FERC, LMP methodology isproven to be an effective mechanism torelieve transmission congestion and toachieve market efficiency LMP is the foundation for market-basedcongestion management

Concept of FACTS Set Point OptimizationMotivation: Further improve social welfare in market clearing by changing lineimpedancesControl of transmission not fully utilized today Transmission assets are treated as static in the short term Current transmission control only for reliability purposes:1.2.Operators change transmission assets’ states on ad-hoc basisSpecial Protection Schemes (SPSs) Alternative to topology control (discrete) [1][2] – less disturbingFlow control via FACTS ARPA-e GENI projects on“Distributed Power Flow Control” [4]Incorporate state of transmission assets into generationdispatch co-optimizationSource: https://www.smartwires.com

Challenges of FACTS Control in Optimization FACTS devices can achieve controlfunctions such as voltage regulation,system damping, and power flow control. FACTS control seem to be a viable way toleverage grid controllability for enhancingsystem efficiency. The problem of SCUC with FACTS controlis a Mixed Integer nonlinear programming(MINLP) problem. The problem of SCED with FACTS controlis a nonlinear programming (NLP) problem.f kt Bk (θ mt θ nt ), k , tBoth are decision variables in the DC load flow equations

Solution Algorithms

Two-stage LP-based Solution Algorithm [5]Assumption: The flow directions of all FACTS lines are knownor can be pre-determined from the initial DCOPFPositive voltage angle difference enforces zk to take 1 as its value, while negative voltage angle difference set to zk to 0.

Power Flow Direction Changes in Power Grids with SignificantRenewable Penetration [6]

Proposed Solution AlgorithmNLPLP1 LP2Bk frozenLP2LP1θ nt frozen

Proposed Solution Algorithm (cont’d)Solve two LP problems iteratively untilthey converge to one solutioni) LP1: Fixed B, solve LP and get Θii) LP2: Fixed Θ , solve LP and get BStep to solve mathematical model1)Fixed B at normal admittance, solve LP1 andobtain Θ2)Fixed Θ, solve LP2 and get a new B,3)Fixed B at new B, solve LP1 again and getnew Θ4)Repeat step 2) - 3), until Δ Θ (differencebetween two iterations) approaches 0i 0B B0Θ0 0P Θ 10yes Δ Θ ThresholdnoStop & OutputResultFix B i ,Solve LP- Θii i 1Δ Θ P Θ- ΘiFix Θi 1 ΘiSolve LP- B i 1P Θ ΘiB i B i 1

Preliminary Numerical Results

Test Cases---5 Bus SystemObjective comparisonBefore FACTS Opt.28947.18After FACTS Opt. Cost Savings ( /hr)1760.6527186.53Cleared Generation comparisonClearedGenerationafter (MW)Unit110110185TARE SOLITUDE UN C Steam229.14311.0324TARE BRIGHTON UN E Steam534.86452.972.5TARE ALTA UN PARKCITY SteamTransmission flow comparisonAdmittance comparisonTransmission LinesSUNDANCE 230 KV D-E LNALTA 230 KV A-B LNALTA 230 KV A-D LNALTA 230 KV A-E LNPNODE B 230 KV B-C LNSOLITUDE 230 KV C-D LNALTA 230 KV A-C LNClearedGeneration Unit Pricebefore (MW) ( -1348MinAdmittanceAdmittance -1037-1037-1037-1037-1037Transmission LinesSUNDANCE 230 KV D-E LNALTA 230 KV A-B LNALTA 230 KV A-D LNALTA 230 KV A-E LNPNODE B 230 KV B-C LNSOLITUDE 230 KV C-D LNALTA 230 KV A-C LNTransmission Transmission TransmissionFlow (after)Flow (before) .5140087.84143.93380

Test case- 37,000-bus, 47,000-branch Test System 370 dispatchable generation units 28 transmission lines violated/binding at limitsObjective comparison:Objective( /hr) Total Offer Cost( /hr) Violation Cost( /hr)Before FACTS After FACTS Optimization991628.73991628.730FACTS Optimization Cost Savings14,650,744.1479,572.9714,571,170.17

Test case- 37,000-bus, 47,000-branch Test System (cont’d)Solution Iterationiteration Objective1 (fix B) Objective2 (fix 29991629 Θi-Θi-1 .4291.8110.641.3220.6370.5270.4950.8320.4680.000 Θmn(i)-Θmn(i-1) 00010010009910001 2 3 45 6 7 89 10 1112 13 1415 16 1718 19 2021 22

Test case- 37,000-bus, 47,000-branch Test System (cont’d)Admittance ComparisonTransmission linesSQBUTTE SQBUTSQBUT23 4 1 LNROCHSTR ROCHST 622QB15 1 LNSQBUTTEW SQBUTSQBUT23 1 1 LNSYCAMOR SYCAMSYCAM34 1 S LNSQBUTTEW SQBUTCENTE23 2 1 LNHIWAY106 HIWAYHIWAY69 2 S LNROXF ROXF WRR 1 A LNBVR CH BVR CBVR C69 1 S LNROCK IS ROCK ROCK 16 1 S LNHAZLTON HAZLTON99534 1 S LNCORYDON CORYDCORYD69 1 S LNRUSHUPA RUSHURUSHN23 1 1 LNAUDUBON AUDUBAUDUB23 1 1 LNKIPPRD KIPPRVEVAY13 1 1 LNOTTOWATP IP-1516 2 LNWHITE WHITEBRKNG34 1 1 LNLKST MER-WAT-2 G LNLAPOINTE Y-43 4 4 LNHAYWARD HAYWARD66716 1 S LNTILDEN T IP-1476 NE LNWRSN WRSN-MASN-4 A LNLAKEFLD LAKEFLAKEF34 1 S LNTAZE CE TAZEW4525TIE 1 1 LNEATH TAP WLWD-GRAY-1 B LNSQIN SQIN BT A LNWLWD WLWD-GRAY-2 A LNHAYWARD HAYWARD66369 1 S LNFT INDUS FI STERLING 1 LN AdjustedadmittanceMax/Min Admittance is about ( /-)30% of admittanceMin 106110291850123143440010981385877301080124347579 .Max Admittance Admittance 01723350862701723190169 52128043440013032006516001303200353171 0098533415410024625012311002462271670

Test case- 37,000-bus, 47,000-branch Test System(cont’d)Branch flow comparisonBefore FACTS Optimization:Transmission LinesALMA2 ALMAWABAC16 1 1 LNARROWHD TR7 TR7 XFBAILLY 138 U8 A LNBNE JCT BNE JAMES11 1 1 LNBR GR PN BR GRPONTI34 1 1 LNBUTTEDES 43021 43 LNCANIFF P3 P3 PSCBLUFFS R922 TR922 XFCOFFEEN IP-4551 COFY A LNCULLEY CULLEGRAND13 1 1 LNDUNKARD DUNKALTV13 1 1 LNEDWARDS2 EDWD KEYS 1397 A LNFRANCESC FRANCPETER34 1 1 LNGOOSECRK IP-4575 1 LNHAZLTON HAZLTBLKHA16 1 1 LNI FALLS 10TR IFAL10TR PSMGN 35351 35 LNMILAN3 MILANAJ MA34 1 1 LNPETERSBU PETERTHOMP34 1 1 LNPLYMOUT2 13819 A LNPORTUNIO PORTUZIMME34 1 1 LNSANDLAKE SAL WAU SA LNSCHAHFER 34508 A LNSCHAHFER 34516 A LNSIOUXCY U1A1 KU1A1 XFSUB 92 SUB 9HILLS34 1 1 LNWMIDDLET 6997 69 LNLONEROCK TX00 LOR PH SHFT0 PSAfter FACTS Optimization:Limits (MW)Flow -13517512-49Transmission LinesALMA2 ALMAWABAC16 1 1 LNBNE JCT BNE JAMES11 1 1 LNLimits (MW)Flow (MW)1051052020BR GR PN BR GRPONTI34 1 1 LN115115CANIFF P3 P3 PS300300COLUMPSI 1 2 1 SAME2 XF8181COLUMPSI 1 3 1 SAME3 XF8181CULLEY CULLEGRAND13 1 1 LN100100DUFF DUFFDUBOIS 1 A LN170170HENNEPIN IP-1556 1 LN107107PINE0000 62101 62 LN2727PINE0000 62201 62 LN2727FLORNCE 62103 62 LN9393341341BELLTAP BELLTVANBT69 1 1 LN68606860ANDOVE ANDOVDCRK69 1 1 LN686068604444686068606767OAKRDG X-91 1 LNELENDLE ELENDOAKGL69 1 1 LNBIXB BIXBRIVER69 1 1 LNMAPLEDPC MAPLEMAPLE16 1 1 LNBefore FACTSOptimization: 7transmission lines areviolated their limits, 21transmission lines arebinding at limits.After FACTSOptimization: 18transmission lines arebinding at limits.