Look-­‐ahead Dispatch in ERCOT: Case Study Le Xie (Lxie@ece.tamu.edu) Department of Electrical and Computer Engineering Texas A&M University Eighth Annual Carnegie Mellon Conference on the Electricity Industry March 13, 2012 Join work with Y. Gu (Texas A&M), D. Obadina (ERCOT), and X. Luo (Formerly at ERCOT) 2 Outline MoPvaPon Look-­‐ahead Dispatch Overview ERCOT Overview Preliminary Case Study Results Based on Benchmark ERCOT System •  Discussion •
•
•
•
3 The OperaPonal Challenge of IntegraPng Variable GeneraPon 4 ERCOT Load and Wind Power Curves 65,000 6,000 ERCOT Load and Wind Power Curves (July 14, 2010) 60,000 Load 50,000 5,000 Wind Power 4,500 45,000 4,000 40,000 3,500 35,000 30,000 0:00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 Load（MW) 55,000 3,000 Wind Power （MW) 5,500 5 With High Wind PenetraPon [1] (Day-­‐ahead forecast) (10-­‐minute ahead forecast) (Substan>al accuracy improvement from Day-­‐ahead to near real Pme) 6 Key Problems with ConvenPonal SCED •  Significant need for fast and expensive units (e.g. natural gas) •  Under uPlizaPon of slow responding units •  PolluPon caused by volaPle ramping of fast units •  Consequently, higher O&M cost and avoidable polluPon •  No incenPves to reduce ramping rate-­‐related costs (socialized UC cost) Recent Literature Review •  Wind forecasPng techniques constantly improving [Boherud, Wang, Miranda, Bessa 2010] [Xie, Gu, Zhu, Genton, 2011] •  Value of real-­‐Pme pricing on cost and value of wind power based on assumed demand elasPcity [Sioshansi, 2010] •  Value of coordinaPng wind with deferrable loads [Papavasiliou, Oren, 2010] •  Industry transiPon from staPc real-­‐Pme dispatch to look-­‐ahead dynamic dispatch [Oh, 2010] •  Preliminary study of look-­‐ahead dispatch with price responsive demands [Ilic, Xie, Joo, 2011] Our presentaPon will be using real-­‐world data based on a look-­‐
ahead dispatch model 7 8
What We Propose QuanPfying System Benefits Using Real-­‐world Data GeneraPon (look-­‐ahead dynamic dispatch) C&I price responsive demand (from ERCOT) Benefits of Look-­‐ahead Dispatch: A Conceptual Example 9 Load G1 Pgmax=80MW 3$/MWh G
G
Wind Coal Natrual Gas Conven@onal SCED 0:00 0:05 Ava. Wind 65MW 80MW G1 65MW 60MW G2 40MW G3 Load G
G2 Pgmax=150MW 30$/MWh 15MW/5mins G3 Pgmax=120MW 40$/MWh Pgmin:5MW 20MW/5mins Look-­‐ahead Dispatch 0:00 Total Cost 227.08$ Ava. Wind 0:05 65MW 80MW G1 65MW 80MW 25MW G2 20MW 5MW 5MW 5MW G3 25MW 5MW 110MW 90MW Load 110MW 90MW Total Cost 198.75$ 9
ERCOT InterconnecPon q Single InterconnecPon, with no synchronous Pes to any other region in North America q 71,800 MW of GeneraPon and 65,776 MW Peak Load System (85% of Texas’s electric load) q Responsible for all transmission equipment operaPng at 60 kV and above (69, 138, 345 kV) q Five asynchronous Ties to external regions o  1 in North Texas (600 MW), 1 in East Texas (220 MW), 3 to Mexico (. ) q Facilitates compePPve markets to help achieve reliability q Operates as Nodal Market as of 1st Dec, 2010 10 11 ERCOT Quick Facts q 85% of Texas load q 550 generaPng units, 3800 staPons q Peak demand: 68379 MW (08/03/2011) q Wind capacity: ~10,000 MW q Wind generaPon record: 7599 MW (03-­‐07-­‐2012), ~22% of load at that Pme 12 ERCOT Nodal Market Design q
Energy Dispatch
o  Resource specific offers
o  Resource specific dispatch
q
Goal
o  Balance generation & demand
o  Manage all Congestion
q
Energy Pricing
o  Local prices for energy
o  Prices reflect all congestion costs
13 ERCOT Nodal System Overview CRR Auction •
•
•
•
Annual & Monthly AucPons CRR Offers and Bids PTP OpPons and ObligaPons Flowgate Rights Day-­‐Ahead Market •
•
•
•
Hourly Market Energy Offers and Bids Ancillary Service Offers DAM PTP ObligaPon Bids Reliability Unit Commitment •  Transmission Security Analysis •  Resource commitment •  Day-­‐Ahead RUC •  Hourly RUC Real-­‐Time Opera@ons •  Network Security Analysis •  Security Constrained Economic Dispatch (SCED) •  5 Minutes Dispatch •  Load Frequency Control (LFC) 14 Components of ERCOT Nodal Market 48 Hours
Short-Term Wind Power Forecast
168 Hours
Look-Ahead Dynamic Ratings
168 Hours
Mid-Term Load Forecast
Large Wind Ramp Alert
Day 1
6 Hours
Day 2 Day 3
SCED
HRUC
DRUC
DAM
Operating Period
10:00
14:30
OH-1 Operating
Hour
This Talk 0 1 16 31 46 61 76 91 106 121 136 151 166 181 196 211 226 241 256 271 286 301 316 331 346 361 376 391 406 421 436 451 466 481 496 511 Genera@on Output at peak load (MW) 15 Benchmark of ERCOT 1400 1200 ERCOT 1000 Benchmark 800 600 400 200 Generator Units Figure 1: Genera@on Output During Peak Load Time 16 Look-­‐ahead v.s. Benchmark SCED Table 1: Comparison of Two Dispatch Methods for a Typical Day (Jul 11, 2009) Benchmark SCED Look-­‐ahead (30 min) Difference EnPre Day $ 26,191,710 $ 26,144,585 $ 47,125 Early Morning $ 3,514,925 $ 3,506,689 $ 8,326 Peak Wind Period $ 1,226,447 $ 1,219,948 $ 6,499 Wind GeneraPon (MWh) 96071 MWh 96432 MWh 361 MWh Early Morning: midnight-­‐8am, July 11, 2009 Peak Wind Period: 3am-­‐5am, July 11, 2009 17 Look-­‐ahead v.s. Benchmark SCED Different Look-­‐ahead Horizon Daily Cost Saving by Looking-­‐ahead 50000 45000 Cost Saving 40000 35000 30000 25000 20000 15000 10000 Benchmark SCED 5000 0 1 2 3 4 5 6 Look Ahead Horizon (Steps: 5 mins) 8 Figure 2: Look-­‐ahead Horizon Response of Total Savings 10 12 18 Look-­‐ahead v.s. Benchmark SCED ComputaPonal Time Computa@on Time for one day (seconds) Computa@on Time for Look-­‐ahead Dispatch (per interval) 210 200 190 180 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0 1 2 3 4 5 6 7 8 Look-­‐ahead Horizon 9 10 11 Figure 3: Computa@on @me over different horizons 12 13 Look-­‐ahead v.s. StaPc System GeneraPon Cost at Each Step 140000 Total Opera@ng Cost ($) 130000 120000 110000 100000 90000 80000 70000 60000 StaPc (Total) 50000 Look-­‐ahead (Total) 40000 1 21 41 61 81 101 121 141 161 181 201 221 241 261 281 Time Step (5 mins) Figure 4.a: Overall System Opera@ng Cost Profile 19 20 Look-­‐ahead versus StaPc System GeneraPon Cost: Zoomed In 110 StaPc (Total) Total System Cost (k$) 100 Look-­‐ahead (Total) 90 80 70 60 50 40 40 42 44 46 48 50 52 54 56 Figure 4.b: Overall System Opera@ng Cost Profile (Zoomed in) 58 60 21 Look-­‐ahead v.s. StaPc Natural Gas Output in a Typical Day Genera@on of Natural Gas Units (MW) 9000 StaPc 8000 Look-­‐ahead 7000 6000 5000 4000 3000 2000 1000 0 40 42 44 46 48 50 52 Time Steps (5 mins) 54 Figure 5: Genera@on of Natural Gas Units (Zoomed in) 56 58 60 22 Look-­‐ahead v.s. StaPc Coal GeneraPon in a Typical Day Genera@on of Coal Units (MW) 26000 StaPc 25000 Look-­‐ahead 24000 23000 22000 21000 20000 19000 18000 40 42 44 46 48 50 52 Time Steps (5 mins) 54 Figure 6: Genera@on of Coal Units (Zoomed in) 56 58 60 23 Look-­‐ahead v.s. StaPc Wind GeneraPon in a Typical Day Genera@on of Wind Units (MW) 12000 10000 8000 6000 4000 StaPc 2000 Look-­‐ahead Available Wind 0 40 42 44 46 48 50 52 Time Step (5 mins) 54 Figure 7: Genera@on of Wind Units (Zoomed in) 56 58 60 Nodal Price: Look-­‐ahead v.s. StaPc Economic Dispatch Four typical buses are selected for detailed nodal price study. Bus Number (ERCOT) Bus Name Base kV Area Number/
Name Zone Number/
Name 1626 LEON5_9 69 2 NORTH 5 NORTH_CE 5396 CALAVER_ 22 4 SOUTH 6 SOUTH_CE 6272 ABOV2A 69 1 WEST 8 WEST 8104 LANCTYPM 138 3 HOUSTON 1 COAST 24 Nodal Price: Look-­‐ahead v.s. StaPc Economic Dispatch 35 25 20 15 10 5 0 More spikes are observed in nodal prices of staPc approach than the ones of look-­‐ahead approach STD Values StaPc Dispatch: 1.96 Look-­‐ahead Dispatch: 6.12 StaPc (Bus 1000) Look-­‐ahead (Bus 1000) 1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161 171 181 191 201 211 221 231 241 251 261 271 281 Nodal Price ($/MWh) 30 Time Steps (5 mins) Figure 8: Nodal Prices at Bus 1626 on July.16th 25 Nodal Price: Look-­‐ahead v.s. StaPc Economic Dispatch 35 30 20 15 10 5 0 -­‐5 -­‐10 A lot more price spikes are observed in staPc dispatch compared with look-­‐ahead dispatch STD Values StaPc Dispatch: 7.37 Look-­‐ahead Dispatch: 4.38 StaPc (Bus 3020) Look-­‐ahead (Bus 3020) 1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235 244 253 262 271 280 289 Nodal Price ($/MWh) 25 Time Steps (5 mins) Figure 10: Nodal Prices at Bus 6272 on July.16th 26 Nodal Price: Look-­‐ahead v.s. StaPc Economic Dispatch Preliminary finding: •  Look-­‐ahead dispatch leads to a more smoothed nodal price pahern •  The nodal prices at selected buses may be higher under look-­‐ahead dispatch than in staPc dispatch 27 28 Summary •  Preliminary study shows the potenPal benefits of look-­‐ahead dispatch in –  Reducing the need for fast responsive units during wind ramping –  Reducing the overall generaPon cost •  ObservaPons for discussion –  Look-­‐ahead horizon: tradeoff b/w computaPonal cost and improved economics efficiency –  Impact on market price: does look-­‐ahead lead to more smoothed nodal price [7]? What should be a good price incenPve for inter-­‐temporal variaPons? More work needs to be done! 29 Ongoing Work •  Empirical study of look-­‐ahead dispatch with price-­‐responsive demands at ERCOT [6] •  Spa>o-­‐temporal wind forecast for look-­‐ahead dispatch with forecast uncertainty and infeasibility handling [7][8] •  CoordinaPon of wind and storage (explicit and implicit) for joint provision of energy balancing and frequency regula>on services [9][10][11] •  MulP-­‐scale integraPon of physics-­‐based and data-­‐driven models of distributed resources for ubiquitous energy storage services [12] 30 Acknowledgement •  Data •  Financial Support References [1] L. Xie, P. M. S. Carvalho, L. A. F. M. Ferreira, J. Liu, B. Krogh, and M. D. Ilić, "IntegraPon of Variable Wind Energy in Power Systems: OperaPonal Challenges and Possible SoluPons," Proceedings of The IEEE, January 2011.
[2] R. Sioshansi, “EvaluaPng the Impacts of Real-­‐Pme Pricing on the Cost and Value of Wind GeneraPon,” IEEE Transac>ons on Power Systems, May 2010.
[3] Papavasiliou and S. Oren, “Supplying Renewable Energy to Deferrable Loads: Algorithms and Economic Analysis”, IEEE PES -­‐GM 2010
[4] A. Oh, “Unit commitment in the PJM day-­‐ahead and real-­‐Pme markets,” FERC Technical Conference on Increasing Market and Planning Efficiency Through Improved SoRware and Hardware, Washington DC, June 2010.
[5] M. D. Ilić, L. Xie and J. Joo, “Efficient CoordinaPon of Wind Power and Price-­‐Responsive Demand Part I: TheorePcal FoundaPons, Part II: Case Studies,” IEEE Transac>ons on Power Systems, Nov 2011 [6] L. Xie, Y. Gu, X. Luo, D. Obadina, “Look-­‐ahead Dispatch: ERCOT Case Study,” Technical Report TAMU-­‐ECE-­‐2012-­‐01, January 2012. [7] Y. Gu and L. Xie, “Look-­‐ahead Dispatch with Forecast Uncertainty and Infeasibility Management,” IEEE PES-­‐GM 2012. [8] L. Xie, Y. Gu, X. Zhu, and M. G. Genton, “Power System Economic Dispatch with SpaPo-­‐temporal Wind Forecast,” IEEE EnergyTech, May 2011 (journal versions under review, Journal of the American StaPsPcal AssociaPon) [9] A. Thahe, and L. Xie, "Towards a Unified OperaPonal Value Index of Energy Storage in Smart Grid Environment," IEEE TransacPons on Smart Grid (accepted, to appear in 2012). [10] L. Xie, Y. Gu, A. Eskandari, and M. Ehsani, "Fast MPC-­‐based CoordinaPon of Wind Power and Bahery Energy Storage Systems,"Journal of Energy Engineering (accepted, to appear in 2012). [11] A. Thahe, F. Zhang, and L. Xie, “Frequency aware economic dispatch,” NAPS 2011. [12] L. Xie, F. Zhang, and M. Ilic, “MulP-­‐scale integraPon of physics-­‐based and data-­‐driven models in electric energy systems,” ICCPS 2012 (invited) 31 32 Thank You Le Xie (Lxie@ece.tamu.edu) Assistant Professor Department of Electrical and Computer Engineering Texas A&M University www.ece.tamu.edu/~lxie