Integrating Variable Wind Power Using a Hydropower Cascade

Transcription

1 Integrating ariable Wind Power sing a Hydropower Cascade Andrew Hamann and Gabriela Hug Power Systems Laboratory, ETH Zürich 18 September th International Workshop on Hydro Scheduling in Competitive Electricity Markets Trondheim, Norway 1

2 Motivation Hydropower is an enormously flexible resource Capable of deploying stored water energy very quickly Fewer start-up and shut-down restrictions than thermal power plants Renewable but not necessarily environmentally friendly) How much flexibility can hydropower provide in real-time operations for balancing the variability from wind power? How do the operations of a coordinated wind-hydro system differ from those of a hydro-only system? 2

3 Motivation Hydropower is an enormously flexible resource Capable of deploying stored water energy very quickly Fewer start-up and shut-down restrictions than thermal power plants Renewable but not necessarily environmentally friendly) How much flexibility can the Mid-Columbia hydropower system provide in real-time operations for balancing the variability from wind power in the Pacific Northwest? How do the operations of a coordinated wind-hydro system differ from those of a hydro-only system? 2

4 Mid-Columbia System Mica Arrow Duncan Corps of Engineers Dams Dams Owned by Others Bureau of Reclamation Dams Portland Seattle Rocky Reach Wanapum PriestRapids Ice Harbor The Dalles Bonneville Chief Joseph Wells Rock Island John Day Hugh Keenleyside ColumbiaR Grand Coulee Lower Monumental McNary Albeni Falls Washington SnakeR Oxbow Brownlee Pend Oreille Lake Hells Canyon Kootenay Lake Little Goose Lower Granite Idaho Noxon Libby Dworshak Flathead Lake Canada S Hungry Horse Montana W illametter Oregon Boise Projects Jackson Lake Palisades SnakeR Source: Army CoE 3

5 Mid-Columbia System Name Type Head m) MW 1 Grand Coulee Federal Chief Joseph Federal Wells Municipal Rocky Reach Municipal Rock Island Municipal Wanapum Municipal Priest Rapids Municipal We have one year of timestamped data for this system, with a temporal resolution of five minutes 4

6 Wind Power in the Pacific Northwest Current and Proposed Wind Project Interconnections S pok an e to BPA Transmission Facilities Washington Wild Horse Expansion Eugene P ier ce Boise Inset Area Raymond Coyote Crest Wind Lewis South Bend Ocean Park Pac ific Gra nt Yak im a 17 Yakima Be nt on Star Point Bakeoven I/II Wa sc o 97 Leaning Juniper I & II Eight Mile Canyon ) " " ) Willow Creek South Hurlburt ) " ) " Olex Harvest Wind Montague I Gilliam County Nook Wind II & III Nine Canyon I/II * # Lower Snake River Co lum bia Stateline PAC) Stateline Nine Mile) WA S H IN G TO N Combine Hills II ansycle Ridge O R E G O N * Combine Hills I PAC) # Helix 37 ) " Pasco Condon Wind Gillia m Klondike III Phase 2 Brush Klondike I Canyon Wind 19 S her m an 197 FG 202 Echanis Wind 31 Juniper Canyon II Phase 2 White Creek Miller Ranch Wheat Field Goodnoe Hills Klic k it at Nook Wind 95 Linden Windy Flats 3 Whistling Tuolumne Windy Pt) Biglow Canyon Ridge ) " Windy Flats - Dooley Golden Hills II & III Patu Ho od The Dalles Riv e rgolden Hills Add'n Lotus Summit Ridge ) " Thresher Klondike III Phase 1 35 Klondike II Golden Hills I Hay Canyon FG ) " 4 Malhe ur Harne y Hampton Wind Burns Map date: 3/9/12 Lower Snake River 3 Hopkins * # Marengo Ridge 124 PAC) 1 & 2 Wa lla Walla Columbia Wind Big Horn Juniper Canyon I Harvest Wind Sand Ridge II East Klickitat 1-4 Juniper Canyon 2, 2A 82 Rattlesnake Road 14 Pebble Springs 730 North Hurlburt 20 Miles ) " Fra nk lin 97 West Butte Lake Des chut es Big Horn II Crook Area So Oregon) Inset Bak er 15 Lower Snake River Lower Snake River 2 Cowlitz Gr ant Wahk iak um 5 Whit m a n Wa llow a 204 Pendleton Harvest Wind Jordan Butte I/II Heppner Ella Wind III/I Horseshoe Bend 395 Montague II 74 Caithness Shepherds Flat II 4 Existing Wind Project - Other tility County Boundary Note: Wind project locations are approximate and intended for graphic purposes only Long Beach Proposed Wind Generation Lew is Saddle Moutain West Existing Wind Generation 23 Wind Project nder Construction * # Coastal Energy Ad am s 26 Kit t it a s Gr ays SWHarbo Wash) r 90 antage Wind 90 Oregon Legend ) " # * * Wild Horse 1&2 # * # Idaho 21 Portland 17 Desert Claim Kittitas alley Kin g Spokane Seattle n ion 244 M orr ow m a ti lla ) Antelope " Ridge Source: BPA 5

7 Modeling: Model Predictive Control xk + 1) = A xk) + B uk) for k = 0, 1,, K 1 System model is developed to predict the response of the system x to a sequence of control inputs u The system is optimized over a time-horizon k = 0, 1,, K The control sequence that gives the best performance over the time-horizon is computed Only the first-step of this control sequence is applied The reaction of the system is observed and the process is repeated at the next time-interval 6

8 Modeling: Hydraulic Model State variables: reservoir elevation, tailrace elevation Control variables: turbine discharge, spill, natural inflow/sideflow Reservoir elevation is a linear function of inflows and outflows Surface area assumed to be constant Storage is only sufficient for a few hours of operation Run-of-river with some flexibility Tailrace elevation is a function of turbine discharge, spill, and downstream forebay elevation ie, encroachment) Tailrace elevation changes more than forebay elevation Travel time of water between plants is considered Tens of minutes Constraints on turbine discharge, spill, forebay elevation, change in turbine discharge, change in spill 7

9 Modeling: Hydropower Generation pq, h) = κ η t q) η g q) q h pj q 4 j,p4 j q 5 j,p5 j Generation MW) q 2 j,p2 j q 3 j,p3 j 0 q 1 j,p1 j 200 hj qj Hydraulic head m) Turbine discharge m 3 /s) h max j h min j qj min qj max Figure 10: Diagram of a piecewise linear HPF This function corresponds to the lower limit for the auxiliary variable q i j, or q i,min j hj) qj i 18) Ideally, for a particular turbine discharge qj, only one auxiliary variable qj i will not be tight to 8 either its upper or lower boundary The upper limit for qj i is thus

10 Modeling: Hydro Power Balance J j=1 p jk) = p hydroload k) Power Demanded GW) Wind Load Hydro Load Hour 9

11 Modeling: Wind power J j=1 p jk) + ϵk) = p hydroload k) + p windload k) p wind k) ϵ is wind curtailment if negative ϵ is load curtailment if positive p wind is wind generation p windload is additional wind load We use BPA wind generation data for p wind and BPA balancing area load data for p windload scaled such that N n=1 p windload = N n=1 p wind 10

12 Modeling: Objective function min qj,s j,ϵ { K 1 k=0 J j=1 [ aj q j k) 2 + c j s j k) 2] + K 1 k=0 d ϵk)2 } a j = c j = ηj η j+1 Ψj+1 Ψ j ) 2 d a j Weight turbine discharge and spill to encourage the transfer of water from large surface area reservoirs to small surface area reservoirs Choose the weight d on ϵ large to put a heavy penalty on load or wind curtailments 11

13 Case Study: 120 Hours in July 2012 Power Supplied GW) Curtail Hydro System Load Wind Hour Load curtailment when wind generation was low and hydro hit its upper capacity limit 12

14 Case Study: 120 Hours in July 2012 Historical Hydro Hydro and Wind Wells Turbine Discharge m 3 /s) Rocky Reach Rock Island Wanapum Priest Rapids Hour 13

15 Case Study: 120 Hours in July Historical Hydro Hydro and Wind Wells 8000 Rocky Reach Spill m 3 /s) 8000 Rock Island 8000 Wanapum 8000 Priest Rapids Hour 14

16 Case Study: Statistics 4884 MW avg of load 2874 MW avg of hydro generation 1664 MW avg of wind generation 346 MW avg of load curtailment 357 MW avg of spilled water 5396 MW peak load and 3838 MW peak wind generation 34% wind energy penetration 71% wind capacity penetration 15

17 Case Study: Ramping Ramping Score j Name Hist Hydro H+W 1 Wells Rocky Reach Rock Island Wanapum Priest Rapids Ramping score is proportional to the sum of the absolute change in the turbine discharge 16

18 Future Work Apply this framework across different hydraulic conditions, system constraints, and wind/load scenarios What are the appropriate metrics? Ramping, unit cycling, spilled water energy, spilled wind energy, etc? What should the additional load from wind look like? Should it be normal load, hourly blocks, on-peak blocks, off-peak blocks? Should thermal power plants be modeled? What does it mean when we say that we want to balance wind using hydropower? What are the best evaluation metrics? 17

19 Funding Acknowledgements Electric Power Research Institute EPRI) Hydro Fellowship, Hydro Research Foundation ) Robert W Dunlap Graduate Fellowship, Steinbrenner Institute for Environmental Education and Research ) Department of Engineering and Public Policy, Carnegie Mellon niversity Bertucci Graduate Fellowship, Carnegie Mellon niversity Spring 2013) 18

20 Questions? Thank you 19