A METHODOLOGY FOR CALCULATING EMISSIONS REDUCTIONS FROM RENEWABLE ENERGY PROGRAMS AND ITS APPLICATION TO THE WIND FARMS IN THE TEXAS ERCOT REGION

Transcription

1 A FOR CALCULATING S FROM RENEWABLE ENERGY PROGRAMS AND ITS APPLICATION TO THE WIND FARMS IN THE TEXAS ERCOT REGION Zi Liu Ph.D., Jeff Haberl Ph.D., P.E., Juan-Carlos Baltazar Ph.D., Kris Subbarao Ph.D., Charles Culp Ph.D., P.E., Bahman Yazdani P.E. Energy Systems Laboratory Texas A&M University November 19, 27 Energy Systems Laboratory 1

2 APPLICATION Outline Introduction Methodology Daily Modeling Procedure Analysis on Sweetwater I Wind Farm Capacity Factor Analysis Application to All Wind Farms Uncertainty Analysis Emissions Reduction Summary November 19, 27 Energy Systems Laboratory 2

3 APPLICATION Background Texas Legislature Requirement Adding 5,88 MW of generating capacity from renewable energy by 215, 5 MW from non-wind renewables. PUCT: A target of 1, MW by 225 TCEQ: Required creditable methodology for computing emissions reductions from renewables to report to the US EPA Energy Systems Laboratory Assist TCEQ to develop SIP-creditable calculations for EE/RE to report to the US EPA. Developing procedures to calculate emissions reduction from all wind farms in ERCOT area Weather normalization of the daily wind power multiple base years selected by TCEQ and US EPA November 19, 27 Energy Systems Laboratory 3

4 APPLICATION Wind Farms in Texas Completed ERCOT: 293 MW WSCC: 1 MW SPP: 122 MW Construction ERCOT: 726 MW SPP: 161 MW Announced ERCOT: 3125 MW ERCOT: Electric Reliability Council of Texas SPP: Southwest Power Pool WSCC: Western Systems Coordinating Council Source: /electric/maps/gentable. pdf (date: 3-6-7) Wind Farms in ERCOT Wind Farms in WSCC Wind Farms in SPP 24 El Paso 1 5 Culberson 49 Jeff Davis Pecos 32 Oldham 46 Ector Martin 41 Upton Hansford 25 Carson Borden 2122 Sterling 4 1 Childress Floyd Cottle ERCOT Power Grid and Wind Farms in Texas Scurry Howard Shackleford 37 Erath Taylor Jack Nolan Source: (date: 3-6-7) Kenedy 48 Wind Projects Completed: ERCOT Region 293 MW 1 Culberson, 35 MW, Texas Wind Power, 1/ Howard, 34 MW, Big Spring Wind Power, 2/ Howard, 6.6 MW, Big Spring Wind Power, 7/ Upton, 75 MW, Southwest Mesa Wind, 6/ Culberson, 3 MW, Delaware Mountain, 6/ Pecos, 82.5 MW, Indian Mesa I, 6/21 7 Pecos, 16 MW, Woodward Mountain, 7/21 8 Nolan, 15 MW, Trent Mesa, 11/21 9 Pecos, 16 MW, Desert Sky (Indian Mesa II), 12/21 1 Upton, 278 MW, King Mountain, 12/21 11 Scurry, 16 MW, Brazos Wind, 12/23 12 Nolan, 37.5 MW, Sweetwater Wind 1, 12/23 13 Nolan, 91.5 MW, Sweetwater Wind 2, 2/25 14 Nolan, 135 MW, Sweetwater Wind 3, 12/25 15 Taylor, 114 MW, Callahan Divide Wind, 2/25 16 Taylor, 12 MW, Buffalo Gap1, 9/25 17 Taylor, 213 MW, Horse Hollow Phase 1, 1/25 18 Taylor, 224 MW, Horse Hollow Phase 2, 5/26 19 Taylor, 299 MW, Horse Hollow Phase 3, 9/26 2 Borden, 84 MW, Red Canyon 1, 5/26 21 Sterling, 124 MW, Forest Creek, 12/26 22 Sterling, 9 MW, Sand Bluff, 12/26 23 Shackleford, 2 MW, Lone Star (Mesquite), 12/26 WSCC Region 1 MW 24 El Paso, 1 MW, Huecon Mountain, 4/21 SPP Region 122 MW 25 Carson, 79 MW, Llano Estacado, 1/22 26 Hansford, 3 MW, Aeolus Wind, Hansford, 4 MW, JD Wind 1, 2, 3, 1/26 Wind Projects Under Construction: ERCOT Region 726 MW 28 Taylor, 233 MW, Buffalo Gap 2 (Cirello 1), 3/27 29 Scurry, 13 MW, Camp Springs, 5/27 3 Nolan, 3 MW, Sweetwater 4, 12/27 31 Scurry, 63 MW, Snyder Wind, 12/27 SPP Region 161 MW 32 Oldham, 161 MW, Wildorado Wind Ranch, 27 Wind Projects Announced: ERCOT Region 3125 MW 33 Cottle, 126 MW, Wild Horse Wind 1, 6/27 34 Cottle, 39 MW, Wild Horse Wind 2, 8/28 35 Floyd, 6 MW, Whirlwind, 9/27 36 Jack, 12 MW, Barton Chapel Wind 1, 1/27 37 Erath, 6MW, Silver Star Phase I, 12/27 38 Shackleford, 2MW, Lone Star Wind (Post Oak), 12/27 39 Scurry, 29 MW, Roscoe Wind, 12/27 4 Howard, 59 MW, Ocotillo Windpower 1, 12/27 41 Martin, 11 MW, Stanton Wind, 12/27 42 Childress, 11 MW, Childress Wind, 5/28 43 Kenedy, 4 MW, Gulf Wind 1, 9/28 44 Kenedy, 4 MW, Gulf Wind 2, 9/29 45 Kenedy, 4 MW, Gulf Wind 2, 7/21 46 Ector, 3 MW, Notrees Windpower, Kenedy, 4 MW, Penascal Wind, MW, Galveston Offshore Wind, 21 Wind Projects Retired: ERCOT Region 7MW 49 Jeff Davis, 7MW, Ft. Davis Wind Farm, 1996 November 19, 27 Energy Systems Laboratory 4

5 APPLICATION Texas Wind Power Generation Installed Capacity as of 3/7 All Texas 9, Texas Wind Power Generation (Source: ERCOT & PUC) 3,6 8, ,2 Installed Capacity as of 3/7 - ERCOT Ozone Season Period (Jul. Sep.) Measured Capacity in ERCOT (MW) Wind Power Generation ( MWH) 7, 6, 5, 4, 3, 2, 1, 2,8 2,4 2, 1,6 1,2 8 4 Installed Cap acity ( MW) Measured Power Generation in ERCOT (MWh) Sep- 1 Dec- 1 Mar- 2 Jun- 2 Sep- 2 Dec- 2 Mar- 3 Ercot Data - MWH Jun- 3 Sep- 3 Dec- 3 Mar- 4 Jun- 4 Sep- 4 Dec- 4 Mar- 5 Jun- 5 Ercot Data - M W Sep- 5 Dec- 5 Total Installed Capacity in Ercot Area by 26 Total Installed Capacity in Texas by 26 Mar- 6 Jun- 6 Sep- 6 Dec- 6 November 19, 27 Energy Systems Laboratory 5

6 APPLICATION Modeling Procedure Overview Hourly Wind Data On-site hourly wind data not available for the base year (not available or the wind farms not existing) NOAA s network of weather stations hourly wind speed data at 1 meter height available Hourly Wind Power Generation Data 15-minute data obtained from ERCOT for each wind farm since 21 Modeling Procedure NOAA hourly wind speed and measured power for the base year and study year (i.e and 25) converted to daily data Daily performance curve developed regressed the daily power generation against the daily average wind speed using ASHRAE s Inverse Model Toolkit (IMT) Weather Normalization: Coefficients from 25 IMT regression applied to the 1999 average daily NOAA wind speed Predicts the daily electricity the wind farm would have produced in 1999 base year. November 19, 27 Energy Systems Laboratory 6

7 APPLICATION Analysis on Sweetwater I Wind Farm Located in Nolan County, Texas Nearest NOAA station - Abilene Regional Airport ABI 37.5 MW Capacity, 25 GE 1.5s Wind Turbines Using hourly model to predict wind power production in baseyear (1999) impractical. GE 1.5s Wind Turbine Specifications Manufacture Nameplate Capacity Cut-in speed Cut-out speed Rotor diameter Tower height at hub Swept area Rotor speed GE 1,5 kw 4 m/s 25 m/s 7.5 m 8 m 394 m^ rpm Wind Power Generation (MW) Hourly Wind Power Generation vs. NOAA-ABI Wind Speed (SWEETWND 37.5 MW) Sweetwater, Nolan County Abilene,Taylor County NOAA Wind Speed (MPH) 25 Hourly wind power plotted against NOAA- ABI wind & manufacture s power curve November 19, 27 Energy Systems Laboratory 7

8 APPLICATION Sweetwater I Predicted vs. Measured Power in 25 Daily data more appropriate for modeling Daily model performed well for the entire year and OSP tracking the measured wind power production Wind Power Generation (MWh/day) July the biggest error (-19.34%) Wind Power Generation vs. NOAA-ABI Wind Speed (SWEETWND 37.5 MW) Measured Data Daily Regression Model NOAA Wind Speed (MPH) Daily Wind Power vs. NOAA-ABI Wind (25) IMT Coefficients Ycp (MWh/day) Left Slope (MWh/mph-day) RMSE (MWh/day) R2 CV-RMSE % November 19, 27 Energy Systems Laboratory 8 Month No. Of Days Average Daily Wind Speed (MPH) Measured Power Generation (MWh) Predicted Power Generation Using Daily Model (MWh) Diff. CV-RMSE Jan ,15 1, % 42.79% Feb ,13 7, % 43.4% Mar ,611 12, % 32.27% Apr ,597 14, % 22.98% May ,29 11, % 3.15% Jun ,323 12, % 2.98% Jul ,465 1, % 35.9% Aug ,882 7, % 31.71% Sep ,62 8, % 36.16% Oct ,167 8, % 35.57% Nov ,94 1, % 37.64% Dec ,322 1, % 34.43% Total , , % 32.76% Total in OSP (7/15-9/15) ,131 17, % 24.2%

9 APPLICATION Sweetwater I Predicted vs. Measured Power in July 5 Measured power not evenly distributed around the prediction Overestimation for the first half of the month Model fit data well in second half of the month Reason unknown Possible causes: Wind Power Generation (MWh/day) Wind Power Generation vs. NOAA-ABI Wind Speed (SWEETWND 37.5 MW) Measured Data Measured Data in Jul 5 Daily Regression Model NOAA Wind Speed (MPH) Curtailment? Wind Power Generation in July 25 (SWEETWND 37.5 M W) Maintenance? Others? 1 M easured Average Daily kwh Predicted Average Daily kwh - NOAA Daily M odel 8 NOAA-ABI Wind Speed /1/5 7/8/5 7/15/5 7/22/5 7/29/5 Date November 19, 27 Energy Systems Laboratory 9

10 APPLICATION Sweetwater I Predicted vs. Measured Power in OSP 5 Daily model performing well in OSP Predicted vs. measured: 3.56% Wind Power Generation (MWh/day) Wind Power Generation vs. NOAA-ABI Wind Speed (SWEETWND 37.5 MW) Measured Data Measured Data in OSP Daily Regression Model NOAA Wind Speed (MPH) Wind Power in 25 Ozone Season Period (SWEETWND 37.5 MW) 1 2 Measured Average Daily kwh Predicted Average Daily kwh - NOAA Daily Model NOAA-ABI Wind Speed Wind Power (MWh/day) July August September Wind Speed (MPH) 7/16/5 7/23/5 7/3/5 8/6/5 8/13/5 8/2/5 8/27/5 9/3/5 9/1/5 Date November 19, 27 Energy Systems Laboratory 1

11 APPLICATION Sweetwater I Testing of the Model Used: 24 measured power output and 24 NOAA wind speed vs 25 daily model coefficients Model sufficiently robust for predicting MWh in base year Nov 16.3% diff. Wind Power Generation (MWh/day) Wind Power Generation vs. NOAA Wind Speed (24) NOAA Wind Speed (MPH) Measured Data Measured in Nov. 4 Daily Regression Model 1 8 Wind Power Generation in November 24 (SWEETWND 37.5 M W) M easured Average Daily kwh Predicted Average Daily kwh - NOAA Daily M odel NOAA-ABI Wind Speed /1/4 11/8/4 11/15/4 11/22/4 11/29/4 Date Two days data missing November 19, 27 Energy Systems Laboratory 11

12 APPLICATION Sweetwater I Predicted Wind Power in hourly NOAA- ABI wind speed converted to average daily wind speed Used 25 daily model coefficients Estimated annual power increased about 15% when compared against 25, OSP power increased 9% Reason: 1999 windier than Estimated MWh/yr 25 Measured MWh/yr 143, , OSP Estimated MWh/day 25 OSP Measured MWh/day Surface plots for comparing ABI 1999 and 25 hourly wind speeds Shows importance of a weather normalization procedure for a more accurate estimation November 19, 27 Energy Systems Laboratory 12

13 APPLICATION Sweetwater I Capacity Factor Predicted 25 capacity factor using 25 daily model vs. measured capacity factor: The daily model performing well in predicting annual and OSP capacity factors (1% error). The biggest error in July (6%). Predicted capacity factor using 25 model for 1999 through 25 More variation in the year-to-year wind speeds than the uncertainty from the model Showed importance of a weather normalizing wind speed back to the base year Capacity Factor Capacity Factor 1% 9% 8% 7% 6% 5% 4% 3% 2% 1% % Jan- 5 1.% 9.% 8.% 7.% 6.% 5.% 4.% 3.% 2.% 1.%.% Capacity Factors Using Daily Model (SWEETWND 37.5 MW) Feb- 5 Mar- 5 Apr- 5 May- 5 Jun- 5 Jul-5 Aug- 5 OSP Sep- 5 Oct- 5 Nov- 5 Dec- 5 Month M easured CF NOAA-ABI Daily Model CF NOAA-ABI Wind Speed Capacity Factors Using NOAA Daily Model (SWEETWND 37.5 MW) 1999 CF 2 CF 21 CF 22 CF 23 CF 24 CF 25 CF Average CF ( ) M easured CF OSP Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Average Daily Wind Speed Per Month (mph) November 19, 27 Energy Systems Laboratory 13

14 APPLICATION Estimated Power Production in 1999 for Each Wind Farm in ERCOT 1999 estimated annual MWh using 25 coeff: all sites increased 1999 estimated OSP MWh/day using 25 coeff: 6 sites decreased, all other sites increased. Note: Blue text shows the wind farms built before 9/21. November 19, 27 Energy Systems Laboratory 14

15 APPLICATION Comparison of Annual 1999 Estimated vs. 25 Measured 1999 estimated annual MWh using 25 coeff: all sites increased Biggest increase: CALLAHAN +3%, H_HOLLOW +31% Highest annual production: TRENT 6, 5, 4, Wind Power Generation in Texas 25 Measured MWh/yr 1999 Estimated MWh/yr Using 25 Daily Model CALLAHAN H_HOLLOW TRENT MWh/yr 3, 2, 1, Wind Farms November 19, 27 Energy Systems Laboratory 15

16 APPLICATION Comparison of 1999 Estimated 25 Measured in OSP 1999 OSP MWh/day: 6 sites decreased, all other sites increased Biggest decrease: SW_MESA, -7% Biggest increase: WOODWRD1_WOODWRD1 1,4 Wind Power Generation in Ozone Season Period in Texas 1,2 25 OSD Measured MWh/day 1999 OSD Estimated MWh/day Using 25 Daily Model MWh/day 1, SW_MESA WOODWRD1_ WOODWRD1 2 Wind Farms November 19, 27 Energy Systems Laboratory 16

17 APPLICATION Started operation in July 25 Horse Hollow Wind Farm (22 MW) Power production during the testing period (July through September) excluded from the analysis 25 model using 3 months data from October to December Wind Power Generation (MWh/day) Wind Power Generation vs. NOAA-ABI Wind Speed (H_HOLLOW_WND1 22 MW) Measured Data Measured Data in OSP Daily Regression Model NOAA Wind Speed (MPH) November 19, 27 Energy Systems Laboratory 17

18 APPLICATION Comparison 1999 vs. 25 For All Wind Farms 6,, Wind Power Generation in Texas 5,, 4,682,682 Annual Total: Increased 17% MWh/yr 4,, 3,, 2,, 4,8, Method Measured 1999 Method 1 Predicted 1,, 25 Total Measured MWh/yr TOTAL 1999 Total Estimated MWh/yr Using 25 Daily Model Wind Power Generation in Ozone Season Period in Texas 16, 14, OSP: Increased 26% MWh/day 12, 1, 8, 6, 4, 8, Method Measured 9, Method 1 Predicted 2, 25 OSD Measured MWh/day TOTAL 1999 OSD Estimated MWh/day Using 25 Daily Model November 19, 27 Energy Systems Laboratory 18

19 APPLICATION Comparison of 1999 and 25 Wind Speed Four weather stations used in the modeling Annually, 1999 windier than 25 for all four weather stations In OSP, 1999 windier than 25 for ABI and FST In OSP, 25 windier than 1999 for MAF and GDP Average Monthly W ind Speed in 1999 and 25- NOAA-ABI Average Monthly W ind Speed in 1999 and 25- NOAA-MAF Wind Speed (mph) 2. OSP Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month ABI: 1% windier in % windier in 1999 OSP Wind Speed (mph) 2. OSP Jan Feb Mar A pr May Jun Jul A ug Sep Oct Nov Dec M onth MAF: 7% windier in % windier in 25 OSP 25. Average Monthly Wind Speed in 1999 and 25- NOAA-FST 25. Average Monthly Wind Speed in 1999 and 25- NOAA-GDP Wind Speed (mph) 2. OSP Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Wind Speed (mph) 2. OSP Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month FST: 9% windier in % windier in 1999 OSP GDP: 2% windier in % windier in 25 OSP November 19, 27 Energy Systems Laboratory 19

20 APPLICATION 1999 Annual and OSP Uncertainty of the Power Prediction The uncertainty of applying the daily models to predict the 1999 annual wind power ranged from 2.4% to 5.5% (Reddy et al. 1992). Daily models reasonably reliable for predicting the performance of the wind farm in the base year within the same range of wind conditions. The uncertainty related to predicting 1999 OSP wind power using annual model are higher than the annual values ( %). A model developed using only the measured power in the OSP would improve the reliability of the wind power prediction in the OSP Annual 1999 Ozone Season Period (OSP) Wind Farm Pred Total Total Rel Pred Total Total Rel 2 ( ) ( ) ( V ) j V 2 Days n Eˆ Variance Estimated = pred, j MSE Eˆ Uncer σ i 1+ 1 Days Variance Estimated Uncer + BRAZ_WND_WND , ,57 3.8% n n 63 5,28 45, % 2 BRAZ_WND_WND , ,97 3.7% ( V63 i Vn2,967 ) 26, % i= 1 CALLAHAN_WND , , % 63 4,31 6, % H_HOLLOW_WND1 * , , % 63 9,917 85, % INDNENR_INDNENRm , ,888 m 4.1% σ ( E pred, j ) ( E pred, j ) ( E pred, total ) 2 ˆ = σ ˆ = σ ˆ 4,642 4, % INDNENR_INDNENR_ ,94 249,34 4.4% 63 4,525 36, % KING_NE_KINGNE j= , ,71 j= 1 3.5% 63 2,789 26, % KING_NW_KINGNW 365 9, ,493 4.% 63 3,781 32, % KING_SE_KINGSE 365 3,492 95, % 63 1,449 12, % KING_SW_KINGSW 365 7,96 29, % 63 3,28 29, % ( Eˆ ) = MSE( Eˆ ) j m 1+ + n m m ( V j Vn ) SWEETWN2_WND , , % 63 3, , % SWEETWND_WND , , % ,756 19, % = 1 TRENT_TRENT σ pred, total ,487 i 563, % 63 6,843 n ( V ) 77, % 2 DELAWARE_WIND_NWP 365 2,864 68, % 61 1,171 i V 7, % n INDNNWP_INDNNWP 363 7, , % i= 163 2,989 23, % INDNNWP_INDNNWP ,436 97, % 63 1,846 14, % KUNITZ_WIND_LGE 365 2,393 43, % , % KUNITZ_WIND_LGE , % , % SGMTN_SIGNALMT 365 4,361 13, % 63 1,89 14, % SW_MESA_SW_MESA 365 9,16 217, % 63 3,778 3, % WOODWRD1_WOODWRD ,193 21, % 63 3,41 29, % WOODWRD2_WOODWRD , ,32 3.5% 63 2,842 27, % November 19, 27 Energy Systems Laboratory 2

21 APPLICATION Emissions Reduction from Wind Power in Texas Using 27 annual and OSD egrid Total MWh savings for each Power Control Area used to calculate the NOx emissions reduction for each of the different county through the USA-EPA prescribed emission fractions. Total MWh savings in the base year 1999 for the wind farms built before September 21 within the ERCOT region: 2,674,858 MWh/yr and 6,652 MWh/day in the Ozone Season Period. Total NOx emissions reductions across all the counties: 1,639 tons/yr 4.8 tons/day for the Ozone Season Period. November 19, 27 Energy Systems Laboratory 21

22 APPLICATION Summary Methodology for predicting electricity from wind farms in the 1999 base year using 25 measured electricity and NOAA wind data. The total wind power production in the base year (1999) and the corresponding emissions reduction calculated from all the wind farms in the ERCOT region using this procedure Result: Improvement of the accuracy of base year predictions using normalization procedure compared to the non-weather normalization procedure. The uncertainty analysis showed that the daily regression models are sufficiently reliable to allow for their use in projecting wind production into other weather base years. Results now combined with EE/RE for integrated NOx emissions reductions Tons/year Tons/Ozone Season Day 16, 14, 12, 1, 8, 6, 4, 2, Annual NOx reduction levels (Preliminary Estimates) All ERCOT ESL-Single Family ESL-Multifamily PUC (SB7) PUC (SB5 grant program) SECO Wind-ERCOT ESL-Commercial Federal Buildings Furnace Pilot Light Program SEER13-Single Family SEER13-Multif amily OSD NOx reduction levels (Preliminary Estimates) All ERCOT ESL-Single Family ESL-Multifamily PUC (SB7) PUC (SB5 grant program) SECO Wind-ERCOT ESL-Commercial Federal Buildings Furnace Pilot Light Program SEER13-Single Family SEER13-Multifamily November 19, 27 Energy Systems Laboratory 22

23 APPLICATION Future Work - ANNs NOAA Wind Speed (MPH) Hourly On-site Wind Speed vs. NOAA Wind Speed Input Layer Hidden Layer Output Layer On-site Wind Speed (MPH) 5 On-site Wind Speed vs. ANN On-site Wind Speed 45 ANN On-site Wind Speed (MPH) On-site Wind Speed (MPH) Multilayer perceptron neural net architecture for relating site wind (output) to (input) variables measured at the NOAA weather site: wind speed, wind direction, dew point temperature and dry bulb temperature November 19, 27 Energy Systems Laboratory 23

24 APPLICATION Future Work MORE ACCURATE WIND DATA On-site 5 meter and 1 meter wind speed data Texas Mesonet, Academy for Advanced Telecommunications and Learning Technology (AATLT) Goal: Hourly analysis, curtailment factors, forecasting, etc. November 19, 27 Energy Systems Laboratory 24

25 APPLICATION Questions? THANK YOU! November 19, 27 Energy Systems Laboratory 25