Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL Photos placed in horizontal position with even amount of white space between photos and header CLIMATE VARIABILITY AND POTENTIAL IMPACTS ON ERCOT ELECTRICITY GENERATION Presentation to ERCOT July 13, 2012

Results of Climate Projections and Hydrologic Modeling for Texas-Gulf Basin  Eugene Yan  Yonas Demissie  Esther Bowen  Christopher Harto  John Gasper  Argonne National Laboratory  Vincent Tidwell  Barbie Moreland  Katie Zemlick  Barry Roberts  Howard Passell  Sandia National Laboratories  Mark Wigmosta  Teklu Tesfa  Andre Coleman  Pacific Northwest National Laboratory  Carey King  Margaret Cook  University of Texas

TASK 6: CLIMATE VARIABILITY IMPACTS ON ELECTRICITY GENERATION  Evaluate impacts of future (2030s) climate variability, drought scenarios, and water demand  Impacts on water availability  Potential reduction or curtailment of power generation  Low lake levels  Thermal effluent limitations  Identify vulnerable watershed at USGS HUC-8 basin level  Support WECC and ERCOT’s long-range transmission planning and studies

Current and Projected Water Picture

Thermoelectric Water Consumption 2011 Total of 392,000 AF

Alternative Futures for Thermoelectric Water Consumption  Wind Case: assumes wind is 13.2% of total production in 2030  No Wind Case: assumes no new wind and decommission of all existing wind. Total of 510,000 AF Total of 450,000 AF

Climate Study Area: Texas-Gulf River Basin and ERCOT Region ERCOT,  Basins having strong potential for loss of electricity generation under drought scenarios  Majority of generation capacity using surface water for cooling, increasing sensitivity to climate variability

Methods of Analysis (1)

Methods of Analysis (2)

Future Climate Projection Data Processing  Climate Downscaled Datasets  USGS CASCaDE dataset: daily projections at 1/8th degree (~12x12 km) spatial scale without bias correction (daily T min, T max, P)  US Bureau of Reclamation dataset: bias corrected and spatial downscaled (BCSD) daily projections at 1/8th degree spatial scale (daily T min, T max, P; monthly average wind speed)  Future Emissions Scenarios  A1B, A2, and B1.  GCMs  NCAR’s Parallel Climate Model 1 (PCM)  NOAA’s Geophysical Fluid Dynamics Lab (GFDL) CM2.1 model  Additional Climate Parameters  Incoming long- and shortwave radiation, estimated vapor pressure, relative humidity, and dew point temperature

Future Climate Projections

Future Precipitation Projections and Drought Scenarios  Percent changes in precipitation in 2030, 2022, and 1956 compared to 2011  Drought scenarios: 2022 and

Hydrologic Modeling – Model Construction  Water Uses  Current consumptive water uses  Projected 2030 consumptive water uses  Five water sources for consumptive uses: river, reservoir, pond, shallow groundwater, and deep groundwater  Construction of Texas-Gulf River Basin Model with Soil and Water Assessment Tool (SWAT) Increase in use of surface water in 2030

Hydrologic Modeling – Model Construction (Cont’d)  125 Reservoirs Included in the Model  Large volume  Close to power plants  Monitored by TWDB  Climate Forcing Data  Historical climate data interpolated from GHCN 2085 stations  Future climate data: USBoR BCSD GFDL A  Drought scenarios: 2022;

Hydrologic Modeling – Model Calibration  Stream Flow Calibration  Calibration period:  Calibration with observed data from 9 USGS gauges

Hydrologic Modeling – Model Calibration (Cont’d)  Reservoir Calibration  Calibration with available reservoir monitoring data

Hydrologic Modeling – Drought Simulations  Baseline: Recent Drought Year in Texas (2011)  Single-Year Drought Scenario  Climate forcing data: BCSD GFDL A2 dataset in 2022  Future water demand in 2030  Multiple-Year Drought Scenario  Climate forcing data: Historic GHCN dataset in  Future water demand in 2030

Hydrologic Modeling Results – Single-Year Drought  Projected Drought Conditions  Severity  Duration  Spatial distribution Percent change in water yield to surface water from 2011 to 2022

Hydrologic Modeling Results – Single-Year Drought (Cont’d)  Projected Reservoir Capacity in HUC-8 Basins

Hydrologic Modeling Results – Single-Year Drought (Cont’d)  Projected Percentage Changes in Stream Flow Rate in HUC-8 Basins from 2011 to 2022

Hydrologic Modeling Results – Multiple- Year Drought

Hydrologic Modeling Results – Multiple-Year Drought (Cont’d)  Projected reservoir storage in HUC-8 basins under drought scenario

Hydrologic Modeling Results – Multiple- Year Drought (Cont’d)  Projected percent changes in stream flow in HUC-8 basins under drought scenario (baseline 2011)

Drought risk to ERCOT power generation – not conclusive  Analysis to date has too much uncertainty to project drought risk to power generation  Assumption too simple: “all reservoirs in HUC-8 basin have same percentage level of storage”  Many reservoirs used for cooling water are operated at constant level  Next SWAT modeling steps need to model ‘human’ interventions  Normal lake management practices (e.g. constant level lakes)  Normal inter HUC-8 basin transfers (?)

Operations near thermal limit

Operations near thermal limit in future summers

Watershed Vulnerability  Water use vs. water yield to surface water 1956

Watershed Vulnerability (Cont’d)  State water plan, water available for new appropriation  Availability for “drought of record” and over 75% availability Outlined basins are locations of projected siting of new power plant capacity. Weight of line designates potential water demand. Note: 500 MW power plant at 85% CF and 600 gal/MWh  ~ 7,000 AFY

Watershed Vulnerability (Cont’d)  Availability based on difference between recharge and pumping  Cost based on well field and pumping Note: 500 MW power plant at 85% CF and 600 gal/MWh  ~ 7,000 AFY

Watershed Vulnerability (Cont’d)  Availability based on irrigated pasture not exceeding 5% of basins total irrigated acreage  Cost based historic payments for water rights transfers out of irrigated agriculture Note: 500 MW power plant at 85% CF and 600 gal/MWh  ~ 7,000 AFY

Watershed Vulnerability (Cont’d)  Availability based on projected 2030 waste water production less current reuse for plants that do not discharge to perennial stream  Cost based conveyance and treatment Note: 500 MW power plant at 85% CF and 600 gal/MWh  ~ 7,000 AFY

Watershed Vulnerability (Cont’d)  Availability based on state projections of brackish water availability  Cost based on pumping and treatment Note: 500 MW power plant at 85% CF and 600 gal/MWh  ~ 7,000 AFY

Watershed Vulnerability (Cont’d) Much of projected expansion is sited in basins with limited water availability Note: 500 MW power plant at 85% CF and 600 gal/MWh  ~ 7,000 AFY

Eugene Yan for Hydrologic Modeling Progress Report, June 20, 2012 Summary of Study and Findings  Twelve sets of climate data were evaluated from two downscaled datasets from PCM and GFDL for emission scenarios A1B, A2, and B1. The BCSD GFDL A2 climate data were used as forcing data for hydrologic modeling, representing the worst climate condition.  Drought simulations include single-year drought (2011 as baseline and 2022) and multiple-year drought ( ).  The 2022 single-year drought scenario:  Drought duration and severity same as or worse than 2011 baseline  Different spatial drought distributions among HUC-8 basins  The multiple-year drought scenario:  Reservoir water storage declining gradually over time in drought  Drought more severe and widespread in its final year than 2011 baseline  The 2011 and multi-year 1950’s drought appear to provide the best available basis for planning over the next years.

Summary of Study and Findings  The SWAT modeling approach is not yet accurate enough to enable us to foresee risks to drought. The reasons are:  Many lakes with power plants are managed to be constant level lakes and we’re not modeling them that way.  Our assumption of all reservoirs in a HUC-8 are at same storage level is too coarse to be informative (a lot due to 1.a above).  We might have to model some transfers of water from one HUC-8 to another (but not sure we have the data to do this)  The effluent temperature historical data analysis and regressions to ‘predict’ if higher air temperatures and higher capacity factors during summer has some good merit for a 1 st -order kind of analysis  We can clearly see a handful of plants at risk to some curtailment (but not too many).  Using future climate data shows that if you assume higher future air temperatures, these plants have more tendency to have to hold back in summer due to effluent temperature limits.

Vincent Tidwell (505)