StarSapphire: Data-driven Modeling and Analysis
WindSENSE

We need to understand wind resources better to manage their increased contribution to the grid (see Figure 1). WindSENSE considers the problem from the viewpoint of the control room operators. Our interviews with them have indicated that they would like 1) more accurate wind power generation forecasts for use in scheduling and 2) additional information that can be exploited when the forecasts do not match the actual generation. To achieve this, WindSENSE has two areas of focus:

Analysis of existing historical data for improved scheduling: We are analyzing historical data to improve our understanding of extreme events, such as ramps. We used simple statistical analysis to understand if ramps occur more frequently during certain months or certain times of the day, if negative ramps are as frequent as positive ramps, and if the different ramp definitions change the severity and frequency of events identified in the data. We also identified weather conditions associated with ramp events, enabling schedulers to plan for days when extreme events are likely. Our on-going work includes the identification of daily patterns in the data to provide guidance to the schedulers (for example, a rise in generation from early evening to mid-night, followed by a gradual decrease in the early morning hours).

Observation targeting for improved forecast: This work, being done in collaboration with AWS Truepower, identifies the locations and types of observations that can most improve short-term and extreme-event forecasts. The wind generation forecasts used by utilities are based on computer simulations, driven by observations assimilated into the time progression of the simulation. It has been found that observations of certain variables at certain locations have more value than others in reducing the forecast error in the variable of interest, at the location of interest, for the appropriate look-ahead period. We are using an Ensemble Sensitivity Analysis approach to identify these locations and variables (see Figure 2). We have applied this approach to the Tehachapi Pass and Columbia Basin regions. We are now validating our analysis using data denial experiments and an observation system simulation experiment.

Figure 1: Actual wind generation (in MW) in the Bonneville Power Administration balancing area from 2007-2009.

 

Figure 2: Average sensitivity of 80-m wind speed at the target box (in white, indicating the Klondike wind farm in the BPA balancing area) to 80-m wind speed elsewhere three hours earlier. Results are from an ESA for a 45 day period in summer 2007.

 

This work is joint with AWS Truepower and a collaboration with Bonneville Power Administration (BPA), California Independent System Operator (CaISO), and Southern California Edison (SCE).

Publications

C. Kamath, “Associating Weather Conditions with Ramp Events in Wind Power Generation,” 2011 IEEE PES Power Systems Conference & Exposition, March 20 - 23, 2011, Phoenix, Arizona.

LLNL-CONF-457176

J. Zack, E. J. Natenberg, S. Young, G. V. Knowe, K. Waight, J. Manobianco, and C. Kamath, “Application of ensemble sensitivity analysis to observation targeting for short term wind speed forecasting in the Tehachapi region winter season,” LLNL Technical Report LLNL-TR-460956, October 2010.

LLNL-TR-460956

J. Zack, E. J. Natenberg, S. Young, G. V. Knowe, K. Waight,J. Manobianco, and C. Kamath, “Application of ensemble sensitivity analysis to observation targeting for short term wind speed forecasting in the Washington-Oregon region,” LLNL Technical Report LLNL-TR-458086, October 2010.

LLNL-TR-458086

E. J. Natenberg, J. Zack, S. Young, J. Manobianco, and C. Kamath, ” A new approach using targeted observations to improve short term wind power forecasts,”AWEA WindPower 2010 Conference and Exhibition, Dallas, TX, May 2010.

LLNL-ABS-416821

J. Zack, E. J. Natenberg, S. Young, J. Manobianco, and C. Kamath, “Application of ensemble sensitivity analysis to observational targeting for short term wind speed forecasting,” LLNL Technical Report LLNL-TR-424442, February, 2010.

LLNL-TR-424442

C. Kamath, “Using simple statistical analysis of historical data to understand wind ramp events,” LLNL Technical report LLNL-TR-423242, February 2010

LLNL-TR-423242

E. J. Natenberg, J. Zack, S. Young, R. Torn, J. Manobianco, and C. Kamath, “Application of ensemble sensitivity analysis to observational targeting for wind power forecasting,” 14th Symposium on Integrated Observing and Assimilation Systems for the Atmosphere, Oceans, and Land Surface (IOAS-AOLS), AMS Annual Meeting, Atlanta, January, 2010

LLNL-ABS-415754

C. Kamath, “Understanding wind ramp events through analysis of historical data,” IEEE PES Transmission and Distribution Conference, New Orleans, April 2010.

LLNL-CONF-416432