WindNET: An advanced wind sensor network to improve short-range wind forecasts for electric utility dispatch and operation
The Hawaiian utilities' current experience and need for improved short-term forecasts, in particular detection of variability and ramp events, is consistent with requirements identified by national and international operators with wind plant experience. Increasing levels of wind penetration on the electric grid in regions like California, Texas, New York, and the Pacific Northwest is driving improvements on predictive models, correlation between grid sensitive periods and wind-driven events, and better integration of results in the control rooms to inform real-time operations, intra-hour market redispatch, and balancing needs. This intra-hour and near “real-time” need is currently not being met by presently available forecasting services and will require additional model enhancements coupled with real-time data.
As part of a multi-state effort sponsored by the U.S. Department of Energy (DOE), efforts are underway with Hawaii utilities as part of the Hawaii Utility Integration for wind to develop better ramp event forecasting tools for Hawaii. In conjunction with the National Renewable Energy Laboratory (NREL) under the Hawaii Clean Energy Initiative (HCEI) and Lawrence Livermore National Laboratory (LLNL) under the WindSENSE program, national laboratory resources are being leveraged to model and characterize wind trends and identify ramp event indicators. The Hawaiian Electric Companies are actively working with industry wind forecasting vendors to investigate control room improvements with enhanced forecasting capability to help manage wind plants in Hawaii. Active participation by utilities is necessary as data must be integrated appropriately into operations, dispatch, and management activities (e.g. energy management systems).
Hawaii utilities are interested in wind sensor networks comprised of in situ and remote monitoring devices that can be deployed at strategic locations to improve the predictability of near-term wind power changes. The specific objective of the WindNET project is to test the concept of deploying such a sensor network to improve the predictability of large wind ramp events in the 0- to 6-hour ahead time frame. The sensor deployment strategy is based on guidance generated using targeted observing techniques developed previously for the LLNL and NREL projects. The techniques included a physics-based atmospheric model approach called ensemble sensitivity analysis (ESA) as well as diagnostic case analyses. The case studies identified physical processes responsible for a diverse set of ramp events and subjectively inferred the type and location of measurements that would likely improve the short-term forecast of these events.
The NREL targeted observation study was conducted for three operational wind farms in the Hawaiian Islands and the results used to formulate sensor deployment plans for each site. Due to resource limitations, the first phase of the WindNET focuses a wind farm at the southern tip of the Big Island (Hawaii). Three sodars and one radiometer will be deployed by September 2010 to make measurements for a 6- to 9-month period through the spring of 2011.
A set of experimental wind power production forecasts will be generated with and without the additional sodar and radiometer data. Results will be used to determine the impact of the new field data on ramp event forecasts and prediction of power production over high-resolution intervals (e.g. 15 minutes) for very-short look-ahead (0- to 3-hour) periods.
The conference presentation will provide an overview of the WindNET project in Hawaii including the instrumentation monitoring and site selection process, field deployment campaign, data analysis, and preliminary results from the forecast experiments. The presentation will also address instrumentation cost, power and maintenance requirements, height, range, vertical resolution, and communication capabilities that are critical for eventual operational use of such equipment by electric utilities to improve short-term wind forecasts.