Initial Assessment of Variable Renewable Energy Forecasting Infrastructure

The parameters relate to both solar and wind power forecasting. The USAID Energy Program has applied the current practice of Australian Energy Market Operator (AEMO) and California Independent System Operator (CAISO) on forecasting, using a centralized Variable Renewable Energy (VRE) Forecasting System. Both AEMO and CAISO have developed specific data requirements for VRE forecasting, which cover the provision of static and dynamic data. Static data includes technical specifications of solar and wind farms, geographical and topographical data, while dynamic data includes real-time measurements of power generation and certain meteorological parameters. For both wind and solar power prediction, SCADA instantaneous measurements are required, with values updated at least every 4-10 seconds. The USAID Energy Program has distributed a questionnaire to the operators and developers of VRE projects to gather feedback on data requirements. The questionnaire has been distributed among the developers of 4 wind projects, and the program is currently collecting and analyzing the content of the feedback. Qartli Wind Farm (QWF) and Imereti and Didgori wind power plant projects are the locations where, due to data availability, the forecast of wind power and/or wind speed could be launched in a short period. To check the applicability of meteorological data input derived from NWPs, the USAID Energy Program organized a Working Group (WG) comprising representatives of the National Environmental Agency (NEA), QWF, and Georgian State Electrosystem (GSE). The WG supported the launch of wind speed and direction forecasting in Test Mode. The first results became available from July 12, and the forecasting results in Test Mode could be characterized by both amplitude and phase shifting error. The USAID Energy Program identified the NEA services that could be utilized as an input for the forecasting system, including historical meteorological data, access to observational data proximity to proposed wind projects locations, and sector-specific forecast of meteorological parameters. The program concluded that data input options from direct measurement, including wind and solar power plants' SCADA system and met masts installed in boundaries of wind and solar power plants, are possible. Additionally, data from existing meteorological radar could be utilized as an input for the forecasting system. More than one month of test mode is required to check how precise the forecasting system is, which comes from numerical weather prediction models. The performance assessment, where uncertainty metrics such as Mean Absolute Error (MAE) and Route Mean Square Error (RMSE) need to be applied, are planned to be performed by the middle of September.