Project Name: Coordinated Ramping Product and Regulation Reserve Procurements in CAISO and MISO using Multi-Scale Probabilistic Solar Power Forecasts (Pro2R)
Funding Opportunity: Solar Forecasting 2
SETO Subprogram: Systems Integration
Location: Baltimore, MD
SETO Award Amount: $1,738,613
Awardee Cost Share:  $434,668
Principal Investigator: Benjamin F. Hobbs

In collaboration with the National Renewable Energy Laboratory, IBM, the University of Texas, Dallas, and the California and Mid-Continent Independent System Operators (ISOs), this project will advance the state-of-the-art in integrating solar power forecasts into ISO operations. The team will use probabilistic solar power forecasts and the consequent net-load ramp forecasts to dynamically procure two types of operating reserves—ramp products and regulation services—which are necessary for the reliable operation of the electric grid. Developing a data-driven platform for probabilistic solar forecasting will enable ISO’s to better integrate and use the new forecasts. The team will test the integration of the forecasts with an offline version of the ISO’s operation environment in order to estimate cost reductions and reliability improvements.

APPROACH

Improvements to IBM’s Watt-Sun probabilistic solar power forecasting system and platform—which is driven by machine learning—will produce both probabilistic solar power and ramp forecasts. This big data platform will make these forecasts available to ISOs and a validation team in the Solar Forecasting 2 program. Specialized procedures, like optimized swinging door algorithms, will enable net-load ramp forecasts to be extracted from solar power forecasts. The resulting probabilistic forecasts will be integrated into ISO procedures for procuring reserve generation in the energy markets to fulfill scheduling requirements. This includes decisions for both short-term and long-term dispatching of generation units. Near real-time visualizations of probabilistic ramp forecasts, along with ramp alerts, will also be developed to provide operators with improved situational awareness.

INNOVATIONS

The project will utilize high performance computing and IBM’s Physical Analytics Integrated Data Repository and Services to accommodate and organize the large amount of generated solar forecasting data. The fusion of disparate data resources, including high-resolution outputs from numerical weather predictions, as well as application of deep-learning techniques, will enable the situation-dependent blending of models. Accuracy of forecasts under challenging conditions, including high-ramp days, will be improved. This will be the first-of-its-kind integration of advanced probabilistic forecasts of solar power and the resulting net-load for the estimation of ramping and regulation requirements.