Improved Modeling Tools Development for High Penetration Solar
- Univ. of California, San Diego, CA (United States)
- Power Analytics Corporation, San Diego, CA (United States)
One of the significant objectives of the High Penetration solar research is to help the DOE understand, anticipate, and minimize grid operation impacts as more solar resources are added to the electric power system. For Task 2.2, an effective, reliable approach to predicting solar energy availability for energy generation forecasts using the University of California, San Diego (UCSD) Sky Imager technology has been demonstrated. Granular cloud and ramp forecasts for the next 5 to 20 minutes over an area of 10 square miles were developed. Sky images taken every 30 seconds are processed to determine cloud locations and cloud motion vectors yielding future cloud shadow locations respective to distributed generation or utility solar power plants in the area. The performance of the method depends on cloud characteristics. On days with more advective cloud conditions, the developed method outperforms persistence forecasts by up to 30% (based on mean absolute error). On days with dynamic conditions, the method performs worse than persistence. Sky Imagers hold promise for ramp forecasting and ramp mitigation in conjunction with inverter controls and energy storage. The pre-commercial Sky Imager solar forecasting algorithm was documented with licensing information and was a Sunshot website highlight.
- Research Organization:
- Univ. of California, San Diego, CA (United States)
- Sponsoring Organization:
- USDOE Office of Energy Efficiency and Renewable Energy (EERE)
- Contributing Organization:
- Power Analytics Corporation, San Diego, CA (United States)
- DOE Contract Number:
- EE0004680
- OSTI ID:
- 1165262
- Report Number(s):
- DOE-UCSD-0004680-1
- Country of Publication:
- United States
- Language:
- English
Similar Records
A Public-Private-Acadmic Partnership to Advance Solar Power Forecasting
A New Approach for Short-Term Solar Radiation Forecasting Using the Estimation of Cloud Fraction and Cloud Albedo