On many days, California now generates more than half of our electricity from renewable energy sources, especially solar and wind farms. Renewables are intrinsically weather-driven, introducing major new uncertainties into the daily balancing of grid load and power generation. Large deviations between forecasted and realized renewable production are frequent and impose significant additional cost on running the grid. I will discuss statistical frameworks for quantifying these costs, and our ongoing work on developing risk allocation methods for fairly ascribing these costs to market participants. To do so, I will review probabilistic forecasting of power generation, and then move to the recently developed stochastic simulation platform for generating hourly joint scenarios across hundreds of renewable assets. I will conclude with our initial results on risk allocation and the relative riskiness of different asset types.
Mike Ludkovski is Professor and Chair at the Department of Statistics and Applied Probability at University of California Santa Barbara where he co-directs the Center for Financial Mathematics and Actuarial Research. Among his research interests are renewable energy markets, stochastic control, and longevity modeling. His work is supported by multiple NSF grants and the ARPA-E PERFORM program. He has worked in the area of stochastic models for energy and commodity markets for 15+ years, including co-editing a Springer volume on “Commodities, Energy and Environmental Finance”. This summer he will be lecturing on this topic at the SIAM Gene Golub Summer School in L'Aquila, Italy.