Integration of wind and solar under a 20% RPS: Stochastic simulation methods and results from California ISO studies

This paper presents two extension to the literature on modeling of wind and solar integration in the California ISO footprint, as well as results from a large-scale regional simulation [2]. First, the paper presents a two-step methodology for simulation of power systems with high penetration of variable generation resources, such as wind and solar. The first step is a statistical model that calculates hourly and sub-hourly operational requirements for the incremental load-following and regulation capacity needed for integration of variable generation resources, developed by [1], [10]. The second step is to add one or more of these capacity reserve requirements to a production simulation model with co-optimization of energy and ancillary services to test the capability of the power system to provide operational capabilities. Both steps start from the same set of 1-minute data on load and wind and solar production as well as forecast errors for modeling to ensure consistency of simulated results. The second extension was to develop a sequential production simulation model that can account for the effect of forecast uncertainty on day-ahead to real-time unit commitment and dispatch over a large number of iterations. This approach was applied to selected days. The simulation begins with unit commitment against day-ahead and then hour-ahead “net” loads that account for load, wind and solar forecast errors, before conducting 5-minute economic dispatch in real-time. The purpose is to evaluate the effect of forecast uncertainty on current ISO market scheduling practices and the capability to conduct real-time net load-following. This is a topic of concern to ISOs when evaluating changes to market products and scheduling practices to elicit further operational capability. Along with additional annual production simulations that examined impacts to fleet operations and revenues, the finding in this study, under the stated assumptions and solution methods, was- that operational flexibility was sufficient to meet intra-hourly and inter-hourly ramps while carrying all needed ancillary services, including additional regulation calculated in the first step. This general modeling framework has been extended further in other recent studies [3].