Sensitivity of Resource Adequacy to Sample Error in Systems with Dependent Renewable Generation

Weather-dependent renewable generation creates new challenges in generation capacity investment planning models and system adequacy studies. The statistical interdependence of renewable generation and customer demand for electricity means that more data are needed than previously necessary for systems based on thermal generators with independent outages. This paper investigates how relationships between load, wind, and solar profiles, together with sample error due to using too few years of data, can distort optimal capacity mix, system cost and system reliability. An optimization model is used to compare results from analyses using perfect information (ten years of hourly matched data) to results when different subsamples of the overall data set are used to determine the capacity investment mix. When computational time requires data reductions, although single years can vary significantly, choosing the best single year of actual matched profile data has significantly less distortion than using typical data sets based on longer historical data averaging.