High frequency irradiance and power output for grid integration and storage evaluations

This paper describes the synthesis of solar photovoltaic output time series for grid integration studies and storage evaluation in locations where direct observations of irradiance are not available. An approach combining the flexibility of a numerical weather prediction model, satellite data, and the accuracy of on-site measurements is used. The method has recently been used to generate data for studies in Hawaii. Modeled data compared well to surface measurements of irradiance and power, however a negative bias in irradiance and positive bias in power was found. Variability at the aggregate of all locations sampled was found to be consistent with measured values. A stochastic-kinematic cloud model was used to simulate cloud movement and create two years of two-second irradiance and power production data in Hawaii. The results suggest the coupled model is able to simulate the behavior of the solar resource as well as its high-frequency fluctuations, or ramps, with acceptable accuracy. The modeled ramp distributions agreed well with the observed distributions for both individual sites and site aggregates at various time intervals. Overall, the modeled data were found to match the statistical characteristics of observed data with acceptable accuracy, and were deemed suitable for use in grid integration studies.