A Statistical Model for Simulating Solar Intensity in New Jersey

Abstract

This thesis is concerned with the construction of a simulator for solar energy generation levels at 24 solar generation plants in New Jersey as well as a comparison of the profitability of grid-level and plant-level storage cells. The simulator is statistical in that it is based off of historical data; specifically the simulator uses ARIMA and/or ARMA time series models to generate solar energy generation quantities. The quality of the simulator is assessed by comparing the cumulative distribution functions of the simulated and the actual total daily solar energy generations. The simulator is then used to compare the profitability of two scenarios for the integration of storage into the electricity grid: the first scenario considered is grid-level storage i.e. a single, massive battery while the second scenario considered is generation plant-level storage i.e. a smaller battery at every solar energy generation plant. The results of the simulation suggest that plant-level storage is a more profitable investment than a single grid-level storage cell.