Systems-based modeling of generation variability under alternate geographic configurations of photovoltaic (PV) installations in Virginia

Abstract

With increased focus on renewable energy in our modern era, it is increasingly important to understand the impact of policies on the performance and reliability of regional energy systems. This research develops a model to understand how geographic dispersion of PV installations impacts the reliability of electricity generated from the total PV network, measured by the variance of the distribution of generated electricity. Using NREL data, beta probability distributions of sunlight (kWh/m 2/day) in various regions of Virginia are estimated using a fitting method that minimizes the Kolmogorov–Smirnov test statistic. A Monte Carlo simulation model is developed to measure PV electricity generation from multiple centralized and dispersed configurations over 100,000 days of probabilistic sunlight. There is a calculable tradeoff between average generation and generation variability, and increased geographic dispersion of PV installations can decrease this variability. Controlling variable generation through policies that promote efficient PV siting can help provide reliable power, minimizing the need for load-balancing peaking power infrastructure and costly electricity purchases from the grid. Using a tradeoff framework of generation and costs, this paper shows that geographically dispersed generation can mitigate the risk of unreliable solar generation that can significantly impact the end-user costs and make PV infrastructure unattractive.