Abstract for: Know Noise: Realistic Random Processes for Simulations
When the integration of renewable electrical generation with a distribution network is modeled, the random aspects of wind and solar power generation must be modeled correctly to make appropriate investment decisions, and to assure reliability. Supply must balance demand on a timescale of much less than one hour, and yet fluctuations in solar and wind can have correlations with timescales exceeding one month. As an example, we analyze hourly generation and demand in the Midcontinent Independent System Operator (MISO) region. After the periodic components are subtracted, solar photovoltaic can be modeled with a fractional Gaussian process, and wind generation and demand can be modeled with Markov– Gauss processes with timescales of one week and one month, respectively. We show the first- order “pink noise” approximation may produce misleading results. Supplemental material provides Vensim code for fractional Gaussian and fractional Brownian processes.