UC San Diego

Solar energy is a promising renewable resource, but the variability of solar photovoltaic (PV) power due to cloud- caused fluctuations is a concern for electric grid operators. Fortunately, though, the relative variability of a PV powerplant will be reduced due to the geographic diversity within the plant. The amount of the variability reduction (VR) will depend on the geometry of the plant, the timescale of interest, and the local meteorological conditions. This work focuses on quantifying and modeling the VR in scaling up from an irradiance point sensor to a PV powerplant. The dependence of VR on timescale is demonstrated using high-frequency data and the wavelet transform. The wavelet variability model (WVM), which simulates the variability of a PV powerplant by estimating the VR at each timescale, is developed and validated. Inputs to the WVM are a point sensor irradiance timeseries, the powerplant layout, and the cloud speed. As an example application, the WVM is used to simulate the numbers of ramps larger than 10% of capacity per minute at various sizes of PV powerplants in Puerto Rico