Jan 9, 2019
Published in: Journal of Renewable and Sustainable Energy, Volume 10 (2018). doi: 10.1063/1.5038028
Posted on RAND.org on November 28, 2018
This article was published outside of RAND. The full text of the article can be found at the link above.
We examined the geographic smoothing of solar photovoltaic generation from 15 utility-scale plants in California, Nevada, and Arizona and from 19 commercial building installations in California. This is the first comparison of geographic smoothing from utility-scale and building-mounted PV and the first examination of solar PV smoothing in this region. Our research questions were (1) how does geographic smoothing of commercial building rooftop PV compare to that of utility scale PV?, (2) is the geographic smoothing found for utility-scale plants the same for the western US as in India?, and (3) how does the geographic smoothing for PV compare to that of wind? By examining the power output of these generators in the frequency domain, we quantified the smoothing obtained by combining the output of geographically separated plants. We found that utility-scale and commercial rooftop plants exhibited similar geographic smoothing, with 10 combined plants reducing the amplitude of fluctuations at 1 h to 18%–28% of those seen for a single plant. We find that combining a few PV sites together reduces fluctuations, but that the point of quickly diminishing returns is reached after [approximately] 5 sites, and that for all the locations and plant sizes considered, PV does not exhibit as much geographic smoothing as is seen for combining wind plants. We present preliminary theoretical arguments for why geographic smoothing of PV plants is less effective than that for wind plants. The slope of the high-frequency part of the PV power spectrum can at best be geographically smoothed (steepen) to an asymptotic spectrum of f-2. This limit for PV has considerably less smoothing than that for wind's geographic smoothing, shown theoretically and from observed data to be f-2.33.