Effects of Solar Resource Sampling Rate and Averaging Interval on Hourly Modeling Errors

Solar energy modeling errors due to time-averaged hourly inputs are significant where solar resource variability and inverter loading ratio are both high. However, predictions of photovoltaic (PV) system performance are most frequently made with hourly solar resource inputs, typically computed from satellite data obtained every 15 or 30 min. Therefore, we studied the effects of solar resource sampling rate and time-averaging interval on hourly modeling errors by using high-frequency measurements from eight different locations across the United States. When we selected minute-average measurements at various sampling rates and averaged them to hourly data, we observed increasing modeling errors for sampling rates 30 min or shorter. At a 30-min sampling rate averaged hourly, we observed an error that was 50% of 1-min samples averaged hourly. As sampling rate approached 60 min, modeling errors decreased, partially canceling out due to the randomness of the low-frequency sampled data. We examined PV systems with dc-ac ratios > 1.3 and observed that clipping errors dominated modeling errors from other sources like transposition to plane-of-array irradiance at sites with greater solar variability. Based on our analysis, we recommend that an hourly modeling correction be applied whenever hourly inputs are used, especially at sites with high solar variability and dc-ac ratios greater than one.