Model predictive on-load tap changer control for high penetrations of PV using high resolution resources assessment with sky imager

A control strategy to reduce OLTC operations resulting from high PV penetration on distribution feeders was developed and applied in this paper. The strategy uses 5-minute-ahead solar forecast to derive future voltage states on the distribution feeder. Unnecessary tap operations (TO) are identified as those that are reversed within 5 minutes, likely because of temporary cloudy or clear conditions over adjacent PV systems. Unnecessary TO are eliminated. On a feeder where TO were abundant at over 750 per day on average, the strategy resulted in the avoidance of 56% of the TO, which would result in significant savings in OLTC maintenance costs. Average extreme voltages were not affected by applying the TO reduction techniques. While daily maximum and minimum voltage excursions over the entire feeder slightly increased on several days, statistical analysis demonstrates that these deviations happened very rarely. The strategy is most effective on partly cloudy days and on voltage regulators with a large number of TO, while the fraction of avoided TO decreased substantially for less than 10 to 40 TO per day. The control algorithm was shown to be robust against forecast errors inherent in state-of-the art sky imager forecasts.