Load commitment of distribution grid with high penetration of photovoltaics (PV) using hybrid series-parallel prediction algorithm and storage

• This paper presents two load forecasting methods for a day ahead and 20 min ahead for distribution system load with high fluctuations of distributed PV generation.
• BESS is used to shave the peak load or follow a reference load curve.
• Load curve smoothing and peak shaving can be performed simultaneously using the proposed short term forecasting methods.

Battery energy storage system (BESS) is one of the promising solutions to deal with intermittency of renewable generation. In this paper, BESS is used for peak shaving and smoothing the distribution load curve of an actual circuit in island of Maui in Hawaii. The distribution circuit has about 850 kW of installed rooftop PV generation. This amount of PV and future expansion, raises some concerns about potential impacts on the transmission system. This paper aims to mitigate these effects. First of all, two load forecasting methods are presented. The forecast load data is then used to control BESS for two main purposes, peak shaving and smoothing. To achieve these goals, two approaches are explained. In approach I, a nonlinear programming method is utilized and equations for simultaneous load shifting and smoothing are derived. In the next approach, a real time control is developed which performs smoothing and peak shaving simultaneously. These methods are applied on 108 days of historical data and pros and cons of each approach is discussed.