Determining the Power and Energy Capacity of a Battery Energy Storage System Utilizing a Smoothing Feeder Profile to Accommodate High Photovoltaic Penetration on a Distribution Feeder

Abstract

Electricity is a perishable commodity; once it is generated it needs to be consumed or stored. Electric energy storage provides both power and energy capacity. Power capacity applications reduce the need for generation, while energy capacity allows for energy consumption to be decoupled from generation. Previous research was done to develop an algorithm for determining the power (MW) and energy (MWh) capacities of a battery energy storage system (BESS) to mitigate the adverse impacts of high levels of photovoltaic (PV) generation. The algorithm used a flat feeder profile, and its performance was demonstrated on the equinoxes and solstices. Managing feeder power leads to fewer voltage fluctuations along the length of the feeder, potentially mitigating load management issues caused by variability of renewable generation and load profile. These issues include lighting flicker, compressor seizing, equipment shut-off, loss of motor torque (τ ∝ V), frequent transformer tap changes and even voltage collapse due to loss of reactive power support (Q ∝ V). The research described in this thesis builds on this algorithm by incorporating a smoothed feeder profile and testing it over an entire year. Incorporating a smoothing function reduces the requisite BESS energy capacity necessary to provide firming and shaping to accommodate the stochastic nature of PV. Specifically, this method is used to conduct a year-long study on a per second basis, as well as a one-minute basis, for a distribution feeder. Statistical i analytical methods were performed to develop recommendations for appropriately sizing the BESS. This method may be used to determine the amount of PV generation that could be installed on a distribution feeder with a minimal investment in the BESS power and energy capacities that would be required to manage the distribution feeder power. Results are presented for PV penetration levels of 10%-50% of the distribution feeder capacity and show that the use of a smooth feeder profile reduces the required energy capacity by a minimum factor of 10 when compared to a flat feeder profile. Results indicated that it is sufficient to have a one-minute sampling rate, as it provides the necessary granularity to model cloud-induced fluctuations. This method can be applied to any distribution feeder where a load profile and a PV profile are available.