The growing popularity of Electric Vehicles (EV) as an alternative to fossil-fuel-driven vehicles has immense environmental appeal. Considering the effects on distribution networks, which were not designed to support such loads, several challenges are bound to be encountered in a future with purely EVs. One of such technical challenges is the effect of charging several EVs at the same time on distribution network voltage. While coordinated charging is one solution, reactive power compensation can be used to support voltage at the point of connection without the need for a centralised control. This paper explores the feasibility of using installed photovoltaic (PV) inverters as voltage compensation devices in Low Voltage (LV) distribution networks. A reactive power controller was developed in Simulink for PV inverters. The case study of a UK LV network for the winter season was used to investigate the feasibility. Results using a cumulative under-voltage index (CUVi) developed to quantify the contributions of the PV inverter reactive power compensation to network voltage support shows that for up to 30% EV penetration, the available PV capacity alone can completely eliminate under-voltage incidents.

Abstract-Load profiles are indispensable in the decision making process of power transmission and distribution companies. Increasing levels of customer-side renewable generation and electric transport will alter the nature of load profiles significantly. Traditional methods relying on historical data will not be suitable for modelling the increasingly complex power networks of the future. In this paper the feasibility of synthesising future load profiles under increasing levels of photovoltaic (PV) generation and electric vehicles (EV) is investigated using an artificial neural network (ANN) based method, trained with publically available data. The performance of the proposed method is evaluated by using a case study developed for a targeted region in the UK. A comparison of results from the ANN model against those using Multiple Linear Regression (MLR) demonstrates the superior performance of ANN over MLR as well as proves the viability of ANN to synthesise future load profiles.