Probabilistic coupled EV-PV hosting capacity analysis in LV networks with spatio-temporal modelling and copula theory

The authors present an innovative approach for probabilistic coupled electric vehicle (EV) and solar photovoltaics (PV) hosting capacity analysis in low-voltage (LV) distribution networks. The challenges posed by system uncertainties and correlations between different parameters, such as PV generation and EV charging demand, are addressed using probabilistic modelling. To appropriately incorporate the geographical distribution and time-variant patterns of EV charging demand, a comprehensive spatio-temporal (ST) model is developed to capture the trip distance, EV arrival, and charging time. The correlation between the PV generation and EV charging demand is effectively captured by copula theory. The proposed models have been validated using actual EV charging and PV generation data from 36 Australian EV users over 1 year. Power flow simulation with actual data and modelled data have identified EV-only and coupled EV-PV hosting capacities in an Australian LV test network. The coupled EV-PV model presents a higher level of accuracy, having an average mean absolute percentage error (MAPE) of 5.97% compared to independent EV profiles having a MAPE of 10.12%. A voltage profile analysis with the EV and PV profiles also validates the same trend, having MAPE of 1.5% and 1.95%, respectively, for coupled EV-PV and independent EV profiles. To appropriately incorporate the geographical distribution and time-variant patterns of EV charging demand, a comprehensive spatio-temporal (ST) model is developed to capture the trip distance, EV arrival, and charging time. The correlation between the PV generation and EV charging demand is effectively captured by copula theory. Power flow simulation with actual data and modelled data have identified EV-only and coupled EV-PV hosting capacities in an Australian LV test network. image