Modelling electric vehicles and setting charging patterns endogenously in energy planning models
Keywords: electric vehicle, charging, endogenously, energy planning, MESSAGE
AbstractIn the UN Emissions Gap Report it is estimated that the action plan agreed in the Paris Climate Agreement is not sufficient to limit the global temperature increase to below 2°C, compared to the pre-industrial period. More stringent measures must be taken to reduce global warming. In order to evaluate the most suitable pathways to decarbonize economies, complex multisectoral models should be used. Electric vehicles have the potential to significantly reduce carbon emissions in the transport sector. However, high penetration of electric vehicles might affect the development of the power sector. Additional energy required to charge these vehicles increases overall electricity demand. Thus, even though a higher share of electric vehicles reduces emissions in the transport sector, overall emission reduction effect depends on the fuels used to generate this additional electricity for charging. In case of fossil fuels, overall emissions might even increase. On the other hand, it is possible to adjust charging patterns according to generation fluctuations in wind and solar power plants. Such energy balancing could allow higher penetration of intermittent renewables. Because of this interlinkage of power and transport sectors through electric vehicles it is beneficial to model both sectors simultaneously in energy planning models, especially when the purpose of the model is to evaluate possible emission reduction pathways. In this paper, a methodology is proposed on how to model electric vehicles in energy planning models. This methodology enables flexible charging of electric vehicles, where charging patterns are set internally by the model. The methodology is based on the modelling of different driving patterns and evaluation of different vehicle states. Furthermore, it is explained how driving patterns and vehicle states can be derived from limited data by using pattern approximation with normal distributions.