Estimation of charging demand for electric vehicles by discrete choice models and numerical simulations: application to a case study in Turin

The electrification of vehicles is one of the recent trends of development for our society to mitigate the problem of air pollution, the critical issues related to climate change and meet the new user needs. Many fields of study and research are involved in this theme, ranging from technological areas to statistical and modelling of transport systems for the management of electric vehicles (EV) and their operations.This study is focused on modelling and simulation of EVs user behaviour to forecast possible electric charging scenarios in cities and understand potential management problems, as well as the room for improvements of EVs and related infrastructures systems. Indeed, many factors may prevent a wide diffusion of electric vehicles. In order to analyse these issues, the case study of conurbation of Turin was selected, to reproduce realistic scenarios applying discrete choice modelling based on socio-economic and transport system data. One of the specific objectives of the study is to identify user’s charging behaviour from a geographic point of view, to model where users prefer to charge in the study area according to the variables that may affect decisions. On the other hand, the estimation of electric vehicles in cities and the characteristics of their user is helpful to complete the picture on electric mobility. Analysing these behavioural issues in a modelling framework can provide a set of tools to realise which are the improvements and the modifications to pursue and facilitate the diffusion of electric vehicles providing an adequate charging infrastructure to users. To perform the analysis, two available models were considered from the scientific literature: (1) to predict the charging demand and applied in Amsterdam districts and (2) to estimate the EVs penetration rate, as considered in a test applied in USA. The first model works using the methodology of Discrete Choice Modelling based on Random Utility Theory and in particular the multinomial logit and nested logit. The independent variables considered includes households dataset and vehicle-related dataset aggregated at zonal level. Its main output is the estimation of the charging demand distribution in the study area. The second model works by using the linear regression technique and requires a disaggregate dataset for individuals to estimate the penetration rate of EVs for the area. Since zonal datasets were available, for the penetration rate model, a Monte Carlo simulation was performed to manage individual data from a zone-aggregated dataset. The two models are applied in an integrated framework to Turin study area and various scenarios are generated. Results show which are the zones in which higher electric charging demand is expected and the zones where the penetration rate of EVs is important. Also, some experimental scenarios are presented to understand which are the most influencing factors on the results and the impact of their changes on the charging demand in the various zones of the conurbation. Observing the model results, the demand is estimated concentred in those zones presenting adequate charging infrastructures supply, a significative number of attraction poles and a particular combination of sociodemographic characteristics.