Modelling, simulation and design of Electric Car Sharing systems with eC2S

The goal of this Master thesis is the implementation of DECASIM: a data-driven, discrete-event simulator for electric car sharing systems in Smart Cities. DECASIM is able to model car sharing demand from data coming from real car sharing systems in 25 cities, and create parametrised simulation scenarios in order to compare different fleet management strategies. The main focus of this thesis is given to charging strategies, which are critical for the correct implementation, deployment and functioning of ECS (Electric Car Sharing) systems. Input data have been collected during 2017 using the software UMAP (Urban Mobility Analysis Platform) and consist in a trace of real origin-destination bookings done by users. These data are used to create a configurable demand model for ECS which represent the main input of the simulator. The model is composed by a parametric estimation of bookings' inter-arrival times based on Poisson processes, and a non-parametric estimation of OD (Origin-Destination) distribution based on Kernel Density Estimate. We performed a quantitative validation of this approach for a more precise assessment about its pros and cons. The demand model is used as the main input of simulator, together with static parameters such as: fleet size, booking requests' arrival rate and number of charging poles. Other more complex parameters are meant to simulate different poles placement policies and different charging strategies. We ran a simulation campaign over 4 cities in order to get insights about charging from the points of view of system operators, users and city planners. Results show that users' contribution in charging processes might be essential to improve the general efficiency of the system and to reduce the cost of charging management for systems operators.