From the standpoint of electric grid, electric vehicles (EV) represent a spatially-distributed energy storage, which can be used for different purposes within a smart charging framework, such as load levelling of electric grid. In order to evaluate benefits of replacing conventional vehicles with electric ones from the standpoint of EV-grid integration, the crucial step is to conduct charging optimisation based on accurate and relatively simple models of EV fleets, which represents the main aim of this thesis. In the thesis, a novel aggregate model of EV fleet is proposed and validated with respect to more precise, but also computationally less efficient distributed EV fleet model. Charging optimisation for each model is conducted by using dynamic programming (DP) algorithm, which guarantees globally optimal results for a general nonlinear optimisation problem. The aim of optimisation is to minimise the cost of electric energy used for EV fleet charging, while satisfying the constraints related to the battery state-of-charge (SoC) and the charging power. In addition, the single-level DP optimisation of aggregate battery charging power is extended to a bi-level optimisation, where the maximum charging power and the SoC-at-departure of an individual EV are optimised at the supervisory level by using a genetic algorithm, with the aim to minimise the operational cost. For the purpose of parameterisation and validation of the proposed aggregate EV fleet model, an extended range electric vehicle (EREV) and a corresponding powertrain control strategy are developed. The EREV control strategy is based on a rule-based controller (RB) which is combined with an equivalent consumption minimisation strategy (ECMS). The SoC sustainability is achieved through SoC control implemented within the RB controller, while the optimality of engine operating points is achieved through ECMS control. The developed control strategy is evaluated by means of computer simulations and by comparing the simulation results with the globally optimally ones obtained by using DP optimisation of powertrain control variables for the case of different driving cycles. Also, data related to driving cycles of the particular delivery vehicle fleet and to the energy system of the corresponding distribution centre are collected and analysed. For the purpose of fleet models parameterisation, the simulations of the developed EREV model are conducted over the recorded driving cycles, as well as over the synthetic driving cycles which represent the set of recorded driving cycles in a statistically reliable way. Finally, based on the developed methods and numerical tools, a pilot study related to the analysis of replacing current conventional delivery vehicles with electric ones is conducted for different scenarios of the share of renewable energy sources and different electricity price models.