Electric vehicles (EV) play a crucial role in transforming the transportation sector towards a sustainable, clean, and energy-efficient model. Their adoption leads to a reduction in air and noise pollution and creates an infrastructure framework that supports environmentally friendly transportation. However, with the growing number of EVs, challenges arise regarding the need for development and improvement of the electric infrastructure. In addition to the potential lack of available charging stations, the increasing use of EVs can also pose a problem from the perspective of exploiting the existing electric energy system, where charging a growing number of EVs leads to a proportional increase in the load on the electrical grid. Taking into account the dynamic needs of users and variable supply and demand conditions, charging management is a crucial issue to ensure reliable and efficient delivery of electrical energy to all users. This master's thesis explores solutions for managing EV fleet charging with the aim of optimizing energy consumption to reduce charging costs. Accordingly, aggregate and distributed models of delivery EV fleets are presented, along with their parameters and input distributions. Since the adopted EV fleet models have linear properties, the mixed integer linear programming (MILP) optimization method recognized in the literature was used for offline optimization, as well as for setting the optimal control strategies. Control strategies for EV charging are based on the model predictive control (MPC) algorithm, with two different approaches considered. In this regard, the hierarchical approach is based on optimizing charging at the aggregate model level, after which the optimal value of aggregate power is heuristically allocated to individual EVs. On the other hand, the distributed approach uses optimization directly at the level of individual EVs. Compared with unmanaged charging, where vehicles are always charged with the maximum available power, the final results show that significant savings can be obtained in both ways. In this regard, the hierarchical approach provides slightly worse solutions with reduced computational complexity, while the distributed approach provides a solution equivalent to the global optimum obtained by the offline optimization on the distributed model. Also, scenarios were considered where the charging cost in a given period is optimized over a known profile of electricity price separately, and in combination with economically viable energy from renewable sources.