The paper deals with charging optimization for delivery electric vehicle fleets based on dynamic programming method. Charging of each individual vehicle within the fleet is optimized separately, thus providing globally optimal solution on the vehicle level. By posing an upper constraint on the grid power used for charging, the individual charging optimizations are coupled together on the fleet level in a suboptimal way. Consecutive optimizations for each vehicle within the fleet are conducted for different orders of individual vehicle optimizations. In this way the sensitivity of optimization results with respect to ordering of charging optimizations can be analysed and a solution closer to global optimum can be found. The obtained optimization results are used for the purpose of validation of previously developed aggregate battery models and corresponding heuristic method dealing with distribution of optimal aggregate power over individual vehicles.