This paper presents a novel nonlinear charge-based aggregate battery model representing the fleet of electric vehicles (EV) connected to a grid. The proposed aggregate battery model is parameterised and validated against a more accurate, but more complex individual battery-based fleet model where each EV is modelled separately as a charge storage. Also, the novel nonlinear model is compared with the linear model commonly used in energy planning and charging management studies, where the battery is modelled as energy storage. The functionality of the novel aggregate battery model is demonstrated through a case study related to a delivery vehicle fleet for which a comprehensive set of recorded driving cycle data is available. The charging management optimisations of the presented battery models are conducted by using the dynamic programming approach, and the comparative analyses of the optimal charging patterns are given.