A model-based optimization of the charging process of batteries in small satellites is proposed. The battery is considered as part of an electrical power system with a peak power tracker architecture and an unregulated bus. Real data are used to determine the parameters of the dynamic model of the system. The battery charging optimization is obtained through the minimization of a cost function which weights charging time, energy losses and temperature variations by taking into account also the dynamic model. Simulation results show the validity of the proposed solution in terms of performance and computational effort for different missions power load, irradiance and temperature operating conditions.