The increasing penetration of distributed generation raises concerns to the optimal use of the available resources. This paper proposes a robust optimal power flow (OPF) which determines the optimal reactive power such that power losses are minimized. A model-based evolutionary strategy was used to minimize the system power losses. In contradistinction to conventional evolutionary strategies, the proposed model-based algorithm gives additional information about relationship between variables since a probabilistic model is constructed in each iteration. In addition it allows to be modified in order to include a robustness feature and to improve the efficiency for real time operation. The methodology is applied to the IEEE 37-bus test system distribution network. Small variation in the system parameters and demands are included in order to test its robustness. Results demonstrate the significant impact of time-variation on power losses and highlights the benefits of Smart-Grid strategies for both loss minimization and renewable penetrations.