Co-generation systems, together with absorption refrigeration and thermal storage, can result in substantial benefits from the economic, energy and environmental point of view. Optimization of operation of such systems is important as a component of the entire optimization process in pre-construction phases, but also for short-term energy production planning and system control. This paper proposes an approach for operational optimization of energy supply systems with small or medium scale co-generation, additional boilers and heat pumps, absorption and compression refrigeration, thermal energy storage and interconnection to the electric utility grid. In this case, the objective is to minimize annual costs related to the plant operation. The optimization problem is defined as mixed integer nonlinear and solved combining modern stochastic techniques: genetic algorithms and simulated annealing with linear programming using the object oriented “ESO-MS” software solution for simulation and optimization of energy supply systems, developed as a part of this research. This approach is applied to optimize a hypothetical plant that might be used to supply a real residential settlement in Niš, Serbia. Results are compared to the ones obtained after transforming the problem to mixed 0-1 linear and applying the branch and bound method