The introduction of new low capacity absorption chillers opens new prospects for the multifunction multisource solar systems (MFSSYS) which exploit the full potential of the solar energy for heating, cooling and production of domestic hot water (DHW) purposes. The solar combisystems (SCS), which preceded the MFSSYS, lacked an adapted procedure for the sizing of their main components: the solar collector and the solar thermal storage tank. From the point of view of the energy and investment cost, a sizing method based on the peak load may fail if the energy source is not guaranteed to be stable. An optimization of the main components by simulation may be then a key solution for an optimal sizing of the MFSSYS. An initial hydraulic schematic is elaborated for the MFSSYS based on information found in the literature about the SCS and the data made available by the chiller manufacture. The modeling of this schematic is complex as redundant simulation and modification were necessary in order to eliminate the potential problems of solution convergence. From this experience, a method of simulation aided design is elaborated. Parametric runs were carried out on the MFSSYS model. They offer needed information for the comparison of the optimization algorithms which are tested later on. The outcome of these parametric runs is a response surface which represents the cost of the MFSSYS as a function of the solar collector surface area and the volume of the solar thermal storage tank. In order to reduce the number of simulations required by a complete parametric runs method, the use of optimization algorithm become a necessity. An optimization algorithm based on the design of experiments (OptDOE) is developed; its performance is compared with the one of a hybrid optimization algorithm in two cases: a reference function of Rosenbrock and the model of the MFSSYS. Compared to the hybrid optimization algorithm, OptDOE has showed good performance. The number of simulations is reduced and the optimized values, found by this method, are close to those of the parametric runs. The main advantage of OptDOE is to describe the behavior of the cost function in the neighborhood of the optimum. This information is valuable especially when the cost function has a valley-like form, which is the case for the systems we studied. In this case, the cost has approximately the same value for a large variation range of the optimized parameters.
