In most of the publications on optimization of energy systems, it is considered that the equipment is available for operation at any instant of time (i.e. it is not subject to failure) except, perhaps, of predetermined periods of maintenance. Thus, it is left to the designer to decide empirically how to provide the system with redundancy, which is necessary in case of equipment failure. However, in this way, the final configuration may not be optimal. In the present work, reliability and availability are introduced in the thermoeconomic model of the system, so that redundancy is embedded in the optimal solution; in addition, more realistic values are obtained for the cost and profit, if any. The state-space method (SSM) of reliability analysis is used. The optimization problem is formulated at two levels: (A) synthesis and design, (B) operation under time-varying conditions. For the solution of the problem at level A and also at level B with no failure, a genetic algorithm coupled with a deterministic one is used. In case of partial failure, the optimization problem is solved by the Intelligent Functional Approach (IFA). The use of IFA combined with SSM is proved to be very efficient for decision making regarding systems under partial failure. It turned out that reliability aspects have a direct and significant impact on the optimal result at each one of the three levels: synthesis, design and operation. (C) 2002 Elsevier Ltd. All rights reserved.