Mathematicians and engineers have developed in a joint research project a solution approach for performing simultaneous structural and design variable optimization in the design of cost-effective complex energy conversion systems. The paper presents a methodology and an application to the design of a combined-cycle power plant that provides fixed amounts of electricity and steam for a paper factory. A superstructure that has embedded several potential configurations of such a cogeneration system is used to minimize the total cost of the plant products. The design problem is formulated as a nonconvex mixed-integer nonlinear program (MINLP), and solved via a rounding heuristic based on an automatically generated convex relaxation of the given problem. Utilities and the process industry are expected to benefit from such a MINLP optimization technique. (C) 2005 Elsevier Ltd. All rights reserved.