Improving Mathematical Optimization Techniques with the Aid of Exergy-Based Variables

Abstract

The design optimization of energy conversion plants requires sophisticated optimization techniques. The usefulness of mathematical programming approaches has been discussed in several papers. Usually, the quality of the computed solutions, concerning global optimality and the convergence speed, is not discussed in these papers and even the existence of local optimal solutions is not mentioned. Indeed, the optimization of nonconvex mixed integer non-linear problems (MINLP), such as the structural and design optimization of power plants, is a very difficult problem. However, knowledge of the real optimization potential can assist the design engineer in better understanding the optimization procedure. This article deals with the use of exergetic variables for improving the quality of results obtained from mathematical optimization techniques and their convergence speed. LaGO, the solver used to compute the discussed results, can evaluate the obtained solution of the discussed minimization problems by calculating lower bounds of the original problem based on a relaxed convex objective function. Here, the use of exergetic variables can help to increase the lower bounds significantly and thus, to improve the evaluation of the computed solutions and the convergence speed. The method is applied to different optimization tasks.

• This paper is an updated version of a paper published in the ECOS'08 proceedings. 

Keywords

• cost minimization, design optimization, cogeneration plant, mixed-integer nonlinear programming, exergy, convergence speed