Planning of Chemical Process Networks via Global Concave Minimization

Abstract

The problem of selecting processes and planning expansions of a chemical complex to maximize net present value has been traditionally formulated as a multiperiod, mixed-integer linear program. In this paper, the problem is approached using an entirely continuous model. Compared to previous models, the proposed formulation allows for more general objective functions. In solving the continuous model, minimizing its nonconvex objective function poses the major obstacle. We overcome this obstacle by means of a branch-and-bound global optimization algorithm that exploits the concavity and separability of the objective function and the linearity of the constraint set. The algorithm terminates with the exact global optimum in a finite number of iterations. In addition, computational results demonstrate that the proposed algorithm is very efficient as, for a number of problems from the literature, it outperforms OSL, a popular integer programming package. We also develop a procedure for generating test problems of this kind.