Strategies for overcoming uncertainties in heat exchanger network synthesis

Abstract

Current grassroot heat exchanger network design techniques involve a three-task procedure: (a) calculation of the minimum utility consumption; (b) selection of a set of process stream matches that satisfy the minimum utility consumption and minimum units criteria; and (c) derivation of a minimum investment cost heat exchanger network configuration. Section 3 of this paper will present an approach for overcoming the uncertainty associated with the network optimization task that arises from the nonconvexities in the network optimization problem. A global optimum search approach is proposed that decomposes the nonconvex network optimization problem into a set of convex subproblems that represent upper and lower bounds and whose solution can lead to the network configuration with the globally minimum investment cost. Uncertainty arises in the selection of matches task when there are many combinations of matches that satisfy the targeting criteria. When there are many such combinations, exhaustive enumeration may be required to determine the optimal combination. Section 4 presents a decomposition methodology circumventing this uncertainty, based upon a proposed hyperstructure that contains all possible network configurations and process stream matches. This hyperstructure is used to derive a mixed integer nonlinear programming (MINLP) formulation that models both the selection of process stream matches and the derivation of a heat exchanger network configuration. A decomposition approach is proposed that efficiently derives the optimal combination of process stream matches and the optimal heat exchanger network configuration.