Two-level optimization algorithm for heat exchanger networks including pressure drop considerations

A mixed-integer nonlinear programming (MINLP) algorithm to account for the effects of the allowable pressure drops for process streams on the structure and cost of heat exchanger networks is presented. Because MINLP models for large problems typically show significant convergence problems, the synthesis model is decomposed into two levels. For a given energy recovery level, an MINLP model is formulated in an inner loop using assumed values for the film heat-transfer coefficients for each stream. In the outer loop, with the use of the areas provided by the MINLP model and thermal-hydraulic models that relate the exchanger area, pressure drops, and film coefficients, updated values of the film coefficients are calculated for each stream according to the specified values for allowable pressure drops. The algorithm provides a network that makes use of the allowable pressure drops specified for the process streams at the total minimum yearly cost. The algorithm shows excellent convergence properties, can incorporate restrictions on matches between streams, and is appropriate for problems with significant differences in the values of film coefficients. Three case studies are used to show the application of the proposed method.