Improved optimization strategies for automated heat exchanger network synthesis through physical insights

Abstract

A heuristic approach to improving current automatic heat exchanger network synthesis (HENS) methods is presented, where the various sets of matches that all achieve the minimum or a specified number of units are rated according to their potential for being able to transfer heat vertically between the composite curves, and thus give minimum total area in the corresponding network. For cases where the film heat transfer coefficients for the streams are significantly different, the requirement of strict vertical heat transfer should be relaxed, and this is done in the model by introducing stream individual contributions to Δ T min. Most published work on HENS to date, including the pinch design method, use a single global minimum approach temperature (Δ T min), not only to set the level of heat recovery, but also as a basic design parameter for the heat exchanger network. It is shown in this paper that relaxing this assumption by using the heat recovery approach temperature (HRAT) only to set the level of heat recovery and allow the exchanger minimum approach temperature to be set at any value (EMAT < HRAT) when designing the heat exchanger network, will result in designs that have lower total annual cost, use less total area for a given (small) number of units and/or require fewer number of units. This is the case even when the streams have identical heat transfer coefficients. Reduced network complexity is another benefit. A new procedure for HENS is presented that uses the heat recovery approach temperature (HRAT) and the number of units ( U) as the two basic optimization variables. The set of matches is found by a new vertical MILP transshipment model, where the exchanger minimum approach temperature (EMAT) can either be completely relaxed (zero) or set to a lower value by the engineer.