MINLP optimization of plate fin heat exchangers

Plate Fin heat Exchanger (PFHE) design is a very complex task. In most cases, heuristic-based procedures are used. In order to improve the company profits, the PFHE design problem is stated according to mathematical programming techniques. First of all, objective functions such as manufacturing cost, physical volume are detailed as well as operating and manufacturing constraints. Finally, optimization variables including the geometrical fin parameters are described. Since most of the geometrical parameters of the exchanger (core number, geometrical fin parameters, etc) have discrete values, this formulation results in a Mixed Integer Non Linear Programming (MINLP) problem. Different solution strategies are discussed. For example, the solution of the relaxed problem using a Successive Quadratic Programming (SQP) algorithm. Another example is the solution of the original MINLP problem using Simulated Annealing (SA) or Branch and Bound (BB) algorithms. The efficiency of the developed tool is illustrated by two industrial case studies: the manufacturing cost reduction is greater than 10%.