Pareto Optimal Solutions in Process Synthesis Problems

K.-M. Björk and T. Westerlund (Finland)


MINLP, Global Optimization, Pareto Optimal Solutions, Heat Exchanger Network Synthesis


Mathematical optimization is frequently used in process synthesis problems. The models are often complex, but still they are simplified in many ways. The annuity factor is very often used to express the trade-off between investment and operational costs. Even with a global optimization approach, it is not always possible to find all good solutions in an optimization problem only by changing the annuity factor. Pareto curves, found in multi-objective optimization, describe the trade-off much better and might therefore be of interest in process synthesis optimization, especially when a number of different solutions must be extracted. The solutions from the optimization problem might form the suggestions for the designing engineer, from which the best is to be selected according to criterions not necessarily found in the optimization model. In this paper, it will be shown that, in some examples, it is not possible to find all interesting structures only by changing the annuity, due to binary variables found in the models. Examples are given from the field of process synthesis and heat exchanger network synthesis along with a brief review of the use of multi-criterion optimization.

Important Links:

Go Back