A PARAMETRIC MINLP ALGORITHM FOR PROCESS SYNTHESIS PROBLEMS UNDER UNCERTAINTY

In this paper, we describe an alogirthm for the parametric solution of MINLP models in the context of process synthesis problems under uncer tainty. The procedure, based on the outerapproximation/equation relaxa tion algorithm, involves the iterative solution of NLP subproblems and a parametric MILP master problem, with which an E-approximate paramet ric solution profile can be obtained which corresponds to the set of o ptimal structures/designs as a function of a scalar uncertain paramete r varying within a closed range. Three example problems are presented in detail to illustrate the steps of the proposed algorithm; its appli cability to address general process synthesis problems under uncertain ty is also briefly discussed.