CONICET | Buscador de Institutos y Recursos Humanos

In this work an optimization based approach is presented to solve the input-output pairing problem for industrial processes. This kind of problems grow in complexity quickly with the process dimension and interaction. Indeed, for a $n \times n$ plant there are $n!$ alternative pairings to be examined. In this case, an alternative formulation is suggested which utilizes the simple and powerful philosophy of the classical relative gain array (RGA) approach within a deterministic MILP formulation in the GAMS environment. This representation favors the systematic evaluation, minimizes the heuristic consideration and could be easily incorporated into more complex problems such as the plant-wide control designs. Furthermore, a comparison with a stochastic global search approach such as genetic algorithms (GA) is proposed and applied on academic as well as industrial cases.