Surrogate-Based Process Synthesis

摘要

Superstructure optimization-based process synthesis is generally regarded asrntheoretically powerful; however, it has not been widely used in practice since itrntypically results in large-scale non-convex Mixed-Integer Non-Linear Programsrn(MINLP) which are very hard to solve effectively. To address this limitation, wernpropose a framework leading to substantially simpler formulations through thernreplacement of complex first-principle unit models by compact and yet accuraternsurrogate models. We show how all the relevant variable relationships established by arnunit model, can be expressed in terms of a subset of the original model variables. Werndiscuss how this subset of variables can be identified, and we present a method torndevelop high quality surrogate models through artificial neural networks. Finally, wernpropose a tailored surrogate model reformulation to incorporate binary variables thatrnallow activation/deactivation of particular units within the superstructure model. Anrnexample is presented to illustrate the application of the proposed framework.