Modelling and real-time optimisation of a heat-exchanger network

Abstract

This work proposes a hybrid mathematical model and an optimisation-based tool to support the management of a heat-recovery section (formed by several heat exchangers) in a fibre-production factory. The purpose of the network is to heat different products using several hot sources, employed as utilities. Furthermore, concerns about the degradation of the equipment due to fouling are explicitly taken into account. Hence, the goals are to allocate the hot sources to heat exchangers and to suggest which heat exchanger should be cleaned to achieve optimal economic operation. Experimental models for the overall heat-transfer coefficients with respect to the flows have been identified, and production constraints are considered too. The problem is formulated such that it can be solved in a real-time optimisation fashion via mixed-integer non-linear programming. The approach has been tested through plant historical situations.