Abstract
This contribution describes a new two-step solution strategy for the synthesis of heat-integrated process water networks (HIPWNs). The proposed strategy involves the solutions of two models. The first, nonlinear programming (NLP) model consists of a water network (WN) model and a simultaneous optimisation and heat integration model. The objective function in model minimises the consumption of freshwater, heating and cooling utilities. The second, a mixed-integer nonlinear programming (MINLP) model combines the WN model from the first step using a heat exchanger network (HEN) synthesis (Yee et al., 1990) model for minimising the total annual cost (TAC) of the overall combined network. According to the proposed strategy, the first targeting NLP is solved during the first step in order to provide good initialisation for the second step, as well as to determine upper bounds for water and utility consumption, assuming a given value for the heat recovery approach temperature (HRAT). Optimal overall network structures are then obtained during the second synthesis step by solving the MINLP model, now having all the temperature driving forces in HEN as optimisation variables. The two-step procedure is repeated for a range of HRAT values. A set of good locally-optimal solutions is thus identified and the best one with minimum TAC is chosen from amongst them. The solutions obtained indicate that the proposed strategy can be successfully applied to the synthesis of HIPWNs. The results of the threshold case-studies are similar to those found in the literature. However, better solutions were achieved in pinched cases because the proposed synthesis model enables the obtaining of appropriate trade-offs between freshwater, utility consumption and investment.
