The successful industrial applications of pinch analysis techniques in energy optimisation and wastewater minimisation have resulted in the recent studies of combined mass and heat integration. Kim and Smith (2001) have demonstrated that operation of cooling water networks in series, rather than the conventional parallel arrangement improve cooling tower and cooling water network performance in new and retrofit design. In this work, utilising a superstructure to determine the mathematical formulation that characterises a cooling water network supplied by multiple cooling water sources, which often occurs in practice, extends this methodology. It is further demonstrated that the optimum cooling water supply to a network of cooling-water-using operations supplied by multiple sources is determined by considering the entire framework of sources and cooling-water-using operations, that is, unified targeting. This optimum is better than that obtained from considering individual subsets of cooling-water-using operations and its respective source, that is, single source targeting. Relevant practical constraints were included in the formulations to enhance robustness and applicability to real life situations. Practical constraints consisted of maximum return temperatures to cooling water sources, as wells as dedicated water sources and sinks of cooling-water-using operations. This concept was applied to an illustrative example and a case study of the Sasol Synfuels (Pty) Limited cooling water system that consisted of individual networks supplied by separate water sources. For the case with maximum water reuse the single source targeting method yielded an improvement of 11.6% over the parallel target for the illustrative example. In comparison, superior results were obtained with the developed unified targeting method, which yielded an improvement of 18.4%. Likewise, for the case with the aforementioned practical constraints 6.8% and 7.6% improvements were forecasted for the single source and unified targeting methods respectively. For the maximum reuse scenario of the case study, improvements of 37.9% and 41.0% over the parallel target were obtained using the single source and unified targeting methods, respectively. Similarly, considering practical constraints improvements of 20.3% and 31.1% were obtained. In both the illustrative example and case study the unified targeting method resulted in superior results than the single source targeting methods.
Dissertation (MEng (Chemical Engineering))--University of Pretoria, 2008.