Process integration as an optimisation tool in multipurpose batch plants

Abstract

Heat integration to optimise energy usage becomes a possibility if a process includes both heat generating and heat consuming operations. Heat integration in batch plants has in the past been largely disregarded as utility requirements are considered less significant due to the smaller scale of batch operations compared to continuous plants. However, utility requirements in some batch plants, such as in the food and drink industries, dairies, meat processing facilities, biochemical plants and agrochemical facilities, contribute largely to their overall costs. This thesis is a continuation of the work published by Stamp and Majozi (2011) and two different aspects of heat integration in multipurpose batch plants are considered. Firstly, wastewater minimisation constraints from the model of Adekola and Majozi (2011) were superimposed into the heat integration model of Stamp and Majozi (2011) and the simultaneous optimisation of scheduling, energy and water was considered. This has not been covered extensively in published literature as the optimisation of all three aspects of a multipurpose batch plant complicates the optimisation. The proposed simultaneous method was compared to a published sequential method and gave an improved profit of 6.78% for a multipurpose example. Secondly, a model for the simultaneous optimisation of the schedule and energy usage in heat integrated multipurpose batch plants operated over long time horizons is presented. The method uses a cyclic scheduling solution procedure. Indirect heat integration via heat storage was included, rather than just direct heat integration. This has not been considered in longterm heat integration models in current literature. Both the heat storage size and initial heat storage temperature were also optimised. The solution obtained over 24 h using the proposed cyclic scheduling model with heat storage for a simple sequential process was compared to the result obtained from the direct solution and an error of less than 1% was achieved.