Abstract
Global energy demand has increased continuously since the last few decades and it has been a critical issue especially in industrial sector. Heat exchanger network (HEN) has received considerable attention for improving heat recovery in industrial processes. In this work, HEN synthesis for multiperiod operation has been studied. Sequential and simultaneous approaches for multiperiod HEN design are proposed and compared by a case study. The most efficient method will be applied to a case study of crude distillation unit (CDU) where different kinds of crudes are used. The objective for both methods is to minimize total annualized cost (TAC) including capital cost and utility cost. The sequential approach consists of three steps. First, an MINLP superstructure-based model is used to generate an initial HEN for a chosen period. Then it will be adapted by NLP model to generate HENs which are fitted to other period conditions. Lastly, HENs for each period are integrated to obtain the multiperiod HEN design. By varying the chosen period in the first step with all periods, it will result in different multiperiod HEN candidates. The best one will be selected as the final solution for sequential method. For simultaneous approach, an MINLP simultaneous model takes into account all periods concurrently and solve at once. Maximum-area-per-period concept is used in area calculation. The results demonstrate that the simultaneous approach showed better performance than sequential approach. Thus, the simultaneous approach is then applied further to the industrial case study of crude preheat train in CDU to assure that the model can deal with larger problem. In this case, an initialization strategy has been carried out to find an initial feasible solution. It showed that the initialization technique can reduce computational time substantially. Moreover, the final solution of HEN will be validated by commercial process simulator, PRO/II, to affirm its feasibility in real process.
