Design of flexible heat exchanger network for multi-period operation

Heat exchanger networks (HENs) increase heat recovery from industrial processes by matching hot and cold streams to exchange heat and reducing utility consumption. The design of HENs is a very complex task which generally involves mixed-integer non-linear programming (MINLP). This work evaluates and critically compares existing HEN design methods. It then presents a systematic methodology in the design of HENs under multiple periods of operation. The model presented in this work is a superstructure-based MINLP model which minimises the total annualised cost containing heat exchanger area cost and utility costs. The model is based on the superstructure by Yee and Grossmann [1990. Simultaneous optimisation models for heat integration—II, heat exchanger network synthesis. Computer & Chemical Engineering 14(10), 1165–1184], which was later formulated for multiple periods by Aaltola [2002. Simultaneous synthesis of flexible heat exchanger network. Applied Thermal Engineering 22, 907–918]. It includes a multi-period simultaneous MINLP model to design the HEN structure, and an NLP model to improve the solution and allow for non-isothermal mixing. Modifications to Aaltola's model include the use of maximum area per period in the area cost calculation of the MINLP objective function, and the removal of slack variables and weighed parameters from the existing NLP improvement model. The new model has been applied to one industrial case study, demonstrating that the new combined MINLP–NLP model can obtain better solutions by not relying on the average area assumption in the MINLP stage.