Accepted version, 892.5Kb
Permanent link to Research Commons versionhttps://hdl.handle.net/10289/15682
The synthesis of heat exchanger networks (HEN) is not a trivial task due to the mixed integer and non-linear nature of the optimisation problem, yet it can provide significant cost and energy use savings to an industrial processing site. Many HEN synthesis models are developed using propriety solvers, which make it difficult for industrial companies to realise the potential of optimisation. This paper introduces a new open-source tool for HEN synthesis built in Python using the GEKKO library, which interfaces with the solver APOPT. The tool features three different synthesis methods, two of which utilise a two-step method where step one aims to initialise or narrow the search space prior to global optimisation in the second synthesis step. A unique feature of this work compared to previous HEN synthesis studies is the application of a true driving force constraint, dQ/dA, (instead of the conventional, but at times crude, minimum approach temperature, ∆Tmin) to provide feasible initial variable values to the second step. To demonstrate the current progress in developing the open-source tool, network solutions obtained by the tool were purposefully compared with the best reported solutions for three common case studies in HEN synthesis literature. For a small problem, the tool achieved a solution within 0.02 % of the best literature solution and attained a solution for the two larger problems within 4 % and 21% of the best literature solution. Before the tool can be made publicly available, further work is needed to implement a non-isothermal mixing model across stream splits, which is expected to yield a more optimal HEN.
©2022 AIDIC. Used with permission.