Simultaneous process optimization and heat integration for ethylene-to-ethylene oxide process: A surrogate model-based approach

Abstract

Since the production process of ethylene to ethylene oxide (EO) is highly energy-intensive, it is necessary to perform a whole process heat integration. However, due to the technical barriers, it is always a challenge to design the heat integration system while optimizing the process parameters by mechanistic models. In response to this issue, this study presented a novel surrogate model-based framework that allows for the seamless integration of process optimization and heat exchanger network (HEN) design. The EO production process is modeled and simulated to provide the essential data for surrogate model training. Considering the inherent complexity of the process and the challenges associated with model training, the whole production process is decomposed into parts and the output parameters relevant to heat integration are included within the surrogate models training by Artificial Neural Network. Building upon these preliminary steps, a flexible optimization framework built upon the Genetic Algorithm, integrating the developed process surrogate model and any HEN synthesis model, is constructed to achieve synchronous optimization. The results show that the obtained process saves 22.93 % utility cost and finally leads to a 2.60 % increase in total profit than the alternative designed by the conventional stepwise method, showing the priority of the proposed method.