Profit-driven Optimization For Consecutive-competitive Chemical Reactions

Consecutive-competitive reaction (CCR) is one of popular chemical reaction forms in chemical industry, for example, the production of ethanolamines and isopropanolamines. The path of CCR can be described by the form Aâ†'Bâ†'C, where its intermediate products would be processed into multiple finished products through further reaction steps. Finally the product-mix of CCR can be separated through a distillation operation. From business point of view, in order to maximize the enterprise's profits, it is critical to determine and manipulate a desired finished-product-mix under an uncertain market place with the changes of demand of finished-products as well as the prices of feedstock and energy. Consequently, the tasks of an integrated business and production system are twofold: determine an optimal product-mix and the relevant control strategy to meet the business goal subject to both business and production constraints.This thesis starts with a general review of the research progresses on system modeling and product-mix optimization for chemical processes. By combining first principles with genetic algorithm (GA), a grey-box modeling approach to complex chemical reaction processes was developed. This study also proposes a GA-based product-mix optimization solution with a bridge between profits, product-mix and operation optimization, and then the proposed solution is applied to the production of ethanolamines, a typical consecutive-competitive reaction system. The production data based simulation results show that the proposed solution could be applied in CCR processes to implement the business and production integration. The main topics studied in this thesis are summarized as follows:(1) This thesis proposes a GA based grey-box model which is fully utilizing the strengths of first-principles model and statistical model. This proposed grey-box model is developed with first-principles oriented model structure and unknown model parameters. The later could be estimated by using GA algorithms. The effectiveness of the proposed grey-box model is proven by the simulation results for a typical consecutive-competitive reaction.(2) This thesis also proposes an integrated solution that combines product-mix optimization with operation optimization. The proposed optimization solution is tested with a production data based grey-box model for an example consecutive-competitive reaction process.(3) The proposed solutions are further applied to a production-scale ethanolamine production line. In this study, a process model presenting the relationship between operation conditions and product-mix, and a profits model presenting the relationship between product-mix and the resulting profits are developed respectively. On the basis of the combining these two models, GA is used to optimize the production profit for the ethanolamine production line.The proposed integrated solutions have great potentials in the optimization control for many other consecutive-competitive chemical reaction processes.