| The traditional chemical process is composed of reaction and separation procedure.Nevertheless,the reactant conversion and product yield are strigently restricted by equibrium-limited reaction.Besides,the separation become more complicated due to existence of multi-component in the discharge stream from reactor,which results in significant increase of energy consumption for multi-product distillation.In order to make the chemical process more energy efficient and sustainable,process intensification is employed to provide critical route to tackle the associated issues.As a vital method for achieving process intensification,reactive distillation could simultaneously improve the conversion and reduce energy consumption/capital cost investment for equilibrium-limited reaction,by removing the target product opportunely and utilizing the reaction heat sufficiently.Optimal design distillation and reactive distillation column shows to be a rigorous mixed integer nonlinear programing(MINLP)problem,which was usually tackled using by coupling simulation software and stochastic algorithm.However,the computation load and convergence problem impose significant challenge to the classical simulation-based optimization problem.In the present work,disjunctive variable was employed to represent integer variable of the process and Generalized Disjunctive Programming(GDP)is introduced to establish mathematical model for reactive distillation.Then logic-based outer approximation(LBOA)was used to decompose the disjunctive model into the master problem-mixed integer linear programming(MILP)and subproblem-nonlinear programming(NLP).Based on the optimization results of reactive distillation,the Kemira-Leonard process of formic acid production process is simulated and compared,taking the hydrolysis of methyl formate to formic acid as an example.The results show that reactive distillation can significantly improve the conversion of methyl formate.It can be seen that the superiority of the equation-oriented method in terms of optimality and computation requirement. |
