هادی سلطانی

In this research, a simple procedure for synthesis of reactor network (RN) is presented. This method could find the best possible optimum solutions with high grantees for the synthesis of RN, which generally is kind of mixed integer nonlinear programming (MINLP) models. In this work, instead of solving the discrete and continuous variables simultaneously, (like MINLP problems), a combination of stochastic algorithm and mathematical methods (i. e. Non-dominated Sorting Genetic Algorithm II (NSGA II) + quasi LP (QLP)) has been used. The reason for using NSGA II is because of the possibility of considering several objective functions instead of a single function in obtaining RN synthesis solutions. This algorithm could provide several optimal structure of RNs as set of final optimal solutions by use of its operators. The isothermal RNs produced by NSGA II are sent to QLP model to compute the amount of objective functions. In the QLP model, whilst all continues variables of RN calculated with just using a linear and a non-linear equations set with outer search loop, their general objective functions are also calculated. Comparison the results with the references shows that the presented method is capable of producing better solutions, which, in some cases improvement of about one percent in concentration of the main material in the product and a decrease of 77 % of the residence time of the materials in the reactor system is observed.