The purpose of this study is evaluation of complicated kinetic models by means of Genetic Algorithm. Because of importance of sulfur removal from oil products to control environmental pollution as well as preparation of low/ free sulfur feed for reforming process, hydrotreating of oil cuts is one of the most important reactions in refining processes. Rigorous HDS reaction kinetics is nonlinear models which are carried out on bi-functional catalysts. Two parallel reactions; Hydrogenolysis and Hydrogenation, are the major reactions which are carried out on the surface of non-homogeneous catalysts as Ni-Mo/?-Al2O3. In this study, appropriate models of Langmuir-Hinshelwood type kinetics for Hydrogenolysis and Hydrogenation were investigated. The experimental rate data at high constant pressure in different temperatures with different concentrations of dibenzothiophene (heavy sulfur compound), hydrogen and hydrogen sulfide were used for each model. By implementation of GA the best model and Arrhenius parameters were obtained, directly.
In this study, a comparison with Levenburg-Marquardt Algorithm has been investigated. It is concluded that LMA is sensitive to the first guesses and there is possibility to trap in local optimum in contrast GA. In conclusion, GA can be used to modeling of rigorous and nonlinear kinetics, without consuming time for writing partial differentiation as we have to do in LMA, and GA is a convergent method to global optimua if we use a reasonable interval for the unknown parameters.
)