Development of CAMD based on the hybrid gene algorithm and simulated annealing algorithm and the application on solvent selection

In this paper, an evolutionary approach, improved CAMD based on the hybrid gene algorithm and simulated annealing algorithm (GASA), is developed. The new approach combines the feature of GA and SA, avoiding the problem of prematurity. With a new category strategy of candidate groups from the Mod UNIFAC group database adopted, a repair operator is introduced to guarantee the integrity of randomly generated molecules and thus the search of straight chain alkane as well as cyclane solutions can be performed together. The properties of molecules are obtained by the group contribution method. An example of extractive solvent selection for a methanol‐methyl acetate system has been illustrated in detail to further explain the method. The stability as well as other properties of the molecules which the authors think important is considered in the fitness function. Results of the example show that the fitness ranking values are better than those in the literature. The CAMD method in this paper can be used in practical chemical processes.