MOLECULAR DESIGN OF ALTERNATIVE REFRIGERANTS USING GENETIC ALGORITHMS

Designing new molecules in desired properties involves time-consuming and expensive steps. Among the reported computer-aided techniques, knowledge based approaches suffer from acquisition of appropriate heuristics whereas mixed integer nonlinear programming formulations are not trivial to generate. Present study describes a new computer-aided refrigerant design using genetic algorithm. Genetic algorithms perform a guided stochastic search and allow improved solutions by sampling areas of the space having higher probability for good solutions. Results, very close to those previously reported by global optimization algorithms using MINLP formulation, illustrate the effectiveness of genetic algorithms to solve such computationally intensive problems with ease of us.