Interaction parameter estimation in liquid–liquid phase equilibrium modeling using stochastic and hybrid algorithms

Metaheuristic methods Simulated annealing (SA) and Genetic algorithm (GA) combined with hybrid functions were investigated to estimate interactions parameters in liquid–liquid phase equilibrium systems. The Rastrigin and Rosenbrock functions with high number of variables were used as test functions to compare the performance of each algorithm. The experimental ternary LLE data for extraction of alcohol with dichloromethane (and diethyl ether) were considered in the NRTL and UNIQUAC activity coefficient model. The interaction parameters were carried out taking into account the closure equation. As a main conclusion, Hybrid algorithms have showed the best performance to estimate the interaction parameters comparatively to stochastic and deterministic algorithms.