Phase equilibrium optimization using differential evolution with the gradient-based repair method

Abstract

Phase equilibrium computations using the Gibbs free energy minimization correspond to mass balance and thermodynamic model of the system are the global optimization problem and usually solved by the Evolutionary Algorithms (EAs). However, the equality constraint-handling techniques of the EAs have been still insufficient. Thus, the aim of this work is to develop the EAs for the efficient phase equilibrium computation by using the gradient-based algorithm to handle the equality constraints in Differential Evolution (DE). Moreover, a new objective function for the phase equilibrium problems modeled with the SAFT (Statistical Associating Fluid Theory) equation of state is proposed in this work. The new objective function uses the thermodynamic fundamental equation information in stead of the fugacity equation. The results show that the solutions obtained from our developed DE have smaller constraint violation than that of traditional DE. In addition, the Gibbs free energy value obtained and the number of function evaluation are also investigated in this work. The solutions obtained by our developed method with the SAFT model agree well with previous research and experimental works for the system of non-associating molecule.