An initial value insensitive constrained linear predictive evolution algorithm for gas–liquid phase equilibrium calculation problems

Abstract

Heuristic algorithms are gradually becoming a type of new promising methods for solving phase equilibrium calculation problems since they do not have to calculate an initial value in advance like traditional methods such as direct Newton's method and indirect thermodynamic method. A new heuristic optimizer, constrained linear prediction evolution algorithm (CLPE) for phase equilibrium calculation under given volume, temperature, and moles (NVT-flash) is proposed in this paper. CLPE employs the total Helmholtz free energy of the NVT-flash problem as its objective function and employs the volume and moles vector of a certain phase as decision variables. Numerical experiments are conducted on four NVT-flash problems. The consistency between the experimental results and those obtained by some traditional methods verifies that the proposed CLPE is effective. The comparative advantage in computational overhead over the similar algorithms indicates the significance of this study. The success of CLPE can drive more heuristic algorithms to solve NVT-flash problems more efficiently, so as to advance the field of phase equilibrium calculation.