An initial value insensitive method for phase equilibrium calculation: constrained quadratic interpolation optimization algorithm

doi:10.1016/j.eswa.2025.129597

Return

An initial value insensitive method for phase equilibrium calculation: constrained quadratic interpolation optimization algorithm

2025-09-02

PRE

Wangyu Tong

Baoduo Su

Y. H. Zhan

DOI：10.1016/j.eswa.2025.129597

Original

Original request for help

Save

Abstract

En 中文

The successful simulation of constrained differential evolution (CDE) algorithm for solving phase equilibrium calculation has first verified that heuristic optimization algorithms are effective ways to solve this kind of problems. Their insensitivity to initial values overcomes the limitations associated with two kinds of traditional methods, i.e., direct solution methods based on Newton’s method and indirect solution methods based on thermodynamic principles. This article proposes a constrained quadratic interpolation optimization algorithm (CQIO) for obtaining the satisfactory solutions of phase equilibrium calculation under given volume, temperature, and moles (NVT-flash). The proposed CQIO regards the total Helmholtz free energy of a NVT-flash problem as its objective function, while the moles vector and volume of a certain phase as its decision variables. The consistency between the four cases’ experimental results of CQIO and those of published articles demonstrates the effectiveness of CQIO in solving NVT-flash problems. Then the computational overhead and algorithmic stability of CQIO were analyzed. In Cases 1, 2 and 3, the average CPU time of CQIO compared to CDE has increased by 46.98 %, 54.56 % and 21.02 % respectively. The Std values of CQIO are significantly smaller than those of CDE in all cases except for Example 2. The proposed CQIO greatly promotes the application of heuristic algorithm in the field of phase equilibrium calculation.

Keywords:

constrained differential evolution

phase equilibrium calculation

NVT-flash

quadratic interpolation optimization

heuristic algorithms