Improvement of two-phase heat transfer correlation superposition type for propane by genetic algorithm

Abstract

The prediction accuracies of the two-phase heat transfer coefficient for the flow in a small channel, which are usually based on the mean absolute error (MAE) between the correlation and experimental data, have remained unsatisfactory. Conventionally, the regression method has been used to determine the correlation that best represents the experimental data. In this paper, an improved heat transfer correlation for the evaporation of propane is developed by applying the genetic algorithm method. A total of 789 data points from 4 sources with circular diameters ranging from 1.0 to 6.0 mm are used to minimise the MAE while searching for the optimum conditions for the suppression factor, S, and convective factor, F, in a selected superposition correlation for two different vapour quality ranges. The optimisation can minimise the MAE at 33% and 25% for Case I and Case II, respectively. The proposed method assists in attaining a precise empirical prediction that fits well with the experimental data.