Modeling of frost thermal conductivity on parallel surface channels

Abstract

Frost buildup on cold plates in heat exchange applications causes an increase in the pressure drop and a decrease in the heat transfer rate. There is a lack of practical and accurate models for evaluation of frost thermal conductivity in the literature. Inaccuracy in estimating the thermal conductivity of frost on parallel surface channels will increase the error in predictions of frost thickness and density. In the present study, a mathematical method known as Least Squares Support Vector Machine (LSSVM) is utilized in order to develop a robust model for predicting the thermal conductivity of frost formation on parallel surface channels. Six key factors affecting the thermal conductivity of frost, namely, frost porosity, wall temperature, air temperature, relative humidity, air velocity, and time duration, are considered as the inputs to the model. Genetic Algorithm (GA) is used to determine the optimized hyper parameters (γ and σ2) of the LSSVM model. The reliability of the GA-LSSVM model is evaluated by comparing the model results with those of existing empirical correlations. The obtained mean square error, root mean square error, and relative root mean square error for the GA-LSSVM model are found to be 1.10e−05, RMSE = 3.3e−03, and RRMSE = 2.42%, respectively. The results show that the GA-LSSVM is a practical and easy-to-use model for estimating the thermal conductivity of frost on parallel surface channels under different conditions.