A study of the thermal performance prediction of embedded pipes in protection engineering based on multiple regression analysis

There are currently a large number of information equipment and similar devices that generate a significant amount of residual heat that must be eliminated through protection engineering. The ground heat exchanger can be useful in the treatment of residual heat due to their high efficiency and adequate concealment. In order to eliminate internal waste heat, the heat exchanger are laid within the inner surface of the surrounding soil at engineering entrances. In this study, the numerical heat transfer model of embedded pipes in protection engineering is developed based on the finite element software COMSOL. In addition, a total of eleven independent variables representing influencing factors of embedded pipe heat transfer were taken into account. The average heat transfer rate per meter, the heat exchanger effectiveness, and the total thermal resistance in the first 20 days of embedded pipe heat transfer were taken into account as dependent variables. Accordingly, a single factor regression model was developed in order to investigate the correlation between the three dependent variables and each independent variable, and the independent variables that had no significant influence on the dependent variables were removed. Then, the orthogonal experiment tables of 10 factor 5 level, 7 factor 5 level, and 4 factor 5 level were established based on the three dependent variables and independent variables, respectively. According to the experimental groups in the three tables, the step-by-step method in multiple regression analysis was employed in order to obtain the prediction formula of the average heat transfer rate per meter, the heat exchanger effectiveness, and the total thermal resistance in the first 20 days of embedded pipe heat transfer. At the same time, the influence degree of each independent variable on the dependent variables is obtained. Three evaluation indexes were selected to analyze the error of prediction results and numerical solutions. It was found that the prediction model was accurate overall. Furthermore, the prediction formula is compared to the analytical solution, and the results obtained by the two methods are comparable. The predictive solution is superior to the analytical solution because of the fewer simplified conditions and a lot of numerical simulation. These results indicate that, it is worth noting that the analysis of the influence of each factor and the predictive solution have a certain reference value for the future application of embedded pipes.