Performance estimation of freeze protection system for outdoor fire piping by using AI algorithm

초록

In the present study, the performance of the metal heater-based freeze prevention system was predicted with respect to the major design variables of outdoor fire piping, including the fire pipe diameter, outer temperature, and insulation thickness. To this end, CFD analysis was conducted to obtain water center temperature data along the pipe-length direction with respect to the major design variables. Subsequently, four AI algorithms, including the deep neural network, decision tree, random forest, and support vector machine, were trained with the collected data, and their prediction performance was compared. Further, each algorithm, once trained, was tested for its ability to make reasonable predictions for the conditions that it had not been trained with. Overall, the deep neural network model exhibited the best prediction performance for both interpolation and extrapolation data. As a result, the model was determined to be the most suitable for the prediction of the water temperature.

키워드