Thermal Conductivity Mechanism Analysis and Neural Network Model of Clay Backfill Mixed with Different Amounts of Steel Slag

Abstract: Aiming at the low recovery rate of waste steel slag in China and the poor heat transfer effect of single clay as a heat pump backfill material, a backfill material for a ground source heat pump was prepared using steel slag particles and clay as the main raw materials. The thermal conductivity coefficient was optimized with the thermal conductivity as the index, and the thermal conductivity of four materials of different grades under different moisture contents and dry densities was tested based on the hot-probe method. The microscopic morphology and thermal phase composition of the materials were analyzed by SEM and CCD combined stereo microscope and other equipment. Based on the test results, a thermal conductivity prediction model of the backfill material with the BP neural network was proposed and then optimized on the basis of an error comparison analysis. The thermal conductivity of the backfill material could be improved with the increase in the steel slag content, and the thermal conductivity of the steel slag–clay combination backfill material showed a positive correlation relationship with the increase in the moisture content and dry density. The main reason is that the number of pores and volume of the porous material were reduced, the contact thermal resistance between the particles was reduced, and the heat transfer mechanism inside the backfill material was improved. The material properties, moisture content, and dry density were introduced into the prediction model of the thermal conductivity as independent variables, with good prediction results. The model produced an average relative error of 1.0007% and an average absolute error of 0.06365 W/(m·K), which are suitable values for calculating the thermal conductivity of porous backfill materials. This study provides a theoretical basis for ground source heat pump backfill engineering and can help improve the utilization rate of waste steel slag