Machine learning‐assisted enhancement of soil thermal conductivity for improved performance of Earth–air heat exchangers under Iraqi climatic conditions

Abstract

Earth–air heat exchangers (EAHEs) are widely used passive geothermal systems; however, their performance is often constrained by the low thermal conductivity of natural soil. The potential of iron filing waste (IFW) as a low‐cost, sustainable additive for modifying the thermophysical properties of the soil surrounding EAHE pipes is investigated in this work. The effects of IFW mixing ratios (10–30%) on the thermal and hydraulic performance of the EAHE system were analyzed using a three‐dimensional Computional Fluid Dynamics (CFD) model. An Extreme Gradient Boosting model was used to predict system performance, achieving an R ² of 0.984 and an mean absolute error of 0.253 while enabling rapid performance assessment. The incorporation of IFW improved the thermal performance of the EAHE compared with untreated soil. Increasing the IFW content beyond 10%, however, produced only marginal improvements in outlet air temperature, heat transfer rate, effectiveness, and coefficient of performance (COP), suggesting diminishing returns. The 10% IFW mixture increased the heat transfer rate by 46% and reduced the outlet air temperature by 7% compared with the baseline case, with no noticeable changes in hydraulic characteristics such as pressure drop, friction factor, and pumping power. At the highest investigated mass flow rate, the 10% IFW mixture achieved a COP of 6.3. Considering both performance and cost, this mixture represents the most favorable configuration, delivering substantial thermal enhancement with minimal material consumption. combining IFW with machine learning also offers an effective approach for the design and optimization of sustainable EAHE systems.