Study on the Effects of Radiation Insulation Film and Backfill Soil on Shallow Ground Source Heat Pumps Within 5 Meters of the Surface

Abstract

Ground Source Heat Pumps (GSHPs) are a sustainable solution for low-energy heating and cooling, yet their performance is significantly affected by insulation materials and surrounding soil properties. This study aims to investigate the effects of radiation insulation film and backfill soil types on the thermal performance of shallow ground source heat pumps (SGSHPs) installed within 5 meters of the surface. A simulation-based, quantitative methodology was implemented using Python in Google Colab, generating 1500 synthetic data points analysed via supervised machine learning models, including SVR, XGBoost, and Random Forest. Results showed that sand-based soils and thicker insulation layers significantly improved heat transfer and efficiency, while SVR achieved the best predictive accuracy, and XGBoost outperformed others in classification tasks. The study concludes that ML models are effective in system optimisation. Recommendations include using AI tools in design phases. The findings provide theoretical insights into geothermal optimisation and practical guidance for SGSHP design, though their generalizability is limited by the absence of real-world validation data.