Optimization of in-situ soil thermal desorption technology based on machine learning and heat transfer process model

Xin Wang¹Yihe Li²Yong Tian³Bowei Zhang³Qinglan Li⁴Shufeng Xi¹Ying Zhao⁵Tingting Zhang⁵Qianting Ye^6*Rong Li^6*

• ¹Shenzhen Technology Institute of Urban Public Safety Co Ltd, Shenzhen, China
• ²Public Utilities Bureau of Shenzhen Shenshan Special Cooperation Zone, Shenzhen, China
• ³Southern University of Science and Technology, Shenzhen, China
• ⁴Chinese Academy of Sciences Shenzhen Institute of Advanced Technology, Shenzhen, China
• ⁵Wisdri City Environment Protection Engineering Limited Company, Wuhan, China
• ⁶South China University of Technology, Guangzhou, China

In-situ soil thermal desorption (ISTD) has been recognized as an effective and promising technology for remediating organic contamination in soil and groundwater. However, high energy consumption poses a major constraint on its remediation costs. In this study, an optimization method for in-situ soil thermal desorption was proposed that combines machine learning with a heat transfer process model. This method optimized the heat flow by effectively predicting the temperature distribution during ISTD, thereby enhancing energy utilization and reducing technical costs. The results show that total energy consumption can be significantly reduced under variable heat flow conditions compared to constant heat flow, with energy savings of 35.93–48.86%. The practical technical implementation requires careful consideration of factors such as heating time, fluctuations at the cold spot temperature, and the intensity of the heat flow. This study provides essential technical support for the further development of in-situ thermal desorption soil technology in practical engineering contexts and the strategic optimization of remediation methods.