Predicting soil thermal properties in freeze-thaw cycles using EFAttNet: A comparative analysis

Area of Science:

• Geotechnical Engineering
• Soil Science
• Thermal Physics

Background:

• Soil thermal properties (conductivity and heat capacity) are crucial for construction and energy efficiency.
• Understanding how freeze-thaw cycles and moisture content affect these properties is essential for accurate modeling.

Purpose of the Study:

• To develop and validate a novel predictive model (EFAttNet) for soil thermal conductivity and volumetric heat capacity.
• To investigate the influence of freeze-thaw cycles, moisture levels, and other factors on soil thermal properties.

Main Methods:

• Developed EFAttNet, a model using embedding and attention-based fusion networks.
• Compared EFAttNet's accuracy against traditional empirical models (de Vries) and baseline algorithms.
• Conducted preliminary measurements under varying temperatures and moisture conditions.

Main Results:

• EFAttNet demonstrated superior accuracy in predicting thermal conductivity (λ) and volumetric heat capacity (C).
• λ increased with moisture content and decreased with temperature; C increased with moisture and freezing temperature.
• Degree of saturation, freeze-thaw cycles, subzero temperatures, and porosity significantly impacted thermal properties.

Conclusions:

• The EFAttNet model effectively captures complex, nonlinear relationships influencing soil thermal properties.
• Findings have significant implications for sustainable construction, energy efficiency, and engineering project design.
• The model's accuracy offers practical applications for managing soil thermal behavior in diverse environments.