Thermal transmittance of slab on ground floors: A comparison of simplified methods and dynamic modelling,insulation thickness,and edge effect

Abstract

Calculation of heat loss through slab-on-ground depends on the methods used. This study evaluates the thermal transmittance of slab-on-ground floors using 14 methods, including validated 2D and 3D transient and steady-state simulation tools and 10 simplified 1D calculation approaches, including EN ISO 13370 standard. The analysis considers insulation thicknesses 0–800 mm, four building geometries, and edge configurations to assess existing methods for modern, highly insulated structures. The results reveal that the EN ISO 13370 standard procedure for typical detached and terraced house sizes demonstrated accuracy within ±3%, and for large buildings within ±6%, as validated by steady-state and dynamic simulations. Methods using zone-based approach resulted in relative differences 54% lower to 271% higher in extreme, primarily uninsulated floors. Other one-equation methods showed even larger maximum relative difference variations, especially in uninsulated cases, ranging from 43% lower to 3120% higher across floor sizes and methods. This highlights the importance of representing ground effects properly, especially for uninsulated cases. Models with typical wall-floor junctions showed that neglecting FEM-calculated thermal bridges and relying solely on EN ISO 13370 for main and edge heat loss could omit up to 40% of heat loss in well-insulated 20 × 8 m floors. Significant inaccuracies were observed in uninsulated cases when evaluating the edge heat loss correction factor itself (excluding wall-floor junction effects) against FEM results. Vertical edge insulation cases calculated using the EN ISO 13370 gave lower edge heat loss, while horizontal edge insulation cases gave higher edge heat loss compared to 2D FEM results. A biased scenario is also discussed, where the edge insulation width equals half of the characteristic length, covering the entire floor area in 2D but not in 3D, emphasising the need to appropriately scale horizontal edge insulation length, particularly in small and primarily uninsulated 2D models.