COMSOL-based parametric simulation dataset for thermal performance prediction of conventional light-gauge steel-framed walls
Description
This dataset contains 5,000 COMSOL-based parametric simulation samples for conventional light-gauge steel-framed walls. The dataset includes geometric parameters, material thermophysical properties, convective boundary conditions, indoor and outdoor air temperatures, and corresponding thermal performance indicators, including thermal transmittance, total thermal resistance, average heat flux density, and average inner and outer surface temperatures. The dataset was used to train and evaluate machine-learning surrogate models for rapid thermal performance prediction of conventional LSF walls.
Files
Steps to reproduce
The dataset was generated using a validated three-dimensional steady-state heat transfer finite element model developed in COMSOL Multiphysics. The conventional light-gauge steel-framed wall model consisted of sheathing panels, insulation infill, C-shaped steel studs, and upper and lower U-shaped tracks. The finite element model was first validated using experimental data reported in the literature. After validation, parametric simulations were conducted by varying wall geometric parameters, material thermophysical properties, convective boundary conditions, and indoor and outdoor air temperatures. For each parametric case, the average heat flux density, inner and outer surface temperatures, total thermal resistance, and thermal transmittance were extracted from the COMSOL results. The resulting dataset contains 5,000 valid samples and was used to train and evaluate machine-learning surrogate models for rapid thermal performance prediction of conventional light-gauge steel-framed walls.