Numerical Estimation of Temperature-Dependent Thermal Properties

Abstract

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.In the present study, the Quadrilateral Optimization Method (QOM) is applied for parameter estimation in an inverse heat transfer problem. A numerical LASER Beam Welding (LBW) experiment of SAE 1020 is the baseline for the estimations. The temperature-dependent thermal conductivity of the steel is assessed. The algorithm accounts for the conductivity as a second-degree polynomial function of temperature. The three parameters of the function are simultaneously assessed. The method regularizes the objective function through Future Time Regularization (FTR) to account for the temporal analysis. Hence, the effect of using different numbers of time steps was analyzed. The most accurate results were found when considering 60 points. Thus, this configuration was set to expand the algorithm to assess the gross heat rate provided by the LASER along with the thermal conductivity function. The results are sensitive enough to represent reliable assessments. Considering the reference and estimated values, the simulated temperatures show good agreement. The present algorithm requires low computational cost due to a GPU’s parallel computation.

Keywords