Coupled Statistical–Numerical Framework for Predicting Droplet Temperature in Gas Metal Arc Welding

Abstract

Accurate prediction of droplet temperature during Gas Metal Arc Welding (GMAW) is crucial for controlling metallurgical outcomes and process stability, yet direct measurement remains a challenge due to the rapid thermal dynamics involved. This study aims to develop and validate both empirical and physics-based models to estimate droplet temperatures during GMAW. A total of 20 experimental runs were performed using mild steel (AISI 1018) under a Central Composite Design (CCD), varying welding current (240–270 A), voltage (23–26 V), and wire feed rate (2.4–3.0 mm/s). Droplet temperatures were measured with a high-resolution pyrometer and modeled using Response Surface Methodology (RSM) in Design Expert and a 1D Finite Element Method (FEM) implemented via MATLAB. The RSM model showed high accuracy with an RMSE of approximately 4.3 °C compared to experimental results, while the 1D FEM model achieved an RMSE of 192.7 °C, capturing the general thermal trend despite simplifications such as constant properties and 1D heat conduction. These findings demonstrate the complementary strengths of empirical and computational approaches: RSM offers precise data-driven predictions, while FEM provides a deeper physical understanding of heat transfer during droplet formation.