Advanced Modeling and Optimization of Weldment Responses Using Statistical and Metaheuristic Techniques

Welding process optimization plays a crucial role in enhancing the material properties of weldments and ensuring high-quality outcomes in industrial applications. This study focuses on developing a robust framework for optimizing welding parameters to improve weldment properties, specifically carbon content. Understanding the effects of welding parameters — current, voltage, and gas flow rate — on carbon content is essential for reducing defects, improving weld quality, and achieving cost efficiency. The experiment was conducted at the Petroleum Training Institute (PTI), Warri, utilizing a Central Composite Design (CCD) to systematically analyze the interactions and effects of the welding parameters. A total of 20 experimental runs, including factorial points, axial points, and central replicates, were performed to ensure comprehensive evaluation and error estimation. Response Surface Methodology (RSM) was employed to develop predictive models, while Particle Swarm Optimization (PSO) was applied to refine the optimization process, leveraging its ability to identify global optima in complex solution spaces. The results demonstrate the effectiveness of combining RSM and PSO for advanced welding process optimization. RSM achieved a minimized predicted carbon content of 0.080 mole, with an experimental validation of 0.0518 mole. PSO further enhanced the optimization, predicting a carbon content of 0.0237 mole and achieving an experimental value of 0.0309 mole, demonstrating superior performance in minimizing carbon content. These findings underscore the potential of integrating statistical modeling with metaheuristic techniques to achieve precise control over welding parameters and deliver actionable insights for industrial applications.