A STATISTICAL APPROACH TO PREDICT AND OPTIMIZE RESIDUAL STRESS WITH THE APPLICATION OF RESPONSE SURFACE METHODOLOGY (RSM)

Abstract

Residual stresses in mild steel weldments critically affect the structural integrity, dimensional stability, and overall performance of welded assemblies. These stresses arise due to the thermal cycles inherent in welding processes and are further influenced by key process parameters such as welding current, voltage, shielding gas flow rate, and ambient temperature. The interplay among these parameters is complex and nonlinear, often making it challenging to predict and control residual stress outcomes accurately. To address this challenge, the present study aimed to model, optimize and predict residual stress variables using an expert statistical systems approach: Response Surface Methodology (RSM). The research not only seeks to identify optimal parameter settings that minimize residual stress but also to evaluate and compare the predictive capability of the modeling approach in the context of welding process design and the experimental results. The methodology involved conducting controlled welding experiments, followed by a comprehensive statistical analysis of the data using Analysis of Variance (ANOVA) to assess the significance and interaction effects of the input parameters on responses (Residual Stress). RSM was employed to develop second-order polynomial models for the response variable, enabling an objective optimization through the desirability function approach. The model was validated using experimental datasets and assessed for accuracy through metrics such as coefficient of determination (R²), residual error analysis, and visual diagnostic tools including scatter plots and time series comparisons. These insights deepen the understanding of how welding settings impact the residual stress behavior of weldments. RSM produced robust multi-response solutions with Optimal parameters (Current: 215.935A, Voltage: 24 V, Gas Flow Rate: 14.68 L/min Ambient Temperature: 28°C, Residual Stress: 214.612 MPa). These settings successfully optimized the Residual Stress and with a desirability index of 0.982 confirms the effectiveness of RSM in balancing complex trade-offs in welding optimization.