A Systematic Data-Driven Framework for Analysis and Control of Weld Defects in Manufacturing

Abstract

Welding defects remain a major source of rework, scrap, production interruption, and reliability risk in manufacturing. Their occurrence is rarely controlled by a single variable because welding quality depends on the interaction of process parameters, joint preparation, material condition, equipment health, operator practice, inspection, and environmental conditions. This study develops a systematic engineering framework for the analysis and control of weld defects by integrating defect classification, structured quality-data collection, Pareto prioritization, process stratification, Fishbone analysis, 5-Why reasoning, risk-based prioritization, and preventive control. The framework is supported by recent industrial and safety evidence and is designed for application to manual, semi-automatic, and automated welding operations. A manufacturing-oriented case application is used to demonstrate how published evidence can be converted into a practical quality-improvement workflow without treating secondary case results as original experimental observations. The analysis emphasizes prevention and root-cause verification rather than end-of-line detection alone. It also shows how conventional quality tools can be extended with process sensing, statistical monitoring, and automated inspection. The resulting framework provides a reproducible basis for welding-quality management and establishes a transition path from defect recording to data-enabled process control.