Integration of DMAIC and Two-Level Fuzzy AHP for Diagnosing and Prioritizing Waste and Potential Causes of Welding Defects in a Sidebeam Bogie Frame

Welding defects remain a critical quality issue in the robotic minor assembly welding process of the TB1014 Sidebeam Bogie Frame, affecting production efficiency and product quality. This study aims to evaluate baseline process performance, identify and prioritize manufacturing waste and its underlying causes, and propose a structured improvement roadmap. A Lean Six Sigma framework using the Define–Measure–Analyze–Improve–Control (DMAIC) approach was integrated with a two-level Fuzzy Analytic Hierarchy Process (Fuzzy AHP). Process performance was evaluated using Value Stream Mapping (VSM), time study, a u′-control chart, Defects Per Million Opportunities (DPMO), and sigma level analysis. We then applied Fuzzy AHP to expert judgments from ten respondents to prioritize four waste categories, followed by a 6M fishbone analysis and a second Fuzzy AHP assessment of potential cause categories. The results identified 485 welding defects across 21 inspected units, corresponding to a DPMO of 136,658.21 and a sigma level of 2.60. Undercut accounted for 57.1% of defects, while Process Cycle Efficiency (PCE) was only 4.88%. Defects emerged as the dominant waste category (0.7174), while Machine (0.4803), Method (0.2890), and Man (0.1596) were the leading potential cause categories. However, weak expert agreement and sensitivity to respondent selection indicate uncertainty in the prioritization results. Based on these findings and corrective-action records, an improvement roadmap is proposed focusing on seam-tracking verification, robot parameter standardization, jig and fixture positioning, tack welding, and fit-up control. The proposed improvements require implementation and empirical validation to establish their effectiveness.