AI-Based Quality Control Framework for Reducing Rework and Scrap in Additive Manufacturing Production

Abstract

Additive manufacturing organizations continue to experience costly rework and scrap because defects are often identified only after substantial material, machine time, energy, labor, and post-processing resources have already been consumed. This study aimed to develop and quantitatively evaluate an AI-based quality-control framework that explains how technological, informational, human, and organizational capabilities improve quality-control effectiveness and reduce production waste. A quantitative, cross-sectional, descriptive, explanatory, and case-based research design was applied across selected additive manufacturing enterprise cases, including industrial facilities, engineering laboratories, and production units using laser powder bed fusion, material extrusion, directed energy deposition, selective laser sintering, binder jetting, and related technologies. From 350 distributed questionnaires, 300 valid responses were retained, producing an effective response rate of 85.71 percent. The framework examined Process-Data Quality and Integration, AI-Based Defect-Detection Capability, Real-Time Process Monitoring and Control, Predictive Quality Analytics and Decision Support, AI-System Integration, Workforce Competency and Organizational Readiness, AI-Based Quality-Control Effectiveness, and Rework and Scrap Reduction Performance. Data were analyzed in SPSS using descriptive statistics, reliability and validity testing, Pearson correlation, ANOVA, multiple regression, and multicollinearity diagnostics. Cronbach’s alpha values ranged from .823 to .918, KMO reached .891, and Bartlett’s test was significant, χ²(378) = 4,126.44, p < .001. AI-Based Defect-Detection Capability recorded the highest mean, M = 4.13, SD = .54. The six predictors jointly explained 68.4 percent of the variance in quality-control effectiveness, R² = .684, with defect detection strongest, β = .27, followed by real-time monitoring, β = .23. Quality-control effectiveness strongly predicted rework and scrap reduction, β = .76, R² = .579, p < .001. The findings imply that coordinated data, intelligent detection, monitoring, predictive analytics, system integration, workforce capability, and organizational readiness can reduce failed builds, rejected components, corrective processing, material waste, and production costs. They also support improved first-pass yield, traceability, resource utilization, and timely intervention.