Adaptive decision-making in multi-stage production: a framework for cost optimization under sampling uncertainty

In multi-stage production, manufacturers face a critical trade-off between the cost of proactive quality control and the risk of downstream defects. This paper challenges the default strategy of early inspection by developing a unified framework to determine when a reactive, end-of-line recovery approach is more cost-effective. Our model uniquely integrates multi-stage production dynamics, the economic trade-offs of reverse logistics, and the statistical uncertainty of sampling inspection. Through optimization with a Genetic Algorithm, we identify specific, data-driven thresholds for market failure cost and initial defect rates where the optimal policy shifts decisively from selective to comprehensive upstream inspection. Furthermore, the analysis quantifies the value of information, demonstrating that higher data accuracy from stricter sampling protocols yields lower long-term costs by stabilizing decision-making. By providing a quantitative tool that adapts to evolving risk profiles, this research offers a practical approach for aligning cost optimization with the principles of Quality 4.0 and sustainable manufacturing.