Confidence-Based Skip-Lot Sampling

An ammunition production facility’s responsibilities are two-fold: produce at a high volume and ensure product quality through adequate testing. When a producer has a long history of high manufacturing quality for a particular product, it is possible that they may be significantly over-testing the lots in production. To address this issue, we have developed a skip-lot sampling procedure modeled after a Bayesian approach to estimate the true quality of each lot in production, even as some lots are skipped. Simulation results demonstrate that this process can both reduce the number of tests required and meet the production facility’s standards for risk exposure. As skip-lot sampling methods introduce two new possibilities (savings by skipping good lots and risk by skipping bad lots), our approach appears more conservative than a common skip-lot sampling method as it prioritizes risk management over savings.