End-to-End Lot & Roll Traceability Depth and Recall Exposure

The model assumes each affected lot is fully consumed by the next stage and that blending is complete, which is the conservative reading and the one auditors apply. Where a lot is only partly consumed the true implicated mass is smaller, but you need consumption records at lot level to claim that, and if you had those you would probably not be estimating. Fan-in values should be the maximum the process permits, not the average, because a recall is decided by the worst case in the window rather than the typical one. The chain is treated as a simple sequence: a genuine bill of materials with parallel warp and weft supply, re-dyed seconds re-entering the stream, or leftover fabric returned to stock forms a graph rather than a chain, and each of those paths adds fan-in this calculation will not see. Yield is applied as a single cumulative figure and assumes waste is representative - if the affected material is preferentially removed, by a metal detector or a contamination sorter, the real exposure is lower, and if it is preferentially retained the figure here is optimistic. The quarantine result sizes containment only; it is not a recall decision, which also turns on where the product has already shipped, what the specification actually requires, and the regulatory regime of the destination market. Finally, the window cap means a broken chain and a complete one can return the same quarantine figure when dispersion already exceeds the window - that is not the records working, it is the exposure being larger than the period you asked about.