How this calculation works
Preserve first-pass counts separately from recovered output and enforce a closed-cohort flow between stages. Report the product of stage first-pass rates as a distinct rate index, not a reconstructed count of never-reworked individuals.
Stage FPY = first-pass good / entering; final cohort yield = final accepted / initial enteringA worked example
A cohort starts at 1,000. Stage one accepts 900 first pass and recovers 50, passing 950 onward. Stage two accepts 900 first pass and recovers 20, leaving 920 accepted units: 92% final yield. There are 70 recovery stage events, not necessarily 70 unique reworked units.
Good to know
A closed sequential cohort with no branching or additions. Recovery sums count stage events. The product of stage FPYs does not identify unique never-reworked units without item histories. No sampling, defect-opportunity sigma conversion or cost inference.
Reference: Reference: model definitions and scope (opens in a new tab) ↗
Recovered output improves accepted yield without rewriting first-pass history
The default cohort starts with 1,000 units. Stage one has 900 first-pass good units and 50 recovered after rework, so it sends 950 onward. Stage two has 900 first-pass good and 20 recovered, leaving 920 accepted units. The final accepted cohort yield is 92%.
Stage-one first-pass yield is 90%; stage two is 900/950, about 94.736842%. Recovered units remain separate. Counting 950 as stage-one first-pass success would erase the rework that the record is intended to reveal.
The OEE calculator uses a first-pass good count in a separate interval measure. Recovered stage output does not become first-pass output retroactively.
A product of rates is not a reconstructed item history
Multiplying the two stage first-pass rates gives about 85.263158%. That conventional rate index is different from the observed final accepted yield of 92%. It also does not tell us exactly how many final items avoided rework at every stage, because the identities of the affected items are missing.
Likewise, the sum of 50 and 20 recoveries is 70 stage events. Some physical units might have been recovered at both stages, so the total is not necessarily 70 distinct reworked items. Item-level histories would be needed to establish that unique count.
A closed cohort makes the flow checks meaningful
Every next entering count must equal the prior accepted output. The tool rejects an unexplained addition or branching flow rather than silently applying a different denominator. Units removed from the modeled flow are counted separately; the result does not infer their disposal method, economic loss or environmental impact.
The overall equipment effectiveness calculator uses a first-pass good count for a separate interval measure. The takt time and capacity calculator uses an entered expected yield for a future planning scenario. This page instead summarizes supplied resolved stage counts, without predicting future quality or converting them into a defect-opportunity sigma level.
A few questions, answered
Does accepted output include recovered rework?
Yes. It is first-pass good plus recovered output, kept distinct from the first-pass rate.
Are summed recoveries unique reworked units?
Not necessarily. A physical item can have a recovery event at more than one stage.
Why must the next entering count match?
The supported model is one closed sequential cohort, with no new units added or branches between stages.
Does the rate product identify final never-reworked units?
No. Individual histories and dependencies are not supplied, so it remains a separate product-of-rates indicator.