Rework Is a Process, Not an Incident
Rework in zipper production rarely arrives as a dramatic event. It arrives as a bench, two people, a tray of sliders and a day that quietly disappears. Because it is absorbed inside normal working hours, it is often invisible in cost reporting even though it consumes labour, machine time and material every shift.
A rework cost model makes that consumption visible and, more importantly, comparable. It allows a manager to compare the cost of fixing a defect with the cost of preventing it, and to compare two repair routes for the same fault.
Classifying Rework Before Costing It
Different rework types have different cost structures, and mixing them into one rate produces a number that cannot be managed. A practical classification for a zipper plant is:
- In-line correction: the operator fixes the piece at the station, usually within seconds, often unrecorded.
- Offline repair: the piece leaves the line for a repair bench, where a slider is replaced or a tooth is trimmed.
- Sorting and re-inspection: a lot is checked at one hundred percent to remove a known fault.
- Re-processing: the piece returns to an earlier step, such as re-plating or re-dyeing.
- Downgrade: the piece is sold as second quality rather than repaired.
Each type needs its own unit cost, because sorting a carton and replacing a slider differ by an order of magnitude in labour content.
The Cost Elements
Five elements make up the cost of a rework event:
- Direct labour: minutes spent multiplied by the fully loaded labour rate, including benefits and supervision share.
- Machine and energy time: the share of equipment cost consumed while the machine produces no saleable output, or while a separate machine is used for repair.
- Consumable material: replacement sliders, pins, stops, tape sections, plating chemicals and cleaning agents.
- Yield loss: material scrapped at the end of an unsuccessful repair attempt, valued at the cost of work already invested.
- Handling and administration: movement, paperwork, re-packing, and the supervision time spent deciding what to do with the lot.
The formula is the sum of the five: rework cost equals labour minutes divided by sixty, multiplied by the labour rate, plus machine hours multiplied by the machine rate, plus consumables, plus yield loss, plus handling. For a lot rather than a single piece, multiply by the number of units processed and add the one-off handling cost once.
Rates, Not Invoices
Most of these elements have no invoice attached, so the model must use standard rates. Two rules keep the rates honest. First, the labour rate must be fully loaded; using a base wage understates rework by a large margin and makes prevention look expensive by comparison. Second, the machine rate must reflect the cost of the capacity consumed, not just electricity, because an assembly machine occupied by repair work is not making saleable product.
Publish the rates alongside the result and keep them fixed for the reporting year. A model that silently changes rates cannot be used to show improvement.
Capturing Hidden Rework
The largest gap in most rework models is the work that never reaches a ticket. An operator who re-threads a chain or replaces a slider without logging it still spends the time. Three practical detection methods work better than asking people to record more:
- Reconcile input against output at short intervals. A gap that is not explained by recorded scrap is unrecorded rework.
- Count consumables issued against units produced. An unexplained rise in slider consumption per thousand units is a rework signal.
- Sample the repair bench physically at random times and record what is being fixed, rather than relying on retrospective forms.
Data source for the recorded portion is the rework ticket and the work order; for the unrecorded portion it is reconciliation and consumable issue records. Both are needed, because the recorded portion typically captures the severe cases and misses the frequent small ones.
Rework Rate and Rework Cost per Unit
Two derived metrics make the model usable in operations. Rework rate is reworked units divided by units produced, expressed as a percentage or in parts per million. Rework cost per unit is total rework cost divided by units produced, which allows comparison across product sizes where the labour content differs.
A third metric, rework cost as a share of conversion cost, is useful for finance because it expresses the same loss in a language that sits next to margin. All three should be reported by process step, not by department, so that the step creating the fault carries the cost.
Repair, Replace or Scrap
Once a unit cost exists, the decision becomes arithmetic rather than habit. Repair is preferred when the rework cost plus the residual risk of a repeat failure is lower than the replacement cost minus the salvage value of the failed piece. Scrap is preferred when the repair consumes more capacity than the margin on the unit can absorb, particularly during peak season when the constraint is machine time rather than material.
The decision rule should be written into the work instruction for each defect code, so that it does not depend on who is on shift. A written rule also prevents the common failure mode where cheap repairs are performed on expensive products and expensive repairs are attempted on cheap ones.
What the Model Changes
A costed rework model usually produces three surprises. The first is that sorting, not repair, is the dominant cost in a defect-heavy month. The second is that a small number of defect codes account for most of the spend, which makes the improvement target obvious. The third is that prevention spending already compares favourably with the rework it would remove, so the obstacle is not economics but priority.
YCC uses the model to compare rework cost per thousand units across product families, which keeps the comparison fair when order mix shifts between simple and complex zippers. The output is not a penalty for the repair team; it is a budget request for the process that generated the defects.