How to calculate scrap rate
Scrap rate is the share of the units you made that ended up in the scrap bin:
Scrap rate = units scrapped ÷ units produced × 100
Rework rate = units reworked ÷ units produced × 100
First pass yield (FPY) = (produced − scrapped − reworked) ÷ produced
“Units produced” means everything made in the period, good and bad. Dividing by good units gives a slightly higher figure and muddles month to month comparisons.
How to calculate the cost of poor quality
The calculator adds up the four failure costs a plant can actually count:
Scrap = units scrapped × cost per scrapped unit
Rework = units reworked × rework cost per unit (or time × labor rate)
Sorting = sorting and re-inspection hours × labor rate
External failure = returns, warranty and chargebacks for the month
Cost of poor quality = scrap + rework + sorting + external failure
The yearly cost is the monthly figure × 12, and cost per good unit spreads the month’s total over the units left: what each unit you ship carries in failure cost.
The four costs of quality (PAF model)
Armand Feigenbaum’s PAF model (prevention, appraisal, failure), along with Joseph Juran’s point that failure costs are “gold in the mine”, splits quality costs into four groups:
| Category | Examples | Type |
|---|---|---|
| Prevention | training, FMEA, poka-yoke, preventive maintenance | cost of quality |
| Appraisal | inspection, testing, gauge calibration | cost of quality |
| Internal failure | scrap, rework, sorting, re-inspection | cost of poor quality |
| External failure | returns, warranty, replacement freight, chargebacks | cost of poor quality |
Prevention and appraisal are spent on purpose, so defects do not happen or do not reach the customer. Internal and external failure are the price of defects that did. This calculator covers only the failure costs, the ones a plant can count each month. Overtime to make up output, lost customers and reputation are real too, but hard to measure, so the result here is a floor, not the full bill.
Worked example
The values preloaded in the calculator are for a stamped and welded part, in a month with:
- 20,000 units produced
- 360 scrapped and 900 reworked
- Cost per scrapped unit: $8
- Rework: 6 min per unit at a labor rate of $30/h, so $3 per unit
- 40 h of sorting suspect lots
- $1,500 in returns and warranty
The math:
- Scrap rate = 360 ÷ 20,000 = 1.8%
- Rework rate = 900 ÷ 20,000 = 4.5%
- First pass yield = (20,000 − 360 − 900) ÷ 20,000 = 93.7%
- Scrap = 360 × $8 = $2,880
- Rework = 900 × $3 = $2,700
- Sorting = 40 h × $30 = $1,200
- External failure = $1,500
- Cost of poor quality = $8,280 a month, or $99,360 a year
- Per good unit = $8,280 ÷ 19,640 = $0.42
A 1.8% scrap rate sounds small, yet it is 35% of the cost. And rework, which many plants never log, is close behind at 33%.
How to read the result
- Look at the split, not just the total. The biggest slice of the bar is where improvement pays back fastest.
- High external failure means defects are getting past inspection. Beyond the money, it hurts the customer relationship.
- High rework with low scrap often hides an unstable process the team “saves” at the end of the line. First pass yield shows it; final yield does not.
- To bring it down, break scrap out by defect type in a Pareto chart, go after the vital few and track the process DPMO and sigma level. Quality loss is also the third factor of OEE.
Common mistakes
- Using the selling price as scrap cost. The cost is the material plus the processing already spent. Lost margin only counts if the scrapped unit leaves a sale unfilled.
- One cost for all scrap. A unit lost at final assembly carries every step before it. If scrap happens at different points, use a weighted average cost.
- Double counting. A unit that was reworked and then scrapped counts only as scrap.
- Leaving out sorting and re-inspection. Hours spent sorting a suspect lot are a failure cost, even if nobody books them as scrap.
- Calling the result the total cost of poor quality. It is what you can count. What does not show up, like a lost customer, is usually bigger.