Casting Rejection Analysis: A Defect-wise Pareto Method
Rejection Pareto
"Rejection is 6% this month" starts a meeting but doesn't end one. To reduce rejection you need to know which defect, on which pattern, from which heat or shift. Here's a method any foundry can run monthly.
Step 1: Measure rejection the same way every time
Weight is better than count for costing, because one large rejected casting can outweigh fifty small ones. Track internal rejection (found at knockout, fettling, inspection or machining) separately from customer rejection; the second costs far more.
Step 2: Record every rejection with its context
For each rejected casting, capture:
- Item or pattern, and quantity and weight
- Defect type, from a fixed list (see below)
- Heat number, mould line and shift
- Stage where it was found: knockout, fettling, inspection, machining or customer
Use a standard defect list so "blow hole", "blowhole" and "BH" don't become three categories. A typical starting list: sand inclusion, blowhole, gas porosity, shrinkage, cold shut, misrun, crack or hot tear, slag inclusion, mould shift or mismatch, dimensional, and hardness or chemistry out of specification.
Step 3: Build the Pareto
Sort defects by rejected weight, largest first, and add a cumulative percentage. An illustrative month with 18 tonnes rejected out of 300 tonnes produced (6%):
| Defect | Rejected (t) | Share | Cumulative |
|---|---|---|---|
| Sand inclusion | 6.1 | 33.9% | 33.9% |
| Blowhole | 4.3 | 23.9% | 57.8% |
| Shrinkage | 3.2 | 17.8% | 75.6% |
| Cold shut | 1.8 | 10.0% | 85.6% |
| Mismatch | 1.2 | 6.7% | 92.2% |
| Others | 1.4 | 7.8% | 100% |
Three defects make up three-quarters of the loss. That's where the month's effort goes.
Step 4: Drill into the top defect
Break the top defect down by pattern, heat and shift. The pattern of clustering points to the cause:
- Clusters on certain heats point to the melt: charge materials, gas pick-up, pouring temperature.
- Clusters on certain patterns point to the method: gating, venting, feeding.
- Clusters on a shift or mould line point to sand and moulding practice: moisture, compactability, ramming.
Step 5: Put a cost on it
A casting rejected at machining has absorbed fettling, heat treatment and machining cost that one rejected at knockout never did. Weight each rejection by the value added up to the stage where it was found, and the priority list often changes.
Step 6: Act, then re-measure
Agree one action per top defect, with an owner and a date. Re-run the Pareto next month on the same basis. Because every rejected casting goes back into the furnace, cutting rejection also lifts casting yield.
Rejection analysis in ICAST
ICAST captures rejections on the shop floor, from a phone, against the item, heat and stage, using your standard defect list. Defect-wise, pattern-wise and heat-wise analysis is a report, not a weekend of Excel. The ICAST AI Engine goes a step further, using your daily production data to flag rejection risk before it happens.
