Fettling absorbs the mistakes of every stage before it. A casting with excess flash, a mismatch or a sand inclusion still leaves as a good casting - it just costs more labour and more consumable to get there. That extra cost lands in a department budget, not against the process that caused it.
Your ERP logs the pieces fettled and the consumables issued.
It won’t tell you which castings came back for a second pass, or which upstream stage is sending them.
Why the gap existsWhere the losses hide
Fettling rarely reports a problem. It absorbs one, and the absorption shows up as hours and abrasives.
A casting that goes round twice is one good casting in the output figure and two passes in the labour. The second pass is invisible unless it is logged as one.
Wheels, discs and belts are issued to the department and consumed against whatever happens to be in front of them. Per-casting cost is the number that would show drift, and it is the one nobody has.
Grinding past the line turns a good casting into a dimensional reject at inspection - scrap created by the operation meant to save it.
The join
Same run, same clock, same asset — three streams that only mean something together.
Pieces fettled, consumables issued to the department, hours booked.
The rework pass logged as a rework, the reason tagged to the upstream stage that caused it, and consumption tied to the machine and the catalogued item.
Rework hours and consumable cost per casting - and the upstream defect they trace back to, which is where the fix actually is.
What Munshi surfaces
Teal is normal, maroon is flagged. Illustrative data.
The reason is tagged at the bench, so the hours land against the stage that caused them rather than against fettling.
Normalised per casting, so a busy week does not read as a problem. The spikes are the batches that came back.
When an outcome moves and a signal moves with it, Munshi shows both and says they moved together. It does not tell you one caused the other. The operator confirms the cause, and the confirmed cause is written back to that asset’s register — so the pattern is already named the next time it appears.
Related stages
Temperature drift, misruns and ladle delays.
Abrasive consumption and wear-driven throughput loss.
Get started
See this on your own line.
Bring a month you already understand. The useful test is whether Munshi finds what you already know — and then what you didn’t.