Foundry · Moulding & core

The mould count is right. The rejects it produced are somewhere else.

Moulding is where a foundry’s rejects are made and where they are hardest to trace, because the defect shows up two stages later at knockout. By then the sand that caused it has been through the muller three more times. Munshi keeps the sand plant’s own readings alongside the reject the operator logged, on the same clock.

Your ERP logs the moulds: boxes made, patterns run.

It won’t tell you the return sand was still hot when it went back to the muller, or which pattern is quietly producing the blow holes.

Why the gap exists

Where the losses hide

The rejects that are made here and found later.

None of these stop the line. They surface at knockout, by which point the conditions that caused them are gone.

Sand that never cooled

Return sand going back hot changes moisture and compactability, and the mould that results is not the mould you set the parameters for. It is a reading at the sand plant hours before it is a blow hole at knockout.

Defects that belong to a pattern

Blow holes, inclusions and cuts cluster by pattern and by shift far more often than they spread evenly. Averaged across a month, that cluster disappears entirely.

Box turnaround and prep delays

Time lost between strip and re-set never appears as downtime, because the line was never technically stopped. It appears as fewer moulds than the shift should have made, with no reason attached.

The join

What the record holds, and what Munshi captures beside it.

Same run, same clock, same asset — three streams that only mean something together.

Your ERP records

Moulds produced, pattern run, sand consumed, and the rejects booked after knockout. Accurate, and silent on which condition produced them.

Munshi captures

Return sand temperature and muller current sampled continuously through the shift; mix cycle time per batch; and the operator’s reject log with the defect tagged at the point it was found.

What that makes possible

A reject rate you can read by pattern and by shift, with the sand-plant condition that ran alongside it already attached.

What Munshi surfaces

The cluster, and what was happening when it formed.

Teal is normal, maroon is flagged. Illustrative data.

Return sand temperature through a shift

The band is where the sand should return. The drift past it is the afternoon nobody logged, and the shift the blow holes came from.

Reject rate by pattern

Tagged by the operator at knockout. One pattern is carrying most of it.

Muller current, batch by batch

The same mix should not need more current. When it does, something upstream changed.

Normal / in-spec Flagged by Munshi -- threshold / average

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

The stages either side of this one.

Melting

Energy per tonne, holding time and chemistry rework, per heat.

Pouring

Temperature drift, misruns and ladle delays.

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.