Scale Loss

Scale loss is the metal lost as iron oxide when steel is heated in an oxidising atmosphere. The oxide forms on the surface, flakes away during handling and forging, and is not recovered.

It is a genuine loss of material. The weight coming out of a forging stage is lower than the weight that went in, and the difference is not a counting error — it is steel that has left the process as scale.

Why it has to be modelled, not estimated#

Any system that reconciles stage inputs against stage outputs will flag a difference at every forging operation. If expected losses are not part of the model, every heat produces an exception, and exceptions that always fire are exceptions nobody reads.

Treating the loss as expected has the opposite effect: the difference between actual and expected becomes the signal. A stage losing noticeably more than usual is telling you something specific — that time at temperature has crept up, that furnace atmosphere has changed, that material is sitting soaking before it is worked.

Why it should not be pooled with other losses#

Scale loss, burning loss and cut loss are frequently rolled into one yield figure. That is convenient and diagnostically useless, because the three have different causes and different fixes:

• Scale loss is surface oxidation, driven by temperature, atmosphere and time.
• Burning loss is material lost to overheating beyond scaling, and points at furnace control or scheduling.
• Cut loss is material removed by cutting — the saw kerf plus unusable ends — and points at the cutting plan and saw setup.

Recorded separately, each one is attributable to a stage and a cause. Merged, a rising total says only that something has got worse.

Where it interacts with inventory#

The practical consequence is that a forging stage cannot be modelled as a simple move of quantity from one place to another. It consumes an input weight, produces an output weight, and accounts for the difference. A system that only supports transfers will either report phantom shortages or quietly let quantities drift.

The same applies to traceability. Losses reduce quantity but must not obscure identity: the reduced output is still material from a specific heat, and the loss is recorded against that heat rather than against an anonymous total.

What it is not#

Scale loss is not scrap, which is material rejected for a defect rather than consumed by the process. It is also not a quality problem in itself — some scaling is unavoidable in hot work. It becomes a problem when it is larger than the process warrants, which is exactly the comparison a separately recorded figure allows.