Heat treatment is energy and time against a specified curve. A cycle that ran cold at one end, held short, or was opened early does not fail visibly - it produces castings that fail hardness later, and a re-treat that costs a second full cycle. In the ERP that is one batch with a note.
Your ERP logs the batch: pieces in, cycle run, energy booked.
It won’t tell you the furnace sat below setpoint for forty minutes, or that this batch is here for the second time.
Why the gap existsWhere the losses hide
The difference between them is where re-treats come from - and a re-treat is close to double the energy for zero extra output.
Soak temperature and soak time are the specification. A cycle that drifts below it, or is cut short to free the furnace, produces parts that will not meet hardness - and the drift is only visible if the furnace’s own readings were kept.
A batch that comes back is charged, heated and soaked twice. One batch in the count, close to two batches of energy and furnace time.
Door seals, loading practice and lining condition all move the energy a cycle takes. The batch count stays flat while the kilowatt-hours behind it climb.
The join
Same run, same clock, same asset — three streams that only mean something together.
Batches treated, pieces in each, energy booked to the department.
Furnace temperature against the specified setpoint through the whole cycle, the operator’s log when a batch is re-treated and why, and energy tied to the furnace.
Energy per batch, deviation from the curve, and a re-treat rate with the reason attached - rather than a batch count that treats every cycle as equal.
What Munshi surfaces
Teal is normal, maroon is flagged. Illustrative data.
The shaded stretch is time spent under the specified soak. The parts do not know it was only forty minutes.
The two outliers are the same batch twice. A re-treat is not a rounding error.
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
Abrasive consumption and wear-driven throughput loss.
Defect Pareto by cause, and the return-to-cause trace.
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.