Rolling mills and forges: changing input, thickness and shape

Incoming material varies from batch to batch

What happens. Slabs, billets or coils arrive with different temperature, hardness and sometimes composition. The mill is set for an average. Whoever stands at the stand sees the thickness wander after a change of batch, and corrects by hand or lets the thickness controller catch up after the error has already happened.

What it costs. Thickness out of tolerance for the first metres, more trimming and downgraded material, and a rolling force that varies from one piece to the next. Operators also run a safety margin on thickness so that the tolerance is not violated.

Why the usual fix fails. A thickness controller only reacts once the error is measured, and at the exit of the stand that is late. A further sensor measures one property at one place. A table of settings per grade helps only as long as the incoming material stays close to the table.

How a fixed-price remote diagnosis checks it. From your stand logs (forces, speeds, gap, temperatures, thickness at the exit) I check how much of the thickness error comes from the material that came in. The fixed-price remote diagnosis answers whether a computed material state can drive the settings forward before the piece is in the stand, and what data that would need.

Background: Computing unmeasured states for control

The core temperature of the slab in the reheating furnace is unknown

What happens. Pyrometers and thermocouples show the furnace and, at best, the surface of the slab. What the core has reached stays a guess. The furnace is run by dwell time and recipe, and a stop in the mill or a cold slab from the yard changes the picture without anyone seeing it.

What it costs. Gas burned on slabs that were already hot enough, more scale on the surface, and rolling problems when a core is too cold: higher forces, uneven thickness, now and then a stopped line. The usual reaction is more time and higher temperature than needed.

Why the usual fix fails. A longer dwell time is safe but expensive, and it grows scale. A fixed heating curve ignores how a slab came in and how long the mill was waiting. Spot measurements in test slabs are rare and cover one situation.

How a fixed-price remote diagnosis checks it. From furnace logs (zone temperatures, gas flows, charging and discharging times, rolling forces) I estimate how the core temperature develops in each slab and how much margin the recipe really contains. The fixed-price remote diagnosis tells you whether a computed slab temperature can lead the furnace, and what that would take.

Background: Predictive control of process plants

Forging dies run too hot or too cold

What happens. The die temperature is checked now and then with a handheld device or by feel. In between it depends on the cycle time, the spray, the stops and the workpiece. After a pause the first parts go into a cold die, and in a fast run the die creeps up.

What it costs. Hot dies wear faster and soften. Cold dies cause cold laps, cracks in the part and a short life. Rejects and extra die changes cost material, time and tool money.

Why the usual fix fails. Spraying by timer fixes one cycle, not the changing rhythm of a shift. An additional temperature sensor in the die is hard to place and to keep alive. Preheating for a long time is safe, but wastes energy and press time.

How a fixed-price remote diagnosis checks it. From press and spray logs (cycle times, stops, spray settings, surface temperature readings, rejects) I check how well the die temperature can be computed from what you already record. The fixed-price remote diagnosis answers whether a computed die temperature can guide spraying and pauses, and what it would take.