Integrated steel plants measure each shop carefully. The trouble is that the margin is lost between the shops, where nobody has a number. We think the plant needs one queue, ranked by cost, across all of them.
An integrated steel plant is a chain of shops: blast furnace, converter, ladle, caster and mills. Each one has targets, and most of them are measured well. The converter tracks tap-to-tap time. The caster tracks sequence length. The mills track throughput. It is entirely possible for every shop to have a good day on its own numbers while the plant has a bad one: a caster breaks sequence because heats arrived in the wrong order, and a customer order slips.
A busy restaurant kitchen has the same shape. The grill station can be fast, the fryer can be fast and the sauce station can be fast, and the table still gets a cold plate because the three parts were not ready at the same moment. Every station hit its target. The diner did not get their dinner. A good head chef does not run the kitchen by asking each station how quickly it worked. They watch the tickets, the orders that need the parts to come together, and call the timing across stations.
In a steel plant, the information each shop needs about the others lives in different systems: process data in the historian, heat and slab genealogy in MES, chemistry and holds in LIMS, orders in ERP, asset health in EAM. They are already measured, but in systems that do not share a clock. So each shop optimises the part it can see.
| What each shop sees | What it cannot see | What it costs the plant |
|---|---|---|
| Converter: its own tap-to-tap | Whether the caster needs this grade next | A heat that waits, or a sequence that breaks |
| Caster: its own sequence | A chemistry hold forming upstream | A slab cast that will be held at the gate |
| Quality: today’s holds | Which sales order each hold affects | A regrade decided without knowing the order at risk |
| Energy: yesterday’s report | Which unit and shift drifted | An overrun nobody could correct in time |
The fix we argue for is not another dashboard per shop. It is one ranked queue of what needs a decision, across the plant. In Steel Twin, every heat is traced from charge to coil, joined on heat_id across MES, LIMS, ERP and the historian, with its grade, sales order, route, chemistry and stage. Holds, constraints and energy variances are ranked by cost and urgency in one action queue, and alerts reach the shift controller while the shift can still act.
Ranking by cost is the important part. A plant has more problems at any moment than people to solve them. If the queue is ordered by which shop shouted loudest, the converter’s issue gets fixed while a hold that puts an export order at risk waits. When each item shows the rupees at risk, the shift controller can make the call the head chef makes: what matters most for the order, not for the station.
Shop numbers are useful, and nobody should stop tracking tap-to-tap or sequence length. The point is that they are inputs to a plant decision, not the decision itself. A converter target that is met by sending heats the caster cannot use is not really met.
That sequence matches how Steel Twin is delivered: steelmaking and casting first, then energy, asset health and iron-making, then mills and dispatch. For how heats are traced, read heat genealogy in steel plants, and for how a twin control room differs from MES, see steel plant MES vs a digital twin control room. More on the Steel Twin guides page.
Because each shop optimises what it can see in its own system. Problems at the boundaries, such as heats arriving in the wrong order for the caster, are invisible to any single shop.
One list of holds, constraints and variances across shops, ranked by cost and urgency. In Steel Twin each item is linked to the heat, grade and sales order it affects.
At the boundary that causes most disputes, usually steelmaking to casting. Steel Twin starts there and extends to energy, assets, iron-making and mills.
See it on your own data. Steel Twin — Every heat, traced to the coil. Book a 30-minute working session with an engineer.
