Today
An induction furnace, a ladle or a rotary kiln is relined when the furnace master thinks it is time. Too early wastes refractory and production time; too late risks a breakout, in the worst case molten metal at the water-cooled coil. Wall thickness measurements, heats, temperatures and patch repairs of each campaign sit in different systems, so the decision stays a matter of a few people's experience.
An example
In the foundry of Rabenhammer Präzisionsteile, induction furnace 2 has 312 heats behind it in its current campaign. Until now it was relined by experience at around 350 heats. The agent links the laser scans taken every weekend with tapping temperatures and patch repairs from the last eight campaigns: in the slag zone the wall has shrunk from 110 to 52 mm, faster than before, since housings in EN-GJS-400-18-LT are tapped more often at 1,520 instead of 1,480 °C. At the current pace it reaches the minimum of 45 mm after heat 334. The agent proposes relining on the weekend before and checks that 2.6 t of ramming mass are in stock. The head of the melt shop sets the date.
Your data, wherever it lives
The state of your furnace and ladle linings is spread out today: heats, temperatures and power in the furnace control, laser scans and thermal images as files and photos, relining records in spreadsheets, patch repairs in the maintenance tickets, material and cost in the ERP. demi reads, understands and links them per furnace and campaign, with your rights. The AI works out in minutes when the minimum thickness will be reached, and your melt shop sets the date.
Where the data lives today
- Machines & sensors
- Photos, video & audio
- Office files
- Tickets
- ERP
How demi solves it
Three layers in this order: connect the data first, then the rights, then the AI.
- 1 · ConnectData Layer
Data sources in this case: laser scans and thermal images of the furnace wall, the furnace control with heats, temperatures, frequency and power, relining records in spreadsheets, patch repairs in the maintenance system, refractory material and cost in the ERP. demi replaces nothing and links it all per furnace or ladle and per campaign. For every lining you see how fast it wears and why.
- 2 · PermitPermission Layer
Melt shop, maintenance and plant management see wear data and forecasts; only purchasing and management see cost and supplier prices for refractory. The responsible person in the melt shop sets the relining date; the agent's proposal and that decision are recorded.
- 3 · ActIntelligence Layer
The agent tracks residual wall thickness per zone over the campaign, calculates the wear rate per heat from the measurements and compares it with earlier campaigns. It shows which factors drive wear, such as tapping temperature, slag volume or holding times, and calculates when the minimum thickness will be reached. Your melt shop sets the limits; the language model only explains the proposal.
How it runs
- 1
After each measurement, demi reads the laser scan, thermal image and furnace data.
- 2
The agent assigns the values to furnace, zone and campaign and projects the residual wall thickness.
- 3
It compares the wear with earlier campaigns and shows what speeds it up.
- 4
It proposes a relining date, with material, team and shutdown window.
- 5
Melt shop and maintenance review the proposal and set the date.
What changes
Relining at the right time instead of by feel
Lower breakout risk and less refractory thrown away
The furnace master's experience captured in data per campaign
What does it bring you?
Check it with your own numbers
Every starting value is an assumption you can change. The calculation sits right below, step by step.
Your numbers
Result
+54 hfreed per year
Assumptions and calculation
demi links measurements and furnace data and projects the wear; you set the relining date.
45 min → 31.5 min per case
How we calculate
20 wear assessments × 45 min × 30% × 1.0 (manual) × 12 months = 3,240 min = 54 h
Value per year, before demi costs
Freed per year+54 h
Assumption: labour cost of €49.50 per hour in German manufacturing in 2025 (Destatis), plus workplace, IT and overheads.
= Value per year, before demi costs
+€3,510
gained per year
An estimate, not a promise. The starting values are assumptions for an industrial company with 150 to 500 people.
Combine with more cases and get the packageQuestions about this use case
Where do the residual wall thicknesses come from?
From the measurements you already take: laser scans of the furnace wall, thermal images of the shell, readings taken during patch repairs or the furnace master's notes. Whether they are files, photos or spreadsheets, the agent reads them and assigns them to furnace, zone and heat. How the measuring devices are connected, we clarify in conversation.
What if we do not measure the furnace wall at all?
Then the agent works with what exists: number of heats, temperatures, electrical values from the furnace control and the condition at earlier relinings. The forecast is coarser then, and it says so. It also tells you which measurement would best back up the decision.
Does this work for ladles, rotary kilns or boilers too?
Yes. The same principle applies to every bricked or rammed unit: track wear per zone over the campaign, link it to operating data and compare it with earlier campaigns. Your specialists set the zones and minimum thicknesses per unit.
Who decides when to reline?
The melt shop. The agent proposes a date with its reasons and shows which measurement and which earlier campaign it rests on. If a measurement falls below the minimum thickness, it alerts the responsible person at once instead of waiting for the next report.
