Planning the tunnel oven as what it is: a work center that runs better if you never stop it.

The problem

The plan is built in a spreadsheet looking at demand — and the oven picks up the bill.

In a tile plant, stopping the kiln is catastrophic: thermal shock cracks the refractories and the invoice is enormous. A bakery tunnel oven does not go that far — it runs at 180-300 °C and stopping it does not break anything all at once — but every start-up has a real cost that almost no plant adds up in one place:

1. Reheating gas. Bringing the oven back up to temperature takes 1 to 3 hours depending on size — burning fuel without a single saleable piece coming out.
2. Transient product. Until the temperature profile settles, the first trays come out pale or over-baked. In breadsticks and crispbread, where color is a retailer specification, that is straight scrap.
3. The wear you do not see. In hearth ovens, every thermal cycle fatigues the refractory. It is not the tile plant's sudden failure, but it shortens the life of the hearth and brings the long maintenance shutdown forward.

And who decides how many times the oven starts up in a month? The production plan. A plan built in a spreadsheet, looking at demand SKU by SKU, tends to fragment: short campaigns, frequent changeovers, line gaps that end in an oven stop. Not because the planner does not know better — they know perfectly well — but because with a spreadsheet they barely get to one viable plan, and comparing alternatives that group campaigns better would take another three days they do not have.

How it fits the IRIS system

Connect absorbs your spreadsheets; the Agent builds the plan that respects the oven.

The iLEAN Planning Agent was born planning work centers that genuinely cannot stop — continuous work centers in heavy industry where continuity is a hard constraint. That same machinery, applied to a bakery tunnel oven, is dialed down: strict continuity if the plant decides so, or a heavy penalty per start-up so the engine minimizes them without forbidding them.

The user drops in their spreadsheets. The agent builds the model of the oven, its lines and its SKUs, and returns an editable receipt. The engine solves with the oven's constraints. The human compares three scenarios and approves with a stamp.

The specific iLEAN piece for the oven of a bread, breadstick or crispbread plant:

• Connect — absorbs the chaotic documents the plant already has: the production spreadsheet with its thousand tabs, the sales demand that arrives by email, the rates by SKU and line. No mandatory template and no prior integration project.
• Planning Agent — builds the planner's database by itself (work center, lines, items, constraints, demand), asks only when something blocking cannot be derived from the documents, and solves with a constraint optimization engine. For the oven it applies the constraints that matter: minimum campaign length, no job changes on public holidays and weekends, each product's affinity with its preferred line, and continuity or start-up penalties depending on how it is configured.
• Three scenarios, not one plan — maximum output, tight inventory and stability (long campaigns, minimum changeovers — the natural scenario when the expensive work center is an oven). Side by side, with KPIs and the differences explained one by one.

The approved plan is exported to the spreadsheet with the production program format the plant already uses. The team's routine does not change on day one — what changes is how long it takes to reach a good plan, and how many times the oven starts up for nothing.

See the full IRIS architecture →

Before and after

Spreadsheet plan vs. plan with the Planning Agent

Aspect	The planner's spreadsheet	With the Planning Agent
Time to build the monthly plan	Days	One morning, with the agent building the model
Oven start-ups and stops	Whatever comes out of fitting it by hand	Minimized by the engine (constraint or penalty)
Campaign length	Fragmented by urgent orders	Long campaigns with a configured minimum
Changeovers on holidays / weekends	They slip in and somebody suffers them	Blocked by a soft constraint
Strategy alternatives	A single viable plan	3 scenarios compared with KPIs + what-ifs
Replanning after an unforeseen event	Another several days	Minutes: recalculated and compared
Planning knowledge	In one person's head	In the model, with plant memory
Format of the approved plan	The usual spreadsheet	The very same spreadsheet — simply exported

Impact estimate

Impact estimate for your oven — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Bread, breadstick or crispbread plant with 1-2 tunnel ovens, multi-SKU, monthly plan built today in a spreadsheet by one person.
• First planning session with the agent on the existing documents — with no prior data project. First comparable plan in one morning.
• Expected reduction of oven start-ups and stops of ≥30% against the manual plan, simply by grouping campaigns the way the engine finds and the spreadsheet does not. (Conservative range — estimate to be validated.) Every start-up avoided is reheating gas and transient trays that do not go in the bin.
• Indicative payback between 4 and 9 months, depending on the oven's consumption, the cost of transient scrap and the planning hours freed up.

The defensible technical anchor comes from the planner's most thoroughly documented real case: a high-demand industrial plant in heavy industry — with a strictly continuous furnace — whose agent built the complete model by itself (one oven with 3 lines, 128 items with their production rates, 5 operational constraints and ~4,600 metric tons of demand) from a 114 MB zip with an 18-tab spreadsheet, and delivered 3 optimized scenarios in under an hour. The same class of continuity constraint, dialed down to what a bread oven needs.^[1]

And the operations director's reasonable doubt

“What if the plan the AI proposes is not executable in my plant?” — that is why there is human verification at every gate. The intake receipt shows everything the agent has understood from your documents and is corrected inline before anything is solved; the scenarios are compared with the differences explained; and the plan only goes into production when the responsible person approves it with a stamp. If the demand does not fit in full, the system does not say “infeasible”: it delivers a partial plan with its coverage figure and flags exactly what is left out, so a person can decide. The AI proposes; the plant signs.

^[1] Real iLEAN case, high-demand plant in heavy industry — verified pilot figures: 114 MB of documents, 18 tabs, 128 items, ~4,600 t, 3 scenarios in <1 h, 99.5% coverage.