Use case · Manufacturing
Dynamic production scheduling with what-if scenarios
Half a working day goes into planning the coming week, every day. At ten a line goes down, and the work starts over — changeover sequence, material, due dates, all of it tied together. We recalculate the plan in seconds and show what it costs before anyone releases it.
Talk about your case →N 01 — Where it breaks
The plan holds until the first phone call.
You plan once and get disrupted daily. And because everything on a shop floor hangs together, no disruption stays local: a breakdown on line 2 shifts the changeover sequence, the sequence shifts the material call-off, the call-off breaks a delivery date. That is why replanning takes as long as planning.
Machine
The line is down
Unplanned downtime, or maintenance running longer than scheduled. Orders have to move to other lines or other shifts — where they trigger different changeovers than planned. The lost hour is rarely the expensive part; what it destroys in sequence is.
Staff
Two people short on early shift
A machine without an operator is not capacity. A sick note does not hit the machine, it hits the qualification: who is allowed to run which line, which shift is left short — and what can actually be swapped.
Customer
The rush order on the phone
A commitment made in two minutes whose consequences nobody can see. Squeezing it in always works. The question is which three orders finish late instead, what that adds in changeover time — and whether one of the three belongs to a bigger customer.
In the end it is not the best plan that wins, but the one somebody could still calculate at ten in the morning.
N 02 — What is taken into account
Four constraints, without which any new plan is fiction
A due date can be moved in any calendar. Whether it holds is decided by four quantities — and the same calculation has to know all four, otherwise it produces pretty bars with nothing behind them.
Changeover times
Sequence is the price. Switching between two variants costs hours that are booked to no order as production; with a wide product range, changeovers exceed the manufacturing itself. Anyone who cannot see the two apart looks for the missing capacity in the wrong place.
Material availability
What is not in the building does not run. Every counter-plan is checked against stock, open purchase orders and lead time — including the uncomfortable question of whether pulling an order forward only works because it takes material away from another one.
Due dates
Not everything can be on time. So the system states what breaks first: which order runs how many days late, and what revenue hangs on it. Priority by due date, ties broken by value at risk — readable afterwards, not decided by gut feel.
Shift and qualification
Capacity is not the machine hour, it is the machine hour with somebody in front of it who is cleared to run the line. Shift model, attendance and qualification enter the same calculation as the equipment.
N 03 — What-if
Calculate first, commit second
The disruption is applied, not entered: the live plan stays untouched, and the calculated counter-plan appears beside it. What gets compared is not one gut feeling against another, but plan against plan.
Apply the disruption
“Line 2 is down for eight hours.” “Two operators missing tomorrow morning.” “400 units by Thursday, customer X.” In plain words, in seconds, without anyone touching a formula.
Recalculate
Backwards from the due date, against all four constraints. Every shift states which constraint forced it — material, changeover, shift or date. Traceability is not an extra here; it is the condition for the plan being used at all.
Compare side by side
Old and new plan in one view, the difference highlighted. What stays, what slips, what no longer fits at all — and several variants of the same disruption next to each other.
The price of the decision
- On-time delivery — which order runs how many days late
- Revenue at risk in euros, not in traffic-light colours
- Additional changeover time consumed by the switch itself
- Missing material — with the date it would arrive
Plus the most uncomfortable figure of all: what fits into no variant at all. A plan in which everything always works out is hiding something.
From a project
How this looked at a food processor
Demand planning, purchasing and production scheduling ran on scattered spreadsheets and one-off queries. Planning the cutting floor was a morning task — until a disruption came in, at which point it became a full-day task. What we built was the chain all the way into production: daily forecast per article, demand derived from it, and the line schedule with changeover times stated separately — plus the open list of what no longer fits into the week.
The disruption simulation described here is built on that same foundation and is still under construction. References are anonymised at our customers' request; names and figures we share in conversation.
N 04 — What it is not
Three promises we do not make
All three would sell well. They just do not hold — and what does not hold shows by the third month at the latest.
→The proposal is calculated; the decision belongs to your planner. Nothing reaches the shop floor that nobody released.
→A result nobody can explain gets worked around within three weeks. We calculate traceably: every shift has a reason the planner can read. Where real optimisation measurably does better, it goes underneath — as a method, not as a sales line.
→If the week is not enough, the system says so instead of massaging the utilisation figure. What does not fit stands as its own list — with the value attached to it.
N 05 — How we get there
Connected to what is already there
This is not a second system next to the ERP, it is the layer missing between the ERP and the shop floor. We read from your systems without intervening. It runs on your hardware.
Weeks 1–2
The data situation
Orders, bills of material, routings, changeover matrix, shift model, stock — read from ERP and MES. What is missing is named as a gap rather than estimated.
Weeks 3–6
First calculation
Line scheduling with changeover times and a material check against real data, replayed against past weeks: would this plan have held?
Weeks 7–12
Disruptions and scenarios
Breakdown, staffing and rush order as applicable cases, comparison and valuation — and the first productive version, operated by your planners.
A clickable demo of this capability is coming to demo.wayfinder-ai.de shortly. Until then we show it live in conversation.
Write an email →support@wayfinder-ai.de