Eyrix — Blog
Finite capacity scheduling: why your ERP plans against capacity you don't have
Most ERP systems plan production the way classic MRP always has: take the demand, offset each order by a fixed lead time, and place it on the calendar. What that calculation never asks is whether the line can physically produce everything landing in the same week. It assumes infinite capacity — and so it hands you a plan that looks complete and is quietly impossible.
The gap: a plan that ignores the constraint that actually binds
A fixed lead time is a guess that a work order takes, say, fourteen days regardless of what else is in the queue. But a line has a finite number of hours per shift, a calendar with downtime, and one order at a time on each machine. When demand rises, the ERP happily schedules three orders into a week that holds two — and reports the plan as feasible. The overload doesn't disappear; it just moves to the place it's most expensive to discover: the shop floor, mid-week, when the line can't keep up.
What finite capacity scheduling does instead
Finite capacity scheduling — the core of what the industry calls APS, advanced planning and scheduling — sequences operations against the resources that actually run them. Concretely, it accounts for:
- Real resource limits. One operation at a time per single-capacity machine; parallel units where a line genuinely has them.
- Working calendars. Shift patterns, holidays and planned downtime as non-working time the schedule has to route around, not naive calendar days.
- Routing precedence. Setup, run time per unit, queue and move time between steps — so the duration reflects the actual batch size, not a flat lead time.
- The bottleneck. Most plants are paced by a handful of constraining resources. Schedule those properly and load-check the rest; that's how real schedulers think, and it's what keeps the problem solvable at scale.
The output isn't just dates — it's a sequence that fits, plus an explicit flag wherever a week is over capacity and why.
"Feasible" has to mean feasible
The difference is sharpest when you compare two plans for the same demand spike. One says make everything in-house and shows a line at 100%+ in weeks 32–34 — a hard constraint violation. The other finds the feasible path: keep the lines inside their limits and quantify what has to move. A finite scheduler won't propose the first plan at all, because it can see the wall the ERP can't. Deciding what to do about that overload is the next question — usually a make-or-buy trade-off the plan should quantify rather than leave to gut feel.
Why this matters more the moment demand moves
An infinite-capacity plan is wrong precisely when you most need it to be right: when demand shifts and the load picture changes. That's the case for planning demand, capacity and sourcing together rather than in separate files — a plan that respects the capacity you actually have is the only one worth committing to the floor.