Rough-cut capacity planning: sanity-checking the schedule before it costs money

Choosing the short list of key resources

The value of the technique collapses if the list grows. Include only resources whose shortage has genuinely delayed delivery: the constraint machine, a heat treatment or curing step, a licensed operator group, a shared test rig, sometimes a single supplier's capacity. Ask maintenance and supervision which resources they fight over, rather than deriving the list from a work centre file. Keep the list visible and dated, and revisit it when the mix or the equipment changes. A list of half a dozen resources reviewed regularly does more good than a comprehensive model refreshed once and then quietly abandoned.

Building a bill of resource that a planner can maintain

For each key resource, record how much of it a representative unit of each product family consumes, including a realistic allowance for setup at typical batch sizes. This is a coarse coefficient, not a routing: the point is arithmetic simple enough to redo whenever the schedule moves. Derive the coefficients from recorded output rather than from standards nobody has verified, and mark which are estimates. When a family's coefficient is contested, that argument itself is informative, because it usually means the family contains products with very different work content and should be split.

Reading an overload without over-reacting to it

A rough-cut overload does not mean the schedule is impossible; it means the schedule is unlikely without action. Look at the shape first. A single period spike surrounded by slack usually resolves by moving work earlier, which costs stock. A sustained overload across many periods is a structural capacity problem that no resequencing will fix, and it should trigger a decision about shifts, subcontracting or declining work. An overload that appears only under a particular mix points at a product family, not at the plant. The response differs in each case, so the pattern deserves a minute of reading before anyone reacts.

Where this fits against detailed capacity checking

Rough-cut runs on the master schedule, before material planning generates orders. Detailed capacity requirements planning runs afterwards, on the resulting works orders, across all work centres and with proper routings and lead time offsets. Both have a purpose: the coarse check prevents a bad schedule from spawning thousands of purchase and works orders, while the detailed check tunes the execution. Plants that only run the detailed version find out too late, when cancelling the consequences is expensive. Plants that only run the coarse version keep discovering local congestion at work centres that never appeared on the key resource list.

Turning the check into a decision, not a report

The output belongs in the schedule review with the people who can act: authorise overtime, move a customer date, release a subcontract order, or reduce the schedule. Name the decision and record it against the period it affects, then check at the next review whether the action was taken and whether the overload actually materialised. That feedback loop is what keeps the coefficients honest, because repeated false alarms mean the model overstates consumption, while repeated surprises mean a resource is missing from the list. Without the loop, the analysis becomes a monthly document that circulates and changes nothing.

Frequently asked questions

Can a small plant do rough-cut capacity planning on a spreadsheet?

Yes, and many should. The method needs a family-level schedule, a coefficient per key resource per family, and available hours per period. That fits comfortably in a spreadsheet maintained by the planner, and the discipline of updating it by hand keeps the resource list short. Move to something more automated only when the number of families or the frequency of schedule changes makes manual refresh the reason the check stops being done.

Should setup time be included in the resource coefficients?

Include it as an allowance based on typical batch size, because on constraint resources setup often consumes a substantial share of available hours and excluding it makes every schedule look feasible. Where batch sizes vary widely within a family, either split the family or carry a separate setup line driven by the expected number of runs. If the plant is actively changing its changeover practice, revisit the allowance rather than leaving the old figure in place.

What if the rough-cut check keeps saying we are overloaded but we always deliver?

Investigate before dismissing it. Either the coefficients overstate consumption, the available hours understate what the plant works, or the plant is genuinely absorbing the overload through unrecorded overtime, deferred maintenance or stock built earlier. The third case is common and matters, because the absorption has a cost that nobody is attributing. Compare a few periods of predicted load against recorded hours on those resources and correct whichever side of the comparison is wrong.