Capacity Planning

Running Rough Cut Capacity Planning as a Weekly Discipline

A master schedule loaded past capacity is not a plan, it is a promise to fail. Here is how to run RCCP so the MPS you release is one the floor can execute.

A master schedule loaded to 115 percent of capacity is not ambitious, it is fiction, and fiction is expensive. The plant that releases it will ship maybe 82 to 88 percent of it, chosen by whoever screams loudest rather than by margin or strategy. Overtime gets thrown at the gap at 1.5 times labor, expedite freight climbs to 2 or 3 percent of sales, and past-due backlog snowballs because next week starts with this week's leftovers. Rough cut capacity planning is the 30-minute check that catches this before release. Skipping it means discovering the overload on the floor, where fixing it costs 5 to 10 times more than fixing it in the schedule.

The mechanics are deliberately coarse. Rough-cut utilization equals required hours from the master schedule divided by available work-center hours, times 100. Work the bottleneck first: suppose the weld cell needs 1,480 hours over the next 4 weeks based on MPS quantities times hours per unit from the bill of resources. Available time is 2 stations running 2 shifts, 5 days, 8 hours, 4 weeks, which is 1,280 gross hours; apply 85 percent demonstrated efficiency and you have 1,088 real hours. That is 136 percent utilization, and the schedule is dead on arrival. The Rough Cut Capacity Planning calculator gives you the percentage; your job is deciding which lever closes the 392-hour gap.

Set planning bands and hold them. Load the bottleneck between 85 and 100 percent; a 5 to 15 percent buffer absorbs the demand mix shifts and absenteeism that always arrive. Between 100 and 110 percent, the plan is executable only with named countermeasures such as approved overtime or a qualified offload, written down before release. Above 110 percent, the MPS must change, full stop. Below 75 percent you have a cost problem, and it is worth pulling demand forward or cutting a shift. Check the top 3 to 5 constrained work centers only; rough cut across all 40 centers just buries the signal in noise.

Five levers close an overload, in order of cost. First, re-sequence the MPS: sliding 200 hours of non-committed demand out 2 weeks costs nothing. Second, alternate routings: qualifying the old brake press for 30 percent of the part family adds 250 hours a month. Third, overtime: two Saturdays add 128 hours at the weld cell but at 1.5 times labor, so it is a bridge, not a plan. Fourth, offload to a vendor, typically at 15 to 40 percent cost premium plus 2 weeks of qualification. Fifth, add staffing or shifts, a 60 to 90 day lever that only S&OP visibility makes possible.

The failure modes are predictable. Planning against rated capacity instead of demonstrated is the biggest: nameplate says 1,280 hours, but 18 months of history says the cell delivers 1,050, and the 20 percent gap becomes weekly past-dues. Ignoring setup is next; on a cell running 12 changeovers a week at 45 minutes each, that is 9 hours, roughly 8 percent of capacity, invisible in a bill of resources built on run time only. Averaging across a department hides the constraint: five machines at 70 percent and one at 130 looks like 80 percent overall while the bottleneck drowns. And stale bills of resources drift 10 to 20 percent within 2 years if never re-timed.

Cadence makes RCCP real. Weekly, before MPS release, the master scheduler runs rough cut on the constrained centers and brings any center over 100 percent to a 20-minute review with operations; nothing releases until the gap has a named owner and countermeasure. Monthly, inside S&OP, run the same math over a 3 to 12 month horizon to drive hiring, capital, and offload decisions while there is still 60 to 90 days of runway. Quarterly, re-validate demonstrated capacity and hours-per-unit against actuals; any center off by more than 5 percent gets its bill of resources corrected before the next cycle.

World-class rough cut planning looks boring, which is the point. The MPS releases at 92 to 98 percent bottleneck load every week, schedule attainment runs 95 percent or better, overtime stays under 5 percent of hours, and past-due backlog holds below 2 percent of monthly output. Capacity decisions happen a quarter ahead instead of a shift behind. Most plants can get there in 90 days, because the tooling is trivial; the discipline of refusing to release a schedule the math says will fail is the entire game.

Published 2026-07-02.