Scheduling

Scheduling to True Capacity, Not Nameplate

Every late order traces back to scheduling against nameplate instead of demonstrated output. Here is the four-factor capacity math, the 95 percent attainment benchmark, and the cadence that lifts attainment 8 to 12 points.

Every late order traces back to the same sin: scheduling to nameplate capacity instead of demonstrated output. A line rated at 1,200 units a shift with 85 percent uptime and 96 percent first-pass yield actually delivers about 979 good units. Promise 1,200 and you are 18 percent oversold before the shift starts, which compounds into overtime at 1.5 times wages, expedited freight at 2 to 3 times, and a backlog that never clears. Plants that schedule to true capacity hit 95 percent plus on-time delivery; plants that schedule to hope typically sit at 75 to 85 and burn 5 to 8 percent of labor cost on unplanned overtime.

The math stacks four numbers: scheduled cycles times units per cycle times uptime times first-pass yield. Run 450 cycles a shift at 2 units per cycle, 88 percent uptime, and 97 percent yield: 450 x 2 x 0.88 x 0.97 equals 768 good units. The Production Schedule Capacity calculator does the multiplication; your job is honest inputs. Use demonstrated uptime from the last 8 to 12 weeks, not the best week ever. Use first-pass yield, not final yield after rework, because rework consumes the same capacity twice. And recount units per cycle after every tooling change; a cavity blocked off in March quietly cuts capacity 12.5 percent on an 8 cavity tool.

Benchmark the inputs, not just the output. Schedule attainment, good units produced versus scheduled, should run 95 percent plus in a healthy plant, measured daily by line; 85 to 95 means your capacity model is optimistic; under 85 means the schedule is a wish list. Load the constraint to 85 to 90 percent of demonstrated capacity, never 100, because a line loaded to 100 percent has zero recovery ability and queues explode nonlinearly past 90 percent utilization. First-pass yield inputs should reconcile with quality data within 1 point, and uptime inputs within 2 points of the OEE system, or someone is keeping two sets of books.

Four levers raise deliverable capacity without capital. Uptime: every point of uptime on the constraint is a point of plant output; a focused changeover program returning 30 minutes a shift on a 20 hour schedule adds 2.5 percent capacity. Yield: moving first-pass yield from 94 to 97 percent adds 3 percent good output and removes the rework load that was eating cycles. Cycle integrity: audit actual cycle time quarterly, because a 5 percent slow drift is invisible day to day and costs 5 percent of the year. Mix: scheduling long runs of low-yield products on the constraint wastes it; route them to non-constraint assets even at higher unit cost.

Failure modes to expect: sales overrides that push scheduled load to 105 percent of demonstrated capacity, which delivers 100 percent of the chaos and about 92 percent of the units; freeze the schedule inside 72 hours except for a named executive override. Averaging capacity across products with 30 percent cycle-time differences makes the model right in total and wrong every single day; build it by product family. Counting rework as output flatters attainment while eating real cycles. Ignoring ramp: a new product at week 4 runs 60 to 75 percent of mature rates, and scheduling it at mature rates guarantees a miss. Each shows up as attainment variance, so investigate every day below 90 percent.

The cadence: daily, compare good units to scheduled units by line, and every gap over 5 percent gets a coded reason in the tier meeting within 24 hours. Weekly: roll up attainment by line and by reason code, refresh demonstrated uptime and yield with the trailing 8 week average, and republish capacity numbers to the scheduler. Monthly: capacity model review with sales and operations, load versus demonstrated capacity for the next 12 weeks, flagging any week over 90 percent on the constraint. Quarterly: full input audit covering cycle times, cavitation, and crew levels. Plants running this loop typically lift schedule attainment 8 to 12 points in six months.

World-class scheduling looks like this: schedule attainment at 98 percent, promise dates quoted from demonstrated capacity in under an hour, a frozen 72 hour window violated fewer than twice a month, and a capacity model refreshed weekly that finance, sales, and the floor all quote identically. The constraint is loaded to 88 percent with the remaining 12 held for recovery and trials. Overtime runs under 3 percent of hours and is scheduled a week ahead rather than begged for at 2 pm. When demand exceeds capacity, the conversation happens in the monthly S&OP meeting with numbers, not on the shipping dock with apologies.

Published 2026-07-02.