Designing for demand that may turn out higher or lower
The demand is a forecast.
Built in A field or setting does it directly.
When
The factory should still work if it turns out 20% lower or 50% higher, and you want to know what that costs in equipment before you buy it.
How
Build the factory for the planned demand as usual. Then, in the Improve goal popover, write Demand scenarios: each outcome as a percentage of the planned demand and a weight for how likely it is, e.g. 80%: 1, 100%: 2, 150%: 1. The planned demand (100%) is always one of them, with weight 1 unless you give another. In the profile: "solverSettings": { "improveScenarios": [{ "factor": 1, "weight": 2 }, { "factor": 1.5, "weight": 1 }] }.
Improve then runs every design once per scenario, with the demand scaled: the takt, every product's arrivals and the quantities on an order list (due times, demand patterns and the mix stay as they are). A design meets the goal only if it meets every limit in every scenario: the takt target and a throughput limit scale with the scenario's demand, a lead-time, WIP or on-time limit holds as written. Output, lead time, WIP and orders on time are traded as the weighted mean over the scenarios; equipment, area, cost and operators are the design's own. Each missed limit names its scenario ("150% demand: Product demand not verified"), the design detail lists every scenario's output and lead time, and the search adds capacity where the highest demand it misses runs short.
On a line paced by the takt the run makes what the line can (units are released as there is room), so a scenario mostly moves the target the line is judged against. In a factory fed by arrivals or orders the runs themselves differ: queues grow at the higher demand.
Only the planned demand gets the stochastic confirmation a factory judged on product demand asks for; each other scenario is one run with the same seed. With Short runs over this run's periods, the screening runs use the planned demand and the designs run at full length afterwards are judged in every scenario.
Check
A shaft arrives every 400 s and the lathe turns one in 300 s. Build layout buys one lathe (75% busy) and the mean lead time is 5 min. With demand scenarios 100%: 2, 150%: 1, one lathe falls behind at 150% (a shaft every 267 s; the lead time grows past 80 min and demand is not met), so Improve recommends a second lathe, which meets the demand in both (test/demand-scenarios.test.mjs).