Modelling guide

Demand peaks, campaigns, seasons and a mix that changes by shift

Demand is not even over the day, the week or the season: a busy shift, a campaign, a peak month, or a mix that is mostly one product in the morning and another in the evening.

Built in A field or setting does it directly.

How

Give the product's arrivals a demand pattern (Factory data → Products → Demand pattern): from an hour of the cycle, a factor, e.g. 0 h: 1, 8 h: 3, 16 h: 1. The factors are shares: the product's rate (Per week, Every) stays its average, and the pattern says how it is spread, here 20%, 60% and 20% of each day's demand over three shifts. In the profile: "arrival_process": { "mean_interarrival_s": 864, "pattern": { "repeat_h": 24, "periods": [{ "from_h": 8, "factor": 3 }, { "from_h": 16, "factor": 1 }] } }. Before the first hour the factor is 1; a factor of 0 is no demand. The cycle repeats every repeat_h hours: 24 for a day, 168 for a week, the run's length for one campaign or season. Random arrivals keep their randomness, at the rate the pattern gives.

A mix that changes by shift is two or more products, each with a pattern of its own (cabinets 1.5 on the day shift and 0.5 in the evening, drawers the other way round, each with half the total rate).

Auto Generate sizes for the mean demand by default: a peak's queue clears after it. Project settings → Size for demand → Its peak buys enough machines for the busiest period, so lead times stay short through it. For a line it takes the products' combined peak, not each product's own (a day product and an evening product never peak together); for work centres each product's own peak, which buys a little more when products peak at different times.

When the mix moves the load from one machine to another, Run details shows Bottleneck over time: the busiest station in each twelfth of the run and when it changed (summary.bottleneckOverTime).

A known large order (a quantity due by a date) is better written as an order list (settings.job_shop_finite_jobs, one row per order with release_s and due_s), which reports orders on time and late.

A line paced by the takt alone (no arrivals) keeps its mix through the run.

Check

Valve bodies are demanded 20 in the first shift, 60 in the second and 20 in the third, every day for three days. The second shift needs 60 x 10 min = 10 h of turning in 8 h. Sized for the peak the shop gets two lathes and the mean lead time is 15 min; sized for the mean (17 h of turning a day) it gets one, and the second shift's queue makes it 62 min. Both make 299 of the 300 bodies.

Capabilities it covers

All capabilities and guides

Try it on your own product.