Modelling guide
Breakdowns and repairs
Machines fail at random and are repaired.
Built in A field or setting does it directly.
How
- On the machine:
mtbf_s(mean time between failures) andmttr_s(mean time to repair). - In settings:
reliability_mode: "per_machine". - The failure clock runs on working time;
breakdown_basis: "calendar"lets it run whether the machine works or not. - A failure stops the unit's work for the repair, then the work goes on with the time it had left. When the work must start again from the beginning (a weld seam, a cure, a test that must run unbroken), tick Work restarts after a repair (
failure_restarts_cycle: true): what was done since the start is done again. - Failures come at random whatever the machine's age. A machine that wears gets a shape above 1 (Hardware → Wear,
failure_shape: 3): it fails more often the longer it has run since its last repair, so failures come at more regular intervals with the same mean. Below 1 most failures come soon after a repair (early failures). A repair makes it as good as new.
Run Stochastic Analysis (Monte Carlo) to see the spread over seeds.
Check
A mill doing 10 min jobs, failing every 30 min of work on average and repaired in 2 min, over 200 h: 374 failures at random; with a wear shape of 4, 377 at much more regular intervals (the spread of the time between them falls from 0.91 to 0.26 of the mean). With restarts each failure costs about 5 min of work done again, and the mill makes 933 housings instead of 1,125.
The coating booth fails every 4 h of work on average and is repaired in 20 minutes: availability 4 / (4 + 1/3) = 92%. It needs 90 s / 0.92 = 98 s of a 120 s takt, so one booth still makes all 1200 brackets in 40 h.