chiaha-SLX Lab 4: A machine that breaks down
Time: 40–50 minutes
You need: a browser and a chiaha-SLX sign-in.
Open it: https://slx.chiaha.com/slx?open=deck/breakdowns
Learning goals
- Model a server that fails and gets repaired, separately from the work it does.
- Compute availability from mean time between failures and mean repair time, and check it.
- See that two machines with the same availability can make jobs wait very differently.
The problem
Jobs come to one machine every 10 ± 3 minutes and take 8 ± 2 minutes of work (both spread evenly). Separately, a failure clock runs: on average every 60 minutes (exponential) the machine goes down, and a repair takes 4 ± 1 minutes. While it's down no new job can start; a job already on the machine is allowed to finish. The failure clock keeps running whether the machine is busy or not. The model runs for 2,400 minutes (five 8-hour days).
Find these lines in the model:
| Line in the model | What it means |
|---|---|
run_length = 2400 |
the run length, in minutes |
iat = rv_uniform(Arrivals, 7.0, 13.0) |
a job every 7 to 13 minutes |
rv_uniform(Service, 6.0, 10.0) |
6 to 10 minutes of work |
rv_expo(Failures, 60.0) |
time to the next failure, mean 60 minutes |
rv_uniform(Repairs, 3.0, 5.0) |
a repair takes 3 to 5 minutes |
served, failures, waited, busy, down |
jobs finished, breakdowns, total minutes jobs waited, minutes of work, minutes under repair |
Measures
Tick Show final values and set Runs to 5. For each run:
- Availability =
1 − down ÷ run_length - Busy =
busy ÷ run_length - Mean wait =
waited ÷ served(minutes) - Breakdowns =
failures
Average each over the five runs.
Before you run
Q1. With failures every 60 minutes on average and repairs of 4, what share of the time should the machine be available? How many breakdowns should happen in 2,400 minutes? What share of the time should it be working on jobs?
Tasks
Keep Seed at 1 and Runs at 5. Each task starts from the original model.
- As written. Run and record the four measures.
- Slower repairs. Change
rv_uniform(Repairs, 3.0, 5.0)torv_uniform(Repairs, 6.0, 10.0)(8 ± 2 minutes). Work out the new availability first, then run. - Often and short. Change
rv_expo(Failures, 60.0)torv_expo(Failures, 30.0)and the repair line torv_uniform(Repairs, 1.5, 2.5). Twice as many breakdowns, half as long. Run. - Rare and long. Change the failure line to
rv_expo(Failures, 120.0)and the repair line torv_uniform(Repairs, 6.0, 10.0). Half as many breakdowns, twice as long. Run. - A long run. Back to the original, change
run_length = 2400torun_length = 240000. Run and compare with Q1. - If you have runs left: faster repairs,
rv_uniform(Repairs, 1.5, 2.5)with failures at 60.
Questions
Q2. Compare task 5 with your answers to Q1. Then compare the five-day runs in task 1: how far off is the breakdown count, and why?
Q3. Tasks 1, 3 and 4 have the same availability on paper. Is it the same in your runs? Is the mean wait the same? Put the three in order of mean wait and explain the order.
Q4. A supplier offers a repair kit that halves repair time but makes failures twice as frequent. Using your results, should the plant take it? What would you want to know before deciding?
Q5. served is the same in every task. Why don't breakdowns reduce how
many jobs get finished here, and when would they?
The answer key is shown to the course's instructor when signed in.