Lesson 4 · Sizing the fleet and comparing runs
How many vehicles does the plant need? The honest answer comes from running the alternatives and comparing them. You will raise the demand to 80 orders an hour, run one, two and three AGVs, and learn which numbers to read to tell a fleet that copes from one that merely survives.
- Time
- 25 min
- Start from
- the checkpoint of lesson 3
- You will learn
- run alternatives one after the other and compare them in the Run comparison table
- tell a saturated fleet from a comfortable one from queue, lead time and utilisation
- find where the time goes with the results sections, charts and heatmaps
Open your project from lesson 3, or its checkpoint.
A third charger
You will compare fleets of up to three vehicles, and every vehicle needs a station of its own to start from. Go back to Draw and add one more charger.
- Place it. Choose Station (7) and click on the bottom side near (500, −6 000), just left of Goods in. Name it
Charge 3and tick Charger. - Keep a gap. Starting stations must be at least the scenario's minimum distance apart (1000 mm by default), so three vehicles do not start on top of each other. 2.5 m between chargers is plenty.
Raise the demand
With an order every 60 s, two vehicles just keep up. Now ask for more. In Simulate, open Orders and set Interval to 45 seconds: 80 orders an hour. Leave everything else as it is, and set Run duration to 1 h.
The principle that makes comparisons trustworthy: change one thing at a time and keep the duration the same. Here the only thing that changes between runs is the number of vehicles.
Three runs
- One vehicle. In Fleet, set AGV count to
1and press Start run. Wait for it to finish; it takes about half a minute. - Two vehicles. Set the count to
2and run again. The previous run was filed automatically; you do nothing to keep it. - Three vehicles. Set it to
3(the step reads3/3 AGVnow that there are three chargers) and run once more.
Read the comparison
Every finished run is added to the Run comparison table in the dock under the map (click its tab to open it). Each row is a run; the columns on the left are the inputs that differ, the rest are results. A ▲ marks the best value of each measure, and only where the runs really differ.
| Run | AGV | Throughput | Max queue | Lead time | Utilisation |
|---|---|---|---|---|---|
| Run 1 | 1 | 30 /h | ≈ 50 | ≈ 20 min | 100 % |
| Run 2 | 2 | 59 /h | ≈ 20 | ≈ 9 min | 99 % |
| Run 3 | 3 | 79 /h | 0 | ≈ 1 min | 90 % |
Read it like this:
- One vehicle delivers about 30 an hour against 80 asked for. The queue keeps growing for the whole hour (the maximum is around 50) and a mission waits for tens of minutes. The fleet is far too small.
- Two vehicles deliver about 59 an hour. Better, but still below 80: the queue never stops growing (around 20 by the end) and lead time is minutes. Utilisation of 99% is the sign: the vehicles are saturated and the surplus goes to the queue.
- Three vehicles deliver about 79 an hour, the whole demand. The queue stays at zero and a mission takes about a minute from creation to delivery. Utilisation is about 90%: no longer saturated, though still with little spare capacity.
The rule of thumb: the right fleet is the smallest one whose queue stays bounded and whose P95 lead time meets your target, with some headroom for the bad day. Here that is three. Throughput that equals the demand is not enough on its own: look at the queue.
The table keeps up to 12 runs of the session. A bin button on each row removes it, Export CSV exports the table and Clear empties it. The runs themselves stay in the Run archive.
Where does the time go?
The comparison says which setup is better. The Results tab says why. Open it after a run and expand its sections:
- Mission times: the average lead time splits into queue wait (before a vehicle is assigned) and execution (assignment to delivery). Almost all of the one-vehicle lead time is queue wait.
- Fleet usage: one bar per vehicle, split by what it was doing: carrying a load, driving empty, loading and unloading, returning, waiting for traffic, charging, idle. It shows whether the work is shared evenly.
- Timeline: charts of throughput and queue over time. A queue line that only rises is saturation made visible; a line that wobbles around a low value is a fleet that copes.
- Bottlenecks: waiting caused by traffic and queues at stations. It is empty here, because a single loop has nowhere to jam; the next lesson gives it something to find.
To see the same information on the plan, choose a layer in the Layout heatmap list of the toolbar: Occupancy shades each path by how much of the time it is occupied, and the others show passes, waits at entries and waiting hotspots. The heatmap changes only the view; it never restarts the run.
For a report, Export (.zip) writes a ZIP with the results as JSON and CSV tables, and PDF writes a short report with findings and plans. Both work on the current run and on any run reopened from the archive.
- Three rows in the comparison table with 1, 2 and 3 AGVs, throughput of about 30, 59 and 79 an hour and the ▲ marks in the third row.
- Set the vehicle count back to 2 and raise Straight speed on the fleet's card in Vehicles from 80 to 120 cm/s. Does a faster pair catch up with the demand? Compare the new run with the previous two.
- Change AGV selection for the mission to the second option with three vehicles and compare it with the default. With one pickup the difference is small: dispatch rules matter more when there are many pickups.
- Reduce the loading and unloading times from 12 s to 6 s. Which lever, a faster vehicle or shorter service, buys more throughput?
Checkpoint
It contains: 8 paths · 6 stations · 3 AGVs.
Download the checkpoint (aurora-04-fleet-sizing.en.sim)
Open it with Project›Import project. It replaces the project you have open, so save yours first (or open it in a new browser tab). Anything you do from here on is a copy: the checkpoint file itself never changes. Use it to continue from here if you skipped a lesson, or to compare your own project with the expected one.