Open the assembly plant (no account needed to open it; running simulations takes Pro). To ship every order within the hour, should the plant speed up the test, add AMRs (autonomous mobile robots) or machine faster? With the same orders, your AI changes them and compares them in a simulation. This article runs on a synthetic plant with the same tools your AI calls over MCP. Every number below comes from that run.
1. The plant and its first hour
The synthetic assembly plant’s first hour of a shift.
- Six machined housings wait in the WIP buffer, and eleven orders are released 110 s apart. For each order an AMR brings a tote of blanks to a free CNC of its part family (an operator sets it up and unloads it), and family A goes on to a free welding robot.
- AMRs take each part to the eight-station line, the unit runs down the stations (each passes it on only when the next is free), and a tester takes it to a free test rig (one in ten fails and is reworked, without a re-test). A packer packs it, and a forklift wraps it and loads it at one of four docks.
- AMRs also run a kit cart to the line-side supermarket every three minutes, and inbound pallets share the forklifts. Each CNC breaks down about every four hours (3 to 8 minutes to repair).
- Today: a 3-minute test, 4 AMRs in service (of 6), and a 5-minute CNC cycle. Not counted: vehicles avoiding each other, operator breaks, conveyor buffers. The figures are assumed.
2. The test time and the AMRs
Your AI ran the saved scenario’s parameters, the test time and the AMRs in service, at every combination with sweep_process_parameters, twenty runs each on the same random numbers. Orders shipped within the hour (of 11, mean of the 20 runs):
| Test time | 2 AMRs | 4 (today) | 6 |
|---|---|---|---|
| 4 min | 6.7 | 8.3 | 8.3 |
| 3 min (today) | 7.4 | 10.8 | 10.8 |
| 2 min | 8.0 | 11.0 | 11.0 |
- With today’s 3-minute test and 4 AMRs, 10.8 ship on average, and all 11 in 16 of the 20 runs (the last then leaving at 57.3 minutes).
- With a 2-minute test, all 11 ship in every one of the 20 runs, the last at 55.0 minutes.
- 6 AMRs change nothing over 4. With 2 the plant falls short: orders wait 324.1 s for an AMR on average (34.8 s with 4).
- With a 4-minute test, no number of AMRs ships more than 8.3.
3. The machining time
With the CNC cycle lengthened from 5 minutes, orders shipped within the hour (of 11):
| CNC cycle | 3-minute test | 2-minute test |
|---|---|---|
| 5 min (today) | 10.8 | 11.0 |
| 6 min | 10.8 | 11.0 |
| 7 min | 10.8 | 11.0 |
- Even a 7-minute cycle hardly changes the orders shipped: the test, not the machining or the AMRs, sets this hour.
4. How to use these numbers
- Give your AI your order release plan and product mix, measured cycle, setup, station, test and packing times, the test failure rate, and your fleets and travel speeds.
- Once the bottleneck is known, trial that step first on site. Finishing earlier saves money only if overtime, shifts or equipment spending actually fall.
- Not counted: vehicles avoiding each other, operator breaks, conveyor buffers, maintenance.