Lesson 4. The experiment: what to speed up
The bottleneck is found, the release is subordinated to it. One question is left, the one every plant manager asks: what should we speed up so that the factory makes more? In this lesson there are three short experiments, no new parts.
4.1. How to run them
- switch SLOW off — we look only at the machines;
- a machine's time — Alt/⌘ + click on it, in turn: 0.5 → 1 → 2 → 4 seconds;
- each shift: power off and on → R → START, after 200 seconds (3 minutes 20 seconds) look at the right display «made».
4.2. Experiment 1: as it is
Touch nothing: Herbie — 2 seconds, the others — 1. By second 200, about 94 are made.
4.3. Experiment 2: speeding up the wrong machine
Machine 1 → 0.5 seconds. Made — again about 94. Machine 1 became twice as fast, and the factory did not make a single extra part: its parts just join the queue in front of Herbie sooner. The idle meter on machine 1 will show that it now stands even more.
Set machine 1 back to 1 second.
4.4. Experiment 3: speeding up Herbie
Machine 3 → 1 second (three clicks: 2 → 4 → 0.5 → 1). Made by second 200 — 128, the whole shift: a third more than in experiment 1.
4.5. Why
The output of the factory is the output of its bottleneck. An hour saved on machine 1 changes nothing: machine 1 was waiting anyway. An hour saved on Herbie is an hour for the whole factory. On the hike Alex lightened Herbie's backpack, and the whole troop walked faster.
After experiment 3 all the machines are equal again — as in FL-01. Make any machine slow and find it with the instrument, without looking: the new bottleneck will again have zero idle time.
4.6. Check
| Experiment | Made by 200 seconds |
|---|---|
| 1. Herbie 2 s | about 94 |
| 2. machine 1 — 0.5 s | about 94 |
| 3. Herbie 1 s | 128 |
Stuck? Something doesn't work →