Lesson 4. The experiment: dice
The factory is built and can measure itself. Now — the experiment the whole thing was for. Two shifts of 128 parts: in the first the machines run like clockwork, in the second a die decides the time of each part. On average the machines of the second shift are just as fast as in the first. Let us see whether the factory notices.
4.1. Idea: a die at the machine
Click the die on a machine — it switches round: grey (off) → blue (½…1½ seconds) → orange (a 1…6 die). Orange is as in the book «The Goal»: before each part the machine «rolls the die», and the part takes
- 1 point — 1/3.5 of a second, the fastest;
- 6 points — 6/3.5 ≈ 1.7 seconds, the slowest;
- on average 3.5 points — exactly 1 second, as without a die.
The release stays the same — a part a second. On paper the line is still balanced.
4.2. The experiment
For each shift write down three numbers at the moment the left display stopped at 128:
- made — the right display;
- work in progress — 128 minus made;
- last — how many seconds after START the right display reached 128.
Shift 1 — no dice. The dice are grey. Power off and on → R → START.
Shift 2 — a 1…6 die. Click the die of each of the five machines until it turns orange (two clicks). Power off and on → R → START.
| Shift | Made by end of release | In progress | Last one done |
|---|---|---|---|
| no dice | about 115 | about 13 | about 140 s |
| 1…6 die | about 106–111 | about 20 | about 155–160 s |
Your numbers will be a little different — a die is a die — but the second shift is always worse than the first.
4.3. What to watch during the second shift
- the lamps of the last machines go dark — a machine has nothing to do although the line is full of parts;
- the first machines' buffer bar at the bottom is full — there is a queue in front of them;
- the difference between the displays creeps up and by the end of the release reaches 20 and more.
4.4. Why
The machine that rolled a six fell behind — and the one after it stood idle all that time. Then that machine rolled a one; it would gladly catch up, but nothing came from the front: the head start was lost. And the delay did not go away — every machine further down the chain got it. That is why a balanced line makes less than the average speed of its machines, and a pile of parts grows inside it.
How Eliyahu Goldratt told about it and what follows — in the section «What happens and why».
4.5. Check
| What you look at | What you see |
|---|---|
| shift 1, right display at 128 | about 115 by the end of the release |
| shift 2, right display at 128 | fewer than in shift 1 |
| shift 2, difference | larger than in shift 1 |
| both shifts, the end | source 128, sink 128, both displays 128 |
¿No te sale? Algo no funciona →