Page 309 - How To Implement Lean Manufacturing

286 Cha pte r Ei g h tee n

but this is not a production rate problem, the cell will still produce to takt. It is just that the
operator at station 4 will sit around a while. On the other hand, when station 4 takes
70 seconds to produce its work, that subassembly is held up and station 5 is starved for
work for ten seconds. This delay passes through all the work stations of the cell in a
wave and that piece is produced on a 70-second cycle time.
So let’s recap… If the station that varies—in this case, station 4—operates faster than takt,
station 4 must wait for the subsequent station to pull the production; however, when station
4 just happened to operate slower than takt, station 4 would slow down the whole cell on that
cycle and there is no recovery with a resultant loss of production rate.
So even though the station may have a 60-second cycle time on average, any time the
cycle time is above average, the production rate drops. This concept is known as the
effect of variation and dependent events. (The dependency is that the “next step”
depends on the “prior step” for supply.)
So the solution is, guess what …? You got it! Add some inventory. We will need to add
inventory both before and after station 4, the one with the variation. We need the inventory
in front of station 4 so when it produces faster than takt, say at 50 seconds, there is raw
material available to keep it producing. We also need the inventory after station 4, so when
it is operating slower than takt, say 70 seconds, there is raw material to supply station 5.
Then, station 4 can have the variation and maintain production at takt on average.
This effect of variation and dependent events is not well understood and is a problem,
always. Two solutions can be employed: Either totally remove the variation, or totally remove
the dependency. To totally remove the variation is an impossibility. Recall that the definition
of variation is, “the inevitable differences in the outputs of a system.” Since it is inevitable,
total removal is an impossibility. Okay, so let’s totally remove the dependency. This means
tons of inventory, the exact thing we are trying to eliminate in a lean solution.
So, guess what? The solution is to find the happy medium and it is best done by first
reducing the variation to a minimum so inventory can then be reduced accordingly.
That’s enough background for now; do the experiment and see firsthand—under
controlled conditions—exactly how this phenomenon plays out.

The Experiment
To do the factory simulation experiment, get 12 ordinary dice, some students; four
teams would be ideal. If you only have enough for two or three teams, do that, the
experiment it totally flexible. We will use dice to get random numbers, and we will vary
the number of dice for each team to modify the amount of variation for each team. Dis-
tribute the dice as shown in Table 18-1. A little math will show that the average values,
independent of the number of dice used will approximate 21. For example, the average
for any die is 3.5, and in each case the multiplier times the number of dice is six, so each
team will average 21. Hence, the rate of the production process simulations for all four

Team No. No. of People No. of Dice Multiplier
1 1 1 6
2 2 2 3
3 2 3 2
4 3 6 1
TABLE 18-1 Team Distribution of Dice