Page 230 - How To Implement Lean Manufacturing

208 Cha pte r T h i r tee n

Recall that the design had 42 seconds of transportation time per unit. We effectively
reduced that to 18 seconds (2 seconds per station × 9 stations) by the use of cells. That
directly translates into a faster cycle time by about 1.5 seconds [(42 – 18)/16]. This does
not sound like much, but on a 16-second work cycle, it is a 9 percent improvement in
production rate. Or viewed from a cost context, we just converted 9 percent more raw
materials into finished goods at no increase in operating expense. (See Chap. 11 for more
on this.)
Next, we eliminated the wait time of the workers. For the first 16 stations, there were
52 seconds of wait time—we turned this into productive time. That accounted for about
three equivalent seconds of cycle time (52/16 = 3.25).
Turning those wastes—the waste of transportation and the waste of waiting; actu-
ally, in this case it was all waiting—into productive time effectively reduced our cycle
time by over four seconds.
But, hey, not so quick! That doesn’t fully explain all the gains.
If the line had been actually producing at a true cycle time of 16 seconds originally,
even with all the waste of waiting, it would have been making 225 units/hr (3600/16 =
225) at 100 percent OEE. The real OEE was not 100 percent, rather the line had 4 percent
scrap and less than 1 percent availability losses, so OEE was really about 95 percent.
Consequently, we should have had about 214 units/hr (225 × 0.95 = 214). We did not
have 214 units/hr; rather, we had only 163 units/hr.
So how do we account for this missing 51 units/hr (214 – 163 = 51)? The answer has
to do with the fact that the line could not perform at the design cycle time of 16 seconds.
And why was that?
You got it! The answer is variation and dependent events! (See Chap. 18 for more infor-
mation on this.)
Yes, this effect is huge, and in this case it is easy to understand. Whether the process
is in lock-step when the process is fully synchronized with a conveyor, as this one was,
or if there is no inventory, the effect is the same. Any time one station performs at a time
above the cycle time, the effect is felt in all stations. This effect accounted for a huge loss
of production, 51 units/h, over 20 percent of the design rate of 225 units/h! Most
people find this interaction of variation and dependent events amazing! Well, amazing
it is, but it is also true, and it is also often an overlooked phenomenon.
Think about this concept of variation and dependent events for just a second. Since
all 21 process steps were synchronized by a conveyor in this case, any time one station
would perform at a time above the design cycle time, there was a time loss for the whole
line—for all 21 stations. It was exacerbated by the short cycle times, but the basic prob-
lem was that 21 people had to be totally synchronized to make this work. In this case,
each of the 21 work stations were dependent upon the other 20, otherwise no station
could maintain its cycle time. That is the nature of variation and dependent events and it
must be understood. Conversely, in the four-person cells there are now only three levels
of dependency, so even if one person slows down, they now only affect three others, not
20, and with 4 cells that one person only slows down
25 percent of the production, not all of it.
Point of Clarity There is In this example is explained one of the beauties
always a loss associated with of cellular design. They are a natural variation
the variation in the system… reduction device and they help us to execute the
flow improvement tactic of, managing the process
Always!!
to absorb deviations.