Page 248 - How To Implement Lean Manufacturing

226 Cha pte r F i f tee n

Establishing Jidoka
We made one other agreement with management, which we knew would be critical.
The agreement was that if we had a quality problem, we would stop the line and not
start it up until the problem was fixed. We knew it would be crucial to have this type of
jidoka concept in place. At first, they baulked at this proposal. We did not even try to
explain logically how important this was, but when we reminded them that they had
characterized the line as having few quality problems, they reluctantly agreed.

Establishing Pull Demand Systems
Our kanban system provided a good pull system within each cell, but we had no pull
signal from the storehouse. We had enough information to design a good heijunka board
with a make-to-stock system. We inquired about implementing a heijunka board with a
production kanban system but they were opposed to taking the time to do it. They said
“Maybe later,” which usually means, “No.” It was clear what they wanted: to increase
the production rate and reduce the labor it took to produce a unit. We obliged.
The Production Run, with Problems Galore

Our Jidoka Concept Works Like a Charm
With these preliminary steps in place, we set up for the first production run. It took less
than one hour for us to see one of the major issues. Station 6 of Cell 1 was now only used
for final visual inspection and packing. The very first unit it received had an incorrect
O-ring installed. Following our jidoka guidelines, we shut down the line, gathered the
cell workers, and began to investigate. It was simple: The changeover had missed that
this model needed both a different shaft and a different O-ring. The shafts had been
changed but the O-rings had not. Fortunately, due to the kanban system, we had less
than 25 units to rework. We reworked the work-in-progress, modified the error in the
changeover procedure, acquired the correct O-rings, and started again. On this run, we
had to stop four more times for quality problems in the first four hours of production
alone. It was clear at this point that process stability was nonexistent. They needed a
good dose of Lean foundational issues—that is, quality control (see Chap. 20).
Back to the test run and the finding of defective products. In each case, with small
lot production, the problem was easily corrected and we only had a few units to rework.
If we had maintained the lots of 50 as before, when the incorrect O-ring was found we
would have had 50 units per work station to rework. That would have been over 400
units on the old nine-station arrangement for each of the five quality problems we
encountered. Combined, that would have been 2000 units reworked with the old cell
design in the first four hours alone—recall that daily production was only 2000 units! It
was becoming obvious to everyone how important the small lot and jidoka concepts
were. And it was equally clear that we had found at least one major source of the extra
time it took to make an order. The rework, in the old arrangement, could easily have
taken as much time and labor as the design workload.
Suspicious that quality might be a larger problem than previously portrayed, we
did a little questioning and found that almost every run had one or two, or sometimes
more “start-up burps,” as they called them, and rework like this was common. In fact,
that was one of the reasons they had everyone so cross-trained: so they could do the
rework. There were several, huge, absolutely unmistakable messages in what we had
just uncovered.