Page 432 - The Toyota Way Fieldbook

Chapter 19. Lean Implementation Strategies and Tactics 405

distributions, box plots, and a four-way probability plot. Also, more
traditional lean tools were used, such as listing the process steps and
determining which could be done externally while the machine was
running and which had to be done internally while the machine was
down. These activities were prioritized, from those taking the most
time to those taking the least. A fishbone diagram of the materials,
man, methods, machine, measurement, and environment causal
factors effecting inefficient changeovers was developed. The top two
causal factors were identified as waiting for a changeover cart and the
process of heating the die, which accounted for 38 percent of the
changeover time, or 1.3 hours per changeover. They also discovered 12
of 22 other steps that could be done while the machine was running
(external).
The Black Belt in training generated a brainstorm of ideas for
improvement with some input from the floor. This was narrowed
down to action items to be implemented:

◆ Schedule mold changes to coincide with lunch breaks so the dies
could be heated during lunch (they could not justify the cost of
equipment to preheat the dies).
◆ Add one additional cart, which would be enough to optimize the
carts needed.

◆ Assign a dedicated changeover team instead of asking operators to
do it, so they could prepare a lot of the external changeover items
while the machines were running.
The results exceeded the goal. Detailed data were collected, put on
run charts and statistically analyzed. It showed significant improve-
ments. The result was a 98 percent improvement resulting in 2,828
parts per million defects (defining a defect as a changeover taking
more than 2.5 hours). The average changeover took 1.2 hours, well
below the 2.5 hour target. Analysis of the savings focused on the
reduced amount of labor for changeovers, which amounted to almost
$300,000 per year. Actually, the number of changeovers done in a
week was over the budgeted number, and they had a parallel program
to stabilize the schedule and reduce the number of changeovers. So
there were arguments about whether her project should get credit for
the labor savings based on the current number of changeovers or on
the anticipated reduced number of changeovers.
So this was a big success, right? Or was it? Let’s consider what’s
wrong with this picture:
1. The total process took several months. Much of that time was
spent on sophisticated statistical analysis and preparation of