Ronald C. alexander, 13642 Sayre Street, Sylmar, Ca 91342 Hm: (818) 367-4676 or 0884, Wk: (714) 798-7703 email: ronnie_a@msn.com http://www.aiprojects.com/



How to Measure the Control of a Factory


For those of you with little background in the ancient art of measuring
factory performance, let me begin with some history.

Most factories are run by an individual known as a plant manager. Where do
these folks come from? Well, to start a factory an investment group needs to
buy some land and build on it. So you go out and find a farm, and talk the
family living there into selling - and becoming a major investor in the new
factory. So now you've got your plant manager - Papa Farmer! So, depending
on just how much the Farmers invested, you've got a few to many years in
which somebody (and maybe Somebody's kids) totally ignorant of how to
measure productivity, will be reporting productivity to you. Good luck!

At Hawthorne Farms when I arrived, we were taking all the defect rates from
all the product lines and adding them together to get an overall measure of
plant quality. So even if you designed production flow to use different
components at different times, etc., you'd often get three or four products
having the same defective component in the same hour - and the cry went up:
"stop the line!" The defect rate tripled or quadrupled! Wow! - of course,
there was nothing much wrong except maybe one bad batch of components, but
the whole plant came to a screaching halt - all 200+ products!

So how do you avoid this? You use a well established theory of product flow
that seems very unintuitive, except to a mathematician - or to someone
actually trained to do this kind of analysis... it's the same mathematics
used for reliability. For some reason Plant Managers let Reliability
Engineers do their job, and insist on doing the same job for the Quality
Engineers themselves - or ask that somebody they've promoted off the line do
the job. If you give a parallel flow analysis job to someone who you hired
straight out of High School, what results do you expect?

So what do you do with parallel product lines? You add the defects up just
like an electrical circuit with parallel capacitors - the capacity of a
product to have defects works just the same! Easy math, and it automatically
flags quality failures which impact the whole Plant to Plant Management. It
doesn't get used by the individual product lines, which directly use the
quality measures for their lines themselves - in fact, they post them and
establish action limits, and we call it statistical quality control (SQC.)

When we established SQC for production, production teams for each product,
and got Plant Management out of the "stop the factory!" mode for an isolated
defect, we saw an immediate and massive improvement in quality everywhere!
If the individual lines didn't catch that defective lot of components, then
the overall indicators would begin to reflect it. At that point somebody
aught to get hanged out to dry, for impacting the whole plant! Throughout it
all, Andy encouraged us by insisting that decisions were to be made at the
lowest level. It's the Plant's monetary dependance on that wrong indicator
that was a mistake.

We filled the warehouse up, since the sales folks wouldn't believe us - they
went so far as taking samples from the warehouse to "prove" we were making
inferior products. The warehouse samples showed an improvement over previous
field reliablity studies, as did retrospective studies of return merchandise
audits. We were still out of work, though; and a trained Plant Manager was
hired (Miller) who knew about this Total Quality Management approach. I had
been accused of inventing (TQM) out of whole cloth by the Plant Manager...
ain't an engineering education great? - especially with my previous BS in
mathematics.

When the factory closed, the ones who stayed were the line assemblers and
new college hires - and folks promoted from those jobs. Certainly not me!

home

© resume