On Wed, Jan 15, 2014 at 3:15 PM, Christian Carlow <
[email protected]> wrote:

>
> One important part I'll have to work out is how to handle the inspection
> department mentioned in the previous post.  I think inspections can be
> handled as separate production runs just like normal ticket movements.  The
> main difference between the inspection and manufacturing dept tickets is
> that inspection dept has the ability to split a manufacturing ticket into
> different quantities destined for different locations.  I think this can be
> handled with the WorkEffortAssoc entity.  Inspection users currently select
> ticketIds from a list to be inspected.  I guess the ticketIds in this case
> would be served by workEffortId of the previous production run.
>

This would have to be flexible, as different manufacturers will have
different protocols.  My father worked all his life in quality control,
just down the hall from the engineer's labs responsible for designing new
products.  For any given product, there were sometimes multiple protocols,
all of which would be used for a given assembly line.  Each would be used
on a random sample of the products coming off the line, and in some of
their facilities, there'd be an extra protocol applied to every unit coming
off the line.  One protocol involved something like function or integration
tests.  Depending on the facility, and how paranoid about quality the
facility's manger was, this would either be applied to a large random
sample, or to all, of the units coming off the line.  A second, including
tests of each component in the unit in addition to the tests in the first
protocol, would be applied to a much smaller, but independent, random
sample.  Please note, a sample generally had a minimum size of 100 units.
In the most recent configuration of that company's QA infrastructure, the
assembly lines was not started for a batch of less than 100,000 units, and
they actually had a computer on the end of the assembly line which
performed all the required tests.  That is to say, every unit coming off
those assembly lines was subjected to at least one suite of tests.  And
there were other protocols also, but I was too young at the time to
remember much else.  But I do know my father would say that the QA testing
was not a separate process from the manufacturing process, but an integral
part of it.  On the other hand, my brother worked in a company that had
it's own machine shop, and that shop would have to gear up to make a single
unit (which of course would be tested thoroughly), because it was needed to
permit another department in the same company repair an insanely expensive
piece of equipment, the parts of which had not been made in decades.  These
techs had the product specs on file, and so, using modern equipment, made
'obsolete' parts to spec, so that the supported old pieces of equipment
could have a few extra years of life; and they had to do it to perfection
since a failure could cost lives.  I would expect that there would be
seemingly countless variants between these two extremes.

I suppose, the question becomes how best to model QA in such a way that it
supports the many varied QA protocols that may be encountered.  Is the
production run adequate, or can you derived from it something more flexible?

I have not studied the code you're working with, or the books to which you
referred, but from the perspective of an end user, in the place of either
my father or brother, I would look for something that was flexible enough
to support either of the cases I described above.

Let me ask you, do you think what you have in mind is flexible enough to
handle such edge cases, and if so, how?

Cheers

Ted

Reply via email to