Very interesting Dan, I can identify with your situation very well.
I had a long evaluation of master project sub/project stucture here
when I was trying to identify what other third party schedule tools could
offer over
Project. The list is long and conceptual. I've become pretty convinced that
what is a crucial part
of the equation is the concurrency and frequency of the actuals recorded
from the various machine
or operational cells. Not many believe me yet.
How to go about getting those actuals is the challenge? There are actuals
that reprsent new projects, new tasks
that become part of the flow load through your operations; and the
collection of progress on those tasks
as well a system of dispatch sheets that need to be directed at operational
cells to make sure they are working
on the correct task at each moment. A sophisticated ERP vendor would present
solutions to those within the
context of a work order system and then maybe a barcoding system. It becomes
a very big animal that effects
every level of the company. But many say it is a necessity for true
production management. I approached it
in the way that the schedule should be a separate tool for resource
management, that the actual resource scheduling
will always be a manual operation rather than one that was derived from a
convergence of what sales activities are
or a rules based schedule or optimization.
I can imagine in a custom kitchen environment you have large saws that
operate in conjunction with a nesting program
to maximize cuts of raw materials, some router operations etc. Each of these
machines and operators must be kept busy
at the scheduled task and must be very flexible. In order for the sub
schedules to be responsive to the demands put on
resources there is a large amount of rescheduling that must be done. This
carries alot of overhead as traditionally each
production plan in project is fully constrained.
You mention bottlenecks and that is a key concept and one which I propose is
best addressed with a hybrid
critical path / critical chain approach. Once one believes that the
effective management of the bottlenecks is the
best goal, then you can actually fully constrain the production plan
subprojects and simply optimise the constraining
resources. The tasks that are not to be accomplished by the bottleneck
resources have an inhereint flexibility to them
as they are not critical constraining resources. Manage the big saws or
staining booths for example and all the other
tasks will be easy to schedule and will not limit delivery in a bad way, for
example you can always hire more folks to
install drawer slides if it becomes an issue that we can't meet deliveries
because we have too many drawer slides to install.
If you are interested in really attacking your bottlenecks the best
reference would
probably be
http://www.focusedperformance.com/. I'm trying a sneak approach
at my company by
showing the benefits of 100% flow optimization at the critical constraining
resources(CCR). Using Project
resource levelling can provide that effectively reducing the CCR buffer to
zero, That 100% is impossible without
extrememly frequent updates of actuals? Other wise I think it would be vary
hard to present some of the concepts
described on that site as fitting in with the way things are done now and in
the past. Maybe kitchens are different than building
custom machines but I know our factory could not operate if the scheduling
was as ridgid as a widget production line, there
is too much hands on craftsmanship required at many of the tasks, even if we
could predict with huge historical accuracy
how long each task would take. Fixing a fully constrained production plan to
those estimates would never allow us to
exceed them without a very responsive system of collecting actuals and
reponding to them.
