Page 45 - Practical Control Engineering a Guide for Engineers, Managers, and Practitioners

20 Chapter Two

2·2·1 What Is the Problem?
Although most engineers working in a manufacturing environment
are formally trained in problem solving, they almost uniformly
bypass the most important first step, which is to clearly and exhaus-
tively define the problem. The number of manufacturing-plant sec-
tion supervisors that I have irritated by persistently and perhaps
obnoxiously asking this question seems countless. They often do not
want to be bothered by such nonsense. After all, they know more
about the process than some staff engineer from headquarters and
have already figured out that there is a need for a control upgrade ...
now, just get busy and do it!
Early in my career I obediently plowed ahead and did the project
manager's bidding. There were some successes-at least enough to
keep me employed-but there were enough failures that I was basi-
cally forced to develop the so-called road map for process improve-
ment shown in Fig. 2-1 and discussed in great detail in the following
sections.
Before jumping into the approach championed in this chapter, it
is critical to convince the project sponsor/manager to develop a team
containing the control engineer as a member. This team should be
diverse, not necessarily in the politically correct ethnic manner, but in
the technical strengths of the members. There is little point in foster-
ing competition so only one member of each important discipline
should be present. Furthermore, the control engineer need not be the
leader; in fact, in my experience it is better to have someone with
more leadership skills than technical skills in that position.
2·2·2 The Diamond Road Map
Figure 2-1 shows a diagram containing four corners of a diamond but
really consisting of many steps.

Compartmentalization and Requirements Gathering
This is a fancy phrase that simply means, "divide and conquer." Man-
ufacturing processes are almost always complex and consist of many
parts, steps, and components. Breaking the process down into all of
its components or dynamic modules is the first step in getting a handle
on improving the performance.
Our method will attempt to decrease the variance of the process
variables local to each module with apparent disregard for the end-of-
line performance. Once each module becomes more controllable, the
targets for each module can be adjusted more precisely to affect the
end-of-line performance beneficially. One usually finds that even
without changing the targets of the improved modules, the decreased
local variance tends to have a salutary effect on the end-of-line prod-
uct characteristics. Therefore, there are three benefits to the localiza-
tion. The first benefit is better control, allowing the local set point to
be adjusted with the confidence that the module will actually operate