HOW WOULD TOC TREAT THE EXAMPLE
WITH THE DELAYED PARTS?
Suppose first the machine we deal with is a non-constraint. In such
a case there should not be any specific
sequence the operator needs to adhere to. If the parts for
Product A are not available, let's start Product B. The next day,
the parts for Product A arrive. Well, we are not going
to switch over to Product A, because we should not follow
any strict sequence. However, Product B takes a whole
week. It may happen that the whole week of being busy
with Product B confines the shipping buffer (or CCR
buffer, depending on which buffer is ahead) too much,
leaving too little time for Product A to arrive on time.
Only when this happens does the order for Product A get
precedence, and we'll need to stop processing B and switch
to A. No harm is done, as wasting time on a non-constraint
has no negative effect.
What if the machine we deal with is the CCR? Then
we should have looked for the parts several days ago!
Such an occurrence should be very rare. Well, if we failed
to bring in the parts for the CCR on time—harm
to be caused. If the setup is not too long, I'd start
with Product B, then switch to A and then back to B.
Otherwise, replanning the CCR is inevitable—trying our
best not to disrupt the deliveries.
As missing parts to the constraint should be a rare
exception, whenever it happens we should inquire what
caused the unfortunate incident, why the problem
wasn't spotted on time (the buffer was fully exhausted—
and we reacted only then). It should not happen again.
There is a temptation to merge planning decisions with
execution decisions. Once you realize you have to deviate
from the planning, it seems the right thing to plan from
scratch. The negative impact on the organization is devastating.
Frequent replanning is the same as no planning.
And without the longer-term and global outlook of good
planners, the performance of the organization becomes
unpredictable. Note also that in such unstable performance
no feedback can be used for learning and improving
the planning-execution processes. Without any clear expectations
regarding performance, the meaning of deviation
from expectations is lost. We usually learn from gaps between
expectations, usually set by a "plan," and actual outcomes. When
expectations are vague, how do you trigger your learning?
What TOC suggests is to look at the planning side in
a new light. Let's acknowledge the uncertainty around, and plan with
enough buffers to allow flexibility in the
execution stage, while still being able to stick to what is
truly important. Once you set such a plan, the execution
decision rules center on following the plan, monitoring
the buffers and fixing the delays that really matter.