A Guide to Implementing the Theory of
Constraints (TOC) |
|||||
The Process Of Change We have seen over the last few sections that we seem
to be caught between a rock and a hard place.
We do have the intuition to know that local optimization doesn’t work
well for the overall system good. And although
this looks as though it is the result of both our own psychology and the fact
that large scale serial process operations are very recent human inventions;
it is also apparent that we do know how to develop a system-wide perspective
even within these systems. We know
this because the people at the top (and some others too) can see the overall
system – and we know that these people are no different from you or me. We also know that, currently, these people
are being served by a “legacy” system; a system of local cost accounting and
local efficiency. What we need is a process by which to facilitate the
change. A process of get us from where
we are now to where we want to be in the near future. Senge described the “where we are now” as
the current reality, and the “where we want to be in the future” as the
vision (1). He noted that if there was
no gap between the current reality and the vision, then there would be no
need to move toward the vision. The
gap between the two becomes a source of creative energy which he termed
“creative tension.” Well there is no shortage of creative tension I am
sure. But that doesn’t actually help
to move forward. In fact Senge notes
that “creative tension often leads to feelings or emotions associated with
anxiety, such as sadness, discouragement, hopelessness, or worry.” Senge described this as “emotional
tension.” The key point is not to
confuse creative tension with emotional tension, otherwise we predispose
ourselves to lowering our vision (1).
We need a process of change to ensure that we move from where we are
now to where we want to be in the near future. Goldratt briefly outlined a process of change in
1990 (2). He characterized it as
follows; (1) What to change. (2) What to change to. (3) How to cause the
change. The first two questions Goldratt considered to be
technical and the last one to be psychological. Why is it that how to cause the change is
considered to be psychological? Well
it almost seems like a mantra, it goes like this (2); § Any improvement is a change. § Any change is a perceived threat
to security. § Any threat on security gives rise
to emotional resistance. § Emotional resistance can only be
overcome by a stronger emotion. Note that not all change is necessarily an
improvement; this might help explain the reason for our healthy skepticism
towards the changes that actually do lead to improvements. The critical supposition is that the emotional
resistance that improvements induce can not be overcome by logic; it can only
be overcome by the stronger emotion, the emotion of allowing the people
involved to deduce the solution for themselves. But let’s not go there just yet, that is
really the subject of the next page – the agreement to change. Let’s confine ourselves here to the first
two steps; (a) what to change and (b) what to change to. In a way we have already intuitively done all of the
first step and part of the second step.
We did this when we formulated our rules of engagement. Let’s repeat the rules here; (1) Define the system. (2) Define the goal
of the system. (3) Define the necessary
conditions. (4) Define the fundamental
measurements. (5) Define the role of the constraint(s). By working through these steps we have partitioned
our problem into what exists now – our current reality, and what doesn’t
exist now but we envision for the near future – our future reality. What we are missing then from our
discussion so far is a method that allows us to address the 5th step;
defining the role of the constraints.
We still need to develop a method to focus on the constraints of the
system. So let’s do that right
now. Why do we want to focus on the
constraints? Because these are the
most fundamental leverage points to move from our current reality to our
future reality. The constraints are
the most fundamental points to enable our process of change. To extend the military metaphor slightly,
to go with our rules of engagement we need a plan of attack.
Goldratt furnished a focusing process in the
earliest versions of The Goal,
however, it was implicit. In later
editions it was made explicit as the five focusing steps. In The Race
published in 1986, we can see the one of the logistical applications –
drum-buffer-rope – essentially fully developed, and another – project
management – at an incipient stage; and yet there is no mention of the five
focusing steps. In addition to the
explicit inclusion in later editions of The Goal, the
five focusing steps were also presented in two other sources dating from 1990
(2, 3). The five focusing steps, exactly as in the original
verbalization, are as follows (2); (1) Identify the system’s constraints. (2) Decide how to Exploit the system’s constraints. (3) Subordinate everything else
to the above decision. (4) Elevate the system’s constraints. (5) If in the previous steps a
constraint has been broken, Go back to step 1, but do not allow
inertia to cause a system constraint. With the verbalization of this process we now have
an iterative means of addressing the fundamental points that will allow us to
move from where we are now, to where we want to be in the future. Even though our rules of engagement may
remain essentially static – the ultimate anchor point in our reality; the
five focusing steps allow us to “loop” and adapt as the constraints in our
environment change in response to our own actions, and in response to the
actions of others that are imposed from outside of the system. The five focusing steps
are without doubt the singular most important
aspect of the Theory of Constraints. A common analogy for the focusing process is the
strength of a chain. Everyone knows
that a chain is only is strong as its weakest link. The weakest link is the limitation upon the
strength of the whole chain (and while a number of people have argued
passionately to me that there is more than one weakest link in a chain, I
remain unconvinced). The chain will
always break at the weakest link, and if we were to rejoin the chain, it will
next break at the next weakest link – and so on. Notice, however, that the 5 focusing steps are
worded for a system and not for a single chain as used in the
analogy. This is a not-so-subtle
aspect that seems to be often missed in moving out from the chain analogy to
a whole system perspective. We need to
look at this in a little more detail. Notice also that in the verbalization above,
“constraints” is plural, not singular.
It is saying that from a system perspective, in any one system, we
should expect more than one constraint.
Initially this too appears to be in direct contradiction with out
simple chain analogy. In fact, I will
suggest that it has been further complicated by a subsequent change in the
verbalization. Goldratt subsequently “put the S
in the word constraints in parentheses,
because there might be a system that had only one constraint (3).” The new verbalization is thus; (1) Identify the system’s constraint(s). (2) Decide how to Exploit the system’s constraint(s). (3) Subordinate everything else
to the above decision. (4) Elevate the system’s constraint(s). (5) If, in the previous steps, a
constraint has been broken, Go back to step 1, but do not allow
inertia to cause a system’s constraint. Is the difference in wording important? I think so.
Maybe by putting the “s” in a bracket Goldratt was in fact trying to
draw attention to the non-singularity of constraints in a system by
suggesting that such an occurrence would be exceptional (and as yet
undiscovered). However, it may also
have inadvertently reinforced the chain analogy with a singular
constraint. In any case, let’s adopt
the earlier verbalization, the one that was first presented above, the one
where plural constraints are explicit.
Except we need to make one change; subordination decisions must also
be plural, as others have also clearly indicated (4). Thus we have; (1) Identify the system’s constraints. (2) Decide how to Exploit the system’s constraints. (3) Subordinate everything else
to the above decisions. (4) Elevate the system’s constraints. (5) If in the previous steps a
constraint has been broken, Go back to step 1, but do not allow
inertia to cause a system constraint. The rationale for this plurality comes from
Goldratt’s own comments which we saw earlier in the page on the bottom line. "We very rarely find a company with a real
market constraint, but rather, with devastating marketing policy
constraints. We very rarely find a
true bottleneck on the shop floor, we usually find production policy
constraints. We almost never find a
vendor constraint, but we do find purchasing policy constraints. And in all cases the policies were very
logical at the time they were instituted.
Their original reasons have since long gone, but the old policies
still remain with us (5)." Although the word “policy constraint” has been more
recently disowned, maybe “policy issue” is more apt; there are still many
more than one constraint per system.
We know, however, that there can only be one weakest link in a chain,
so if there is more than one weakest link in a system we are in effect asking
how many chains there are in a system.
Let’s examine this. In order to answer this question. Let’s take our simple systemic model as an
example. And let’s draw it with a
physical constraint as we did on the measurements page, and let’s add two
products to our process. Here we have; one system, one
process, two products, and one constraint. Let’s make this a little more explicit. Let’s put a boundary around our system and
label the constraint. Both product A and product B share
the same set of physical resources or the same process layout, however we
could consider that we have in effect a chain for process A and that we have
a chain for process B, and that both of these chains share a common weakest
link. Let’s now add a third product – product C. Product C does not go through the process
constraint – after all, to do so would mean reducing the capacity to produce
products A & B. Therefore, the
demand for product C must be limited so as to not cause a capacity constraint
on any of the other common resources that it shares with products A or
B. If the demand is limited then
product C must be market constrained. Let’s draw this. So we now have; one system, one
process, three products, and two constraints. Another way of looking at this is that we
now have 3 chains in our system and 2 constraints. Let’s step this argument out one more level – an important
level conceptually. What if we have
two processes in our system? Let’s have a look. Now we have, one system, two processes, six
products, and four constraints.
Or to put it another way; we now have 6 chains in our system and 4
constraints. In industry, having two discrete processes is not so
uncommon. Consider for example a drill
bit manufacturer making the same range of products in both high-speed steel
and also in tungsten carbide. There
will be two similar but quite separate lines – maybe even totally separate
personnel. Too often we tend to think that the five focusing
steps – our plan of attack – applies to an individual chain, and in a very
simple system it might, but most often it applies to a system of chains. Moreover, even if the physical constraints
are limited, the policy constraints that underlie them will be numerous. We need the plural “s” in constraints. One last point.
All of the constraints in our diagrams above exist simultaneously in
time. And if we overcome each of these
constraints then the next set of weakest links will present themselves – also
simultaneously in time. The plural “s”
in constraints doesn’t apply to this series of sequentially uncovered
constraints in one chain; it applies to the co-existing constraints in
several chains. Let’s now look at each of these five steps in detail
and then we will examine some of the broader ramifications. In the process we will also draw some
distinctions between logistical constraints (physical constraints) and
non-logistical constraints (policy constraints) and how we approach them. Identify seems straight forward, especially if it is
a physical constraint in a logistical solution – a true bottleneck. If you ask people who don’t have direct
responsibility for the constraint but who do have responsibility for making
sure that orders or work are delivered on time, they will have a fair idea of
where the constraint is. The planners
will have a fair idea also of where the constraint is. If the situation is in sales or marketing
or some other non-logistical area, then the constraint will be a policy and
the best way to identify it is by using the Thinking Process to derive a core
problem or a core conflict. In the previous section we noted the cynics response
that “bottlenecks are at the top” in reference to the leadership. We decided this was unfair and indeed
totally untrue. Let’s invert the
analogy. Instead of a bottle, let’s
use a funnel analogy (6). We are
looking for a choke point, the neck, the narrowest point in the funnel
through which everything must flow. Exploit means that once a constraint has been
identified, all efforts must be made to properly utilize the capacity of the
constraint – to make the very best of the existing situation using the
resources that are at hand. If our
constraint was a funnel for instance, we would make sure that there are no
blockages lodged in the narrowest point.
We would also make sure that the material flows smoothly and
continuously through the constraint.
In fact, a funnel does this automatically to some extent by having a
reserve in the mouth above the neck – if you like a sort of buffer to ensure
constant supply to the neck. For a
physical constraint however, writing a plan or a schedule for best utilizing
the available capacity is a good way to exploit the existing capacity. A plan is really just a set of instructions
that provides for a timely and appropriate output. Taiichi Ohno had a very effective way of describing
how to exploit the existing capacity of a work center (7); Present Capacity = Work + Waste Stated in this way, exploitation of a constraint can
also be seen as eliminating waste and this is central to the way just-in-time
exploits capacity everywhere. The
particular wastes that apply in instance of a constraint are; the waste of
waiting, the waste of over-processing, the waste of over-production, and
waste of making defective products. As
the waste component is reduced at the constraints the work component can and
should be increased. We should also
make sure that there are not existing policy constraints or just plain old
bad habits that unreasonably limit the available working time either. Removing these will increase the work
component even further. How do we over-produce on a constraint? Easy, instead of making just what is
required for the “foreseeable” future, we make a bit more because it is more
efficient. A knowledge of the earning
capacity of different products as described in the measurements section is
also a good way to ensure full exploitation of an internal constraint. Non-logistical implementations may not have physical
constraints but they will have policy constraints. It is unlikely that we will try to exploit
or subordinate to these because the simplest way to overcome an erroneous
policy is to replace it. It doesn’t
cost money, it just takes thought, so most people will go from identify to
elevate automatically. However there
could be a situation where a policy is externally imposed – Government for
instance, in which case we probably won’t be able to change it any time soon
and we may indeed have to make the best of the existing situation that we can
– exploit the externally imposed policy. Focusing on the constraints will provide
unbelievable leverage. This leverage
occurs because for once we know exactly where to look and exactly what to
concentrate on. We have a very limited
number of places where concentrating on the detail complexity gives us a
unique handle on the dynamic complexity of the whole system. If “exploit” is the instruction for the constraint,
then “subordinate” is the instruction for all of the non-constraints –
everything else, both physical and policy.
Proper subordination is the key to effective
implementation of Theory of Constraints. Proper subordination means that the
non-constraints only do what is required to ensure maximum exploitation of
the constraint. We need to ensure that
the parts are subordinated to the whole, or more correctly in larger-scale
enterprises, that the subsystems are subordinated to the system. Once an exploitation plan has been decided upon,
then, take care to remember that there are two ways that we can deviate from
this plan (8). Deviating from the plan
absolutely means improper subordination and consequently less than fully
effective exploitation. Deviation from the plan here results from; (1) Not
doing what was supposed to
be done. (2) Doing what was not
supposed to be done. Often the most important part of subordination isn’t
just ensuring that we do what is supposed to be done; it is ensuring that we
don’t do what is not supposed to be done.
We could think of this as active restraint. Making sure that we don’t find work for
non-constraints just to keep them busy is a particularly important example of
active restraint. It avoids another of
Ohno’s wastes; the waste of stock-on-hand. To use our funnel analogy, this means pouring
material into the funnel at the same rate as it flows out, leaving just
enough in the mouth of the funnel to ensure the neck is always at full
capacity. Pouring the material in
faster will fill up the mouth of the funnel to overflowing, but won’t cause
any more material to flow through the neck.
We all know this from personal experience; we all blame the stupid
funnel, when actually it is always the stupid pourer! To put this in another way, pouring the material in
faster than the rate of the constraint is a waste. It is the localized waste of
over-production. We could look at
non-constraint capacity then as follows; Present Capacity = Work + Localized Waste of Over-production
+ Other Wastes The point in stressing the localized waste of
over-production is not because it produces excessive inventory and longer
lead times, but rather because it destroys sprint capacity. Sprint capacity on non-constraints is an
integral part of buffer management.
Buffer management in turn is an integral part of ensuring the
timeliness of our plan. We will return
to these concepts again in our logistical examples. Subordinating the local and system measurements to
the constraint and the goal of the system is also a key and under appreciated
facet of Theory of Constraints. If we
provide the wrong feedback then we can’t be surprised if we get the wrong
behavior. This means not only ensuring
that we do measure and communicate the effectiveness of the constraint,
buffer management, and shipping, but also that we don’t measure local
efficiencies or utilization of the non-constraints. The local performance measures of
inventory-dollar-days waiting and throughput-dollar-day late are important
parts of ensuring the correct subordination of subsystems to the goal of the
whole system. We will return to subordination issues in more
detail shortly. Elevation is the stage at which we bring in
additional resources in a logistical solution and raise the capacity of the
constraint in some way. This action will require some operating expense and/or
investment. The intention is to leverage off
the constraint so that a small elevation results in a substantial increase in
output. In a for-profit organization
we are trying to decouple net profit from operating expense. If in our funnel analogy we could bore out
the neck of our funnel, without having to buy a whole new funnel for
instance, we could elevate the constraint.
If we bought a second funnel we would also elevate the constraint. In the non-logistical solutions elevation means a
new break-through idea that enables us to do things that were previously not
possible. Generally speaking this is
not capital intensive, however as Newbold notes; if the leverage point is a policy, political capital may be needed to
be invested (9).” The last step is important. The first part; “If in the previous steps a
constraint has been broken, go back to step 1,” is straightforward enough, it
introduces a looping structure. It
reminds us that if we break a constraint then clearly we will have another
newer one to hunt out and deal with in the same manner. However, the meaning of the second part;
“but do not allow inertia to cause a system constraint,” is less clear. In this case the meaning is that when we do
break a constraint we should be careful not to let our satisfaction with the
new level of performance stop us from seeking out even higher levels of
performance. Colloquially, inertia is often interpreted to mean
that it is difficult to start moving something from rest, but it also has a
meaning that once something has started moving at a constant rate it is
difficult to change it to a new direction or to a new rate of movement. It is this second part of the meaning that
is important in this context. The best description of inertia comes from Mark
Woeppel (10). “How about a 43% annual
ROI? Could you sit there awhile? I
know a company that did. How about
taking your order fulfillment cycle from 3 weeks to 3 hours and stalling
there. I know another company that did
that.” These are rather fine examples
of inertia. The message is; once you
break a constraint, go and look for the next one; don’t accept the new
current and higher level of performance as satisfactory. There is something else blocking you from
going even higher. Go back and look
for it. Don’t let the satisfaction of
the current improvement constrain us from seeking even more improvement. I read the 5th step not only as an instruction to go
back and identify the next constraint – physical or policy, but also as an
admonishment to Don’t Stop. This is a process of on-going improvement –
POOGI – it’s a journey not a destination. There are several analogies that are used to
illustrate the focusing process. We
used funnels above, let’s use an hourglass (or a minute timer – depending
upon your experience) to summarize what we have learnt. Step 1: Identify. Step 2: Exploit. Step 2: Exploitation continued… Step 3: Subordinate. Step 4: Elevate. Step 5: Don’t stop. Don’t let inertia become a constraint to
the system. Hopefully this analogy, simple as it is, will help
us to remember what we are about in our focusing process; taking a system,
identifying the constraint to increasing the output and then maximizing
output within the context of the constraint and/or overcoming the constraint
completely – but not stopping there because there will always be room for
further improvement. In the measurements section we learnt to define; the
system, its goal, the necessary conditions, the fundamental operating
measurements, and the role of the constraints. This, it might be assumed, is what makes
Theory of Constraints systemic. All of
these factors are necessary but, of themselves, not sufficient to make the
approach systemic. If we recall
Goldratt’s definition from the introduction; a system approach is a warning
against “concentrating on a local optima (in place or time) and, by that,
jeopardizing the performance of the system as a whole (11).” Thus it is
subordination that makes Theory of Constraints truly systemic. It is subordination that stops us from
concentration of local optima. In addition to being systemic, Theory of Constraints
is also of course systematic. The
steps that we saw in the introduction of defining the system, the goal and so
forth, plus the process of change introduced in this section and the 5
focusing steps are the entities that make Theory of Constraints systematic as
well. Indeed the 5 focusing steps are the heart of Theory
of Constraints. You might wonder how
such a simple method can be applied to manufacturing or supply chain or
projects or marketing or strategy – but it can and it is. When you find an implementation drifting
astray you can guarantee that somewhere, someone, has lost sight of the 5
focusing steps. It is this simple 5
step focusing process that allows us to move from local optimization
everywhere to global optimization of the system – because now we know exactly
what to look for and exactly what to do when we find it. In the previous section on people we were left
hanging over the cause of the conflict between transfer points. We agreed that; Conflict
arises not because people are failing to do their best, but because everyone is
doing their best. There
certainly seemed no way out of this dilemma, at least not under the
reductionist/local optima approach of efficiency everywhere. However, now that we have seen the 5
focusing steps we are in a much better position to understand that the
definition of what is best is actually different for different parts of the
system under a systemic/global optimum approach. Our erroneous assumption, the assumption
that gave rise to the conflict, was that we must have maximal utilization
everywhere. Of course we can not have
maximal utilization everywhere but we didn’t know that, or at least we
weren’t allowed to acknowledge it. We now
appreciate that what is best actually depends upon whether the area is a
constraint or a non-constraint.
Constraints should be utilized as effectively and as efficiently as
the market demands. Non-constraints should
be utilized as effectively and as efficiently only as much as the constraint
demands. As non-constraints by
definition and fact have more capacity than constraints they will not be
utilized as much as the constraints are.
And if the physical constraints are very few in number, then in
relative terms the non-constraints must be very many in number. Thus most resources are going to have
significantly less productive time. The previous conflict was due to the erroneous
assumption that doing our best, is the same as maximal efficiency
everywhere. In fact, we could never
achieve that because of dependency and variability – but that wasn’t the
cause of the conflict. The cause of
the conflict was the assumption that WE MUST be efficient everywhere in order
for the whole system to be efficient. Now we know that for the system as a whole to be
efficient, we must be efficient in a few critical places and less efficient
everywhere else. What do you think
will happen to conflict between spans of control is we add this to our
understanding? It will be
substantially reduced. It will never
be totally removed because there will always by detail complexity issues
between spans of control, but now we have true goal congruence. Everybody knows what to do in order to do OUR BEST for the system. There is only
one place, the constraint, that can truly operate at full utilization all of
the time. All other areas must operate
at less than their full utilization for most of the time. The only time non-constraints can operate
at full utilization is when there has been an interruption and there is
“catching-up” to be done. At this time
non-constraints dip into their sprint capacity. At all other times non-constraints must practice
active restraint. Maybe active
restraint explains better what non-constraints must do, but now we are left
with different definitions of doing our best depending upon whether we are
talking about a constraint or a non-constraint. Stein, however, offers an all-embracing
definition of doing our best (12); Resources are to be utilized in the creation
or protection of throughput, and not merely activated. Now we need
only to ask whether a resource is a constraint or a non-constraint. It is both the constraint and
non-constraints that are involved in the creation of throughput, but it is
only the non-constraints that are involved in the protection of
throughput. This is
the definition of doing our best according to the goal of the system. Let’s apply
this idea to our diagram of a simple system – and now that we know of the 5
focusing steps let’s also include a constraint as we did in the section on
measurements. When the non-constraints are not
utilized in the creation of throughput they are utilized in the protection of
throughput (in fact, conserving their sprint capacity). The constraint however is always utilized
in the creation of throughput. “What
makes a constraint more critical to the organization is its relative
weakness. What distinguishes a non‑constraint
is its relative strength, which enables it to be more flexible (13).” We must protect that flexibility in order to
protect throughput. What about the situation where the constraint is
external to the system? Let’s have a
look at that also. Now, all areas within the system are non-constraints and either
utilized in the creation of throughput or they are utilized in the protection
of throughput. There is no capacity
constraint within the system; everything within the system is subordinated to
the external constraint. There is
another way we can look at Stein’s description of resource utilization –
especially resource utilization of subordinated non-constraints, and this is
by way of a traffic light analogy. When a
non-constraint is working it is creating throughput, when a non-constraint is
not working it is creating protection for throughput. We can consider this as stop and go, or on and off, or creation and
protection. However, there is another
common but more insidious situation, a half way position, a go slow position,
let’s look at that. When we slow down we do two things.
We often fail to fully create throughput and we often fail to fully
protect throughput. It is a double
edged sword. Basically we destroy
throughput. We destroy throughput on
two different fronts at the same time.
We need to gain a much better appreciation of this. When the goal
is adequately articulated; and the role of the constraints, the
non-constraints, and the policy and measurement systems are aligned with the
goal, then people will subordinate.
It’s not against human nature to do so. The non-constraints will find some
startlingly productive things to do.
System productivity won’t go down, it will go up – you’ll see. There are 3 generic logistical solutions, they are;
manufacturing/processing, supply chain, and project management. They are generic in the sense that the same
basic approach to each class of problems will yield a workable solution to
all cases in that class. In each class
the objective is to move the system towards the goal of the system – most
usually increased monetary throughput or physical output. This is done through an understanding of
the system’s constraint and knowledge of the system dependencies and
variation. In the introduction we discussed batching in terms
of quantity of goods or quantity of time.
When can also use this same criterion – quantity of goods or time – to
differentiate the logistical solutions into those that aim to increase
physical output (and hence monetary throughput) or increase available time
(and hence monetary throughput). Thus
manufacturing and supply chain aim to increase physical output (goods sold)
by identifying, exploiting, and subordinating to a process constraint. Project management aims to increase
available time (shorten project durations) by identifying, exploiting, and
subordinating to a duration constraint – the critical path. Moreover, within manufacturing and supply chain,
manufacturing most often exploits the constraint by addressing (and often
reducing) batching of goods. Supply
chain most often exploits the constraint by addressing (and often reducing)
batching of time. The graphical trees of the thinking process are used
to customize the solution to the specific circumstance and communicate and
obtain agreement. In the non-logistical solutions; marketing, sales,
strategy, and similar problems, there are no generic solutions. There are certainly very consistent
approaches to these problems, but the solutions are always situation
specific, or as the Japanese would say; “case by case.” The graphical trees of the thinking process are used
here to; first analyze the situation, then develop and customize a situation
specific solution and communicate and obtain agreement. Senge talks about micro worlds and transition
objects, we know these better as games and simulators (14). Such things are very important parts of the
process of change. Aircraft,
especially commercial aircraft, are very expensive and therefore training
(and continuous retraining) is carried out on simulators – a less expensive
“version” of the aircraft secured to a concrete floor. Before a commercial pilot ever sets foot
inside a large aircraft we would expect full command of a simulator in which
the pilot has been exposed to all the known contingencies that might possibly
be met in the course of service. Companies, especially commercial companies, are also
very expensive and therefore training (and continuous retraining) is carried
out on …… Well, actually its not done
on simulators is it? It’s done in real
time, real life, seat-of-your-pants, and as for retraining, let’s not go
there. There is of course nothing
wrong with experience – if you can get it.
But how do you gain experience for all those small problems that end
in a system crash (insolvency) without actually going there? Fortunately the system, a business, is a lot slower
and a little more forgiving than an aircraft.
Access to the collective history of other business failure and success
is available via the print media and personal mentoring by others who have
“been there, done that.” But even with
access to that experience we can’t always internalize the experience for
ourselves (short of real time, real company) without running some sort of
pilot. This is because we are dealing
with dynamic complexity – cause and effect removed in time and space from the
decision process. A business simulator
gives us an opportunity to run a pilot on a personal computer and experience
first-hand the effects of our decisions. Within Theory of Constraints there are a number of
excellent simulators designed to allow us to build these experiences with
safety. Indeed the P & Q analysis
is nothing more than a very unsophisticated (but elegant and effective) pen
and paper simulator. With more
sophisticated simulators you can indeed run a process into the red and turn
the computer off at the end of the day – equally, and in fact much more
likely, is that as a consequence you will learn to run a system superbly –
even in the presence of multiple dependency and variation. The production self-learning kit (see
resources) is still the best around for manufacturing/process learning. There are also simulators available for
project management and distribution.
Two more generalized simulators are available for drum-buffer-rope
(15) and enterprise resource planning (16).
These are important learning tools and part of the reason for rapid
and substantially success in implementing a Theory of Constraints’ based
solutions in a business. We began our discussion of the bottom line with a
suggestion that many people believe that in order to increase profit we must
increase production, and that in order to increase production me must
increase operating expense. Really
this is saying we must elevate the system.
Let’s draw this idea. In fact, in discussing
measurements we made the observation that this idea is so entrenched that
many people will reel off that; in order to make money you have to spend money. Let’s
examine this in more detail. There is a saying in English that goes; “if I had a
dollar for every time that …., then I would be quite rich.” You can add whatever is bothering you to
complete the sentence. Let me complete
the sentence with something that bothers me. If I had a dollar for every time that I was told
that “yes we understand about bottlenecks, we have
identified several, and if any one of them becomes serious enough we will
spend some money and increase the capacity,” then I would be quite
rich. Do you see what such a statement tells you? This is only one step removed from simply
raising production across the board.
One step removed because there is at least an awareness that some
resources appear to be more constraining than others. Let’s draw this. Indeed, if capital is spent and
additional capacity is purchased, what do you think the chances are of
meeting the original payback calculations?
Very little. So, at one step removed
from the stage of simply increasing production; we are at least aware of the
identity of some apparent constraints and that elevating them will have a
positive effect. For many
reductionist/local optima proponents the focusing process is just this; identify,
elevate, identify, elevate, etc… In
fact, the identification isn’t active, it’s passive, and the apparent
constraint must stand out and shout to identify itself. Let’s extend the corollary then to; if I had a
dollar for every time that I was asked to wait until new capacity was install
and productivity increased, then I would be even richer. What does your experience tell you? New capacity never attains the anticipated
production levels put forward in the CAPEX; output is always lower, and
payback is always longer than planned.
Do you ever wonder why this should be so? Basically, we fail to properly exploit our
new capacity. Although it must be said
that companies with limited capital resources, companies which must be
frugal, may be better at this. Often when there is a great deal of work-in-process
we have phenomenon of “wandering bottlenecks” which is really to say that we
have a phenomenon of wandering large batches.
Without some skill, and batch size reduction, it is hard to know where
the real bottlenecks lie. We may
simply have broken (elevated) something that wasn’t really a constraint, or
broken something with more capacity than was needed. Therefore, at two steps removed from simply
increasing production; we are aware of the identity of apparent constraints,
we know elevating them will have a positive effect – and we know that we
should at least try to exploit the new situation maximally. Let’s draw this. However, once again, when it comes
time to elevate the constraint the results will still be less than
anticipated. This is the level that
most companies using a reductionist/local optima approach will reach. The focusing steps are; identify, elevate,
exploit – in that order. To be truly
effective we must go one step further.
We must subordinate as well. At three steps removed from simply increasing
production; we are aware of the identity of the constraints, we know to
exploit them, and we know that elevating them will have a positive effect –
but we also know that first we should try to subordinate the existing
situation maximally. This means not
doing things that we don’t need to do, and making sure that we do things that
we do need to do. In production for instance this means releasing work
at the rate and lead time necessary to ensure that the constraint is utilized
effectively – and to monitor and measure the work that doesn’t arrive at the
constraint in good time. This and this
alone will tell us where the next true constraint is. Without knowledge of the next constraint we
can’t properly make a decision on how to elevate the current constraint and
we will nearly always over-estimate the anticipated increase in
productivity. This is the
systemic/global optimum approach. This
will give you the biggest bang for your buck.
Let’s draw it. Be careful, our reductionist/local
optimization history and experience is action based, not doing things that we
don’t need to do requires restraint and understanding, a conscious non-action
action. Yet, this is the key to
effective implementation. Let’s state
it; subordination is the key to effective implementation. Subordination is the
step most misunderstood and most misapplied. The reductionist/local optima approach will always
fall short of the full potential achievable by the systemic/global optimum
approach. In the introduction we learnt that under Theory of
Constraints “Local optima do not add up to the optimum of the system as a
whole.” Identify, exploit, and elevate
alone are just local optimizations.
Subordination makes the 5 step focusing process systemic; “The key to
know what to do locally is the realization of the role the system constraints
are playing (17).” We can’t know where
the system constraints are without proper subordination. The focusing process is iterative; Goldratt calls
this POOGI – a Process
Of On
Going Improvement. POOGI is quite a common term amongst Theory
of Constraint practitioners but rather scarce in published sources and
therefore although the term “a process of on-going improvement” is used in
1986 (18) the period at which it morphed into POOGI is obscured. In a sense POOGI is similar to the PDCA cycle of TQM
except that maybe POOGI is more likely to be applied to the dynamics of the
process or system and PDCA is more likely to be applied to the detail of a
product or individual operation. These
are differences of kind, not absolute.
Let’s examine this a little more closely. Stata showed that in an environment of continuous
improvement, the rate of improvement in individual improvement projects
tended to be logarithmic nature (19).
This should be no great surprise; logarithmic rates of change are
common in nature. What it really says is that early on you get large
rates of change for a given effort; later on you get less and less additional
improvement for the same amount of continuous effort. We all know from our own personal
experience that when we learn or do something new, that after the first
tentative steps we progress quickly, and then as we become more proficient
further improvement seems slower and slower.
Let’s graph this trend. This is a graph
of incremental improvement – its fine if
you can get it. This is largely the
trend that you would get applying the PDCA cycle to an individual project, so
let’s add that as well. How is POOGI different from that?
Well, often we break a constraint before we finished completely
exploiting it – or maybe even beginning to elevate it. The improvement in productivity in the
early stages is sufficient that something else becomes more binding on the
system than the constraint that we are currently working on, and we must turn
our attention to the next constraint.
Let’s draw that in. So in fact we “jump” the system up a bit. That doesn’t mean we necessarily stop our
continuous improvement initiatives, but it does mean than we have moved our
system focus somewhere else. If we
continue to do this – break constraints – we can get significant improvements
in very short periods of time. Now where does POOGI fall in here?
Let’s see. This is fundamental
improvement. POOGI best
describes the individual fundamental improvements that keep jumping the
system up to new levels of achievement.
Now ask yourself, for any given amount of time, would you prefer to
follow the curve outlined by PDCA, or would you prefer to follow the curve
outlined by POOGI? Debra Smith considers that the
Theory of Constraints can be summed up simply with two words: focus and leverage
(20). We focus on each new constraint
and leverage off of it. That is what
POOGI does and that is what the diagrams above show. We could also imagine a physical location for the sequence for this
leveraging process regardless of whether the constraints are physical or
policy. The locations could be
departments within a plant, or they could be plants or functions within a
corporation. Let’s draw it. Each time we focus and elevate a
constraint we elevate the whole system, we leverage off of the
constraint. Leverage points and
constraints are two sides of the same coin.
Senge notes that small changes can
produce big results – but that the areas of highest leverage are often the
least obvious (21). Naturally, if we
had already seen the leverage points/constraints we would already have done
something about them! It comes back to
our maps of reality, if our reality is one of local optimization we won’t see
the leverage points that arise from global optimization. Now, however, we have in the
Theory of Constraint logistical and non-logistical solutions a means to hunt
out constraints – the very non-obvious leverage points that Senge is looking
for. We’ve covered a lot of ground in the last few pages. In the page on measurements we introduced
the rules of engagement; (1) Define the system. (2) Define the goal
of the system. (3) Define the necessary
conditions. (4) Define the fundamental
measurements. (5) Define the role of the constraints. In the page on measurements we also defined each of the items in some
detail but we were left hanging over the role of the constraints. As we moved into the page on people we were
able to resolve some of the conflict over local versus global optimization
but once again we couldn’t entirely resolve the conflict over the roles of
various parts of the system – essentially we still couldn’t resolve the role
of the constraints. In fact we had to
wait until this page to completely answer some of the questions that have
been bothering us. We did this firstly
by introducing the process of change, which is characterized as; (1) What to change. (2) What to change to. (3) How to cause the change. The process of change concept caused us to address the “what to change
to” more critically than previously.
Using the concept of leverage we were able to narrow-in or focus-on
the constraints. We introduced
Goldratt’s generic sequence of five focusing steps; (1) Identify the system’s constraints. (2) Decide how to Exploit the system’s constraints. (3) Subordinate everything
else to the above decisions. (4) Elevate the system’s constraints. (5) If in the previous steps a
constraint has been broken Go back to step 1, but do not allow inertia to cause a
system constraint. In other words; Don’t Stop Note that we have also now appended the warning about not stopping to
the last step. In fact the five focusing steps – our plan of attack – basically
“falls out” of the rules of engagement; a cascade or a drop-down. We can illustrate this graphically. Let’s have a look.
With the five focusing steps in place we were able
on this page to define the role of the constraints fully for the first time,
and by difference, the role of the non-constraints. However, initially, we still couldn’t resolve
the last conflict; how to ensure that everyone could do their best when there
are both constraints and non-constraints present. In order to resolve this dilemma we had to make sure
that the concept of subordination was fully understood. Too often the steps in the focusing process
looks too similar to the steps in the reductionist/local optima approach and
we fail to recognize how absolutely critical subordination is to the success
of the systemic/global optimum approach – the approach of Theory of Constraints. With a full appreciation of subordination we can at
last understand the measure for “doing our best” regardless of position
within the process; Resources are to be utilized in the creation
or protection of throughput, and not merely activated. This is
the definition of doing our best according to the goal of the system. With such knowledge we can examine any problem from
first principles and derive a satisfactory solution that will enable the
parts of the system to move in unison towards obtaining the common goal of
the system. However, in
order to move in unison towards obtaining the common goal of the system, we
must first obtain agreement to the proposed change. And that is the topic of the next page. (1) Senge, P. M., (1990) The fifth discipline: the
art & practice of the learning organization. Random House, pp 150-151. (2) Goldratt, E. M., (1990) What is this thing called Theory of
Constraints and how should it be implemented?
North River Press, pp 3-21. (3) Goldratt, E. M., (1990) The haystack syndrome: sifting information out of the data
ocean. North River Press, pp 58-61. (4) Schragenheim, E., (1999) Management dilemmas:
the Theory of Constraints approach to problem identification and
solutions. St. Lucie Press, pp 5-7 (5)
Goldratt, E. M., (1990) What is
this thing called Theory of Constraints and how should it be
implemented? North River Press, 162
pp. (6) Woeppel,
M. J., (2000) Manufacturer’s guide to implementing the theory of
constraints. St. Lucie Press, pp
16-17. (7) Ohno, T., (1978) The Toyota
production system: beyond large-scale production. English Translation 1988, Productivity
Press, pg 19. (8) Goldratt,
E. M., (1990) The
haystack syndrome: sifting information out of the data ocean. North River Press, pg 146. (9) Newbold,
R. C. (1998) Project management in the fast lane: applying the Theory of
Constraints. St. Lucie Press, pg 232. (10) Woeppel,
M. J., (2000) Manufacturer’s guide to implementing the theory of
constraints. St. Lucie Press, pg 41. (11) Goldratt,
E. M., In: Cox, J. F., and Spencer, M. S., (1998) The constraints management
handbook. St Lucie Press, pp ix-xi. (12) Stein, R. E., (1994) The next phase of
total quality management: TQM II and the focus on profitability. Marcel Dekker, pg 105 (13) Schragenheim,
E., and Dettmer, H. W., (2000) Manufacturing at warp speed: optimizing supply
chain financial performance. The St.
Lucie Press, pg 33. (14) Senge, P. M., (1990) The fifth
discipline: the art & practice of the learning organization. Random House, pp 313-325. (15) Schragenheim,
E., and Dettmer, H. W., (2000) Manufacturing at warp speed: optimizing supply
chain financial performance. The St.
Lucie Press, 342 pp. (16) Ptak, C.
A., with Schragenheim, E., (2000) ERP: tools, techniques, and applications
for integrating the supply chain. St.
Lucie Press, 424 pp. (17)
Goldratt, E. M., In: Cox, J. F,
and Spencer, M. S. (1998) The constraints management handbook. St Lucie Press, pp ix-xi. (18) Goldratt, E. M., and Fox, R. E., (1986)
The Race. North River Press, pg
143-150. (19)
Stata, R. (1989) Organizational learning – the key to management
innovation. Sloan Management Review,
Spring, pp 63-74. (20)
Smith, D. (2000) The measurement nightmare: how the Theory of Constraints can
resolve conflicting strategies, policies, and measures. St. Lucie Press, pg 137. (21) Senge, P. M., (1990) The fifth discipline: the art & practice of the learning organization. Random House, pp 63-65. This Webpage Copyright © 2003-2009 by Dr K. J. Youngman |