Vous êtes sur la page 1sur 7

Concept of Variation

Common causes and special causes of improvement. Stable System.

Shewhart used the term “assignable cause” of variation where as I use the
term “special cause.” I prefer the adjective “special” for a cause that is
specific to some group of workers, or to a particular production worker, or to
a specific machine, or to a specific local condition. The word to use is not
important; the concept is, and this is one of the great contributions that Dr.
Shewhart gave to the world.

Common causes belong to the system.

What is the system?

To people in management, the system consists of

Style of management
Employees – management and everybody
The people in the country
Their work experience
Their education
The unemployed
Government
Taxes
Reports
Tariffs
Impediments to trade and industry
Requirements to fill positions by quota, not by competence
Quotas for import and export
Foreign government
Quotas for import and export
Manipulation of currency
Customers
Shareholders
Bank
Environmental constraints

I should estimate that in my experience most troubles and most possibilities


for improvement add up to proportions something like this:

ARM Page 1
SQC
94 % belong to the system (responsibility of management)

6 % special

No amount of care or skill in workmanship can overcome fundamental faults


in the system.

Discovery of a special cause of variation, and its removal, are usually the
responsibility of someone who is connected directly with some operation
that yields data (for the control chart).

Some special causes can be removed only by management. For example,


production workers sometimes need engineering assistance to remove
problems connected with malfunction of machinery in use. It is the
responsibility of management to provide assistance when needed.

Fig. 1. Portion of a run chart for


miles per gallon between fillings
of the tank
30
28
Miles/gal

26
24

22
20

1/1/00 1/8/00 1/15/00


Date

The run of nine points below the average reflects a change. The cause was
attributed to bad spark plugs.

Another instance of management's responsibility for special causes occurs in


the current state of chaos in dealing with vendors. Production workers are
sometimes forced to use unacceptable or inconsistent raw materials or parts.
It is management's job to take corrective action to work with vendors to
improve the quality of incoming materials, and to halt the practice of
switching from one source to another.

ARM Page 2
SQC
Poor sales may stem from a faulty product, or from high price.

The worker at the machine can do nothing about causes common to


everybody on the job. He is responsible only for the special causes
chargeable to him. He can not do anything about the light; he does not
purchase the raw materials nor the tools. His job is to try to use them. The
training, supervision, and the company's policies are not his.

A riot occurs in a certain prison. Officials and sociologists turn out a detailed
report about this prison, with a full explanation of why and how it happened
here, ignoring the fact that the causes were common to a majority of prisons,
and that the riot could have happened anywhere.

Quality
improves
by removal
of special
cause

Proportion of items Stabilized


Found faulty at final
Audit
Continued
improvement Wishful
is expected thinking
but will not
happen.

1981 1982 1983 1984

Fig. 2. Typical path of frustration. Quality improves dramatically at first;


then levels off, becomes stable. The responsibility for improvement of quality
shifts more and more to the management, and finally almost totally to the
management, as obvious special causes are one by one detected and
removed, and quality becomes stabilized, unfortunately at an unacceptable
level.

ARM Page 3
SQC
Two kinds of mistake

We may now formulate two sources of loss from confusion of special causes
with common causes of variation.

1. Ascribe a variation or a mistake to a special cause when in fact the


cause belongs to the system (common cause).

2. Ascribe a variation or a mistake to the system (common causes) when


in fact the cause was special.

Over-adjustment is a common example of mistake no. 1. Never doing


anything to try to find a special cause is a common example of mistake no.
2.

It is easy to establish a clean record on either mistakes: never make mistake


no. 1, or never make mistake no. 2. But in avoiding one mistake, one
commits the other mistake as often as possible. There is no hope to avoid
both mistakes all the time.

The action required to find and eliminate a special cause is totally different
from the action required to improve the process.

Need for rules

Shewhart (about 1925) recognized the fact that good management consists
of one mistake now and then, and the other one now and then. What was
needed, he saw, is rules that can be put into practice by which to try to
achieve minimum net economic loss from both mistakes. To this end, he
contrived the 3-sigma control limits. They provide, under a wide range of
unknowable circumstances, future and past, a rational and economic guide to
minimum economic loss from both mistakes.

The Control Chart sends statistical signals, which detect existence of a


special cause (usually specific to some worker or group or to some special
fleeting circumstance), or tell us that the observed variation should be
ascribed to common causes, Chance Variation attributable to the system.

ARM Page 4
SQC
Statistical control

A stable process, one with no indication of a special cause of variation, is


said to be, following Shewhart, in statistical control, or stable. It is a random
process. Its behavior in the near future is predictable. Of course, some
unforeseen jolt may come along and knock the process out of statistical
control. A system that is in statistical control has a definable identity and a
definable capability.

In the state of statistical control, all special causes so far detected have been
removed. The remaining variation must be left to chance - that is, to
common causes - unless a new special cause turns up and is removed. This
does not mean to do nothing in the state of statistical control; it means do not
take action on the remaining ups and downs, as to do so would create
additional variation and more trouble. The next step is to improve the
process, with never-ending effort. Improvement of the process can be
pushed effectively, once statistical control is achieved and maintained (so
stated by Joseph M. Juran many years ago).

Removal of common causes of trouble and of variation, of errors, of


mistakes, of low production, of low sales, of most accidents is the
responsibility of management.

Sound understanding of statistical control is essential to management,


engineering, manufacturing, purchase of materials, and service. Stability, or
the existence of a system, is seldom a natural state. It is an achievement, the
result of eliminating special causes one by one on statistical signal, leaving
only the random variation of a stable process.

Many people that use control charts suppose that statistical control is the end
success of all effort. I have seen, for example, statistical control of
contamination, where the big problem is to get rid of contamination.

ARM Page 5
SQC
Marbles

  Movable Funnel


  Total
  Variation Felt-padded
base
Variation at Variation at
position A position B

Fig. 3. Deming’s funnel experiment to demonstrate the implication of


ascribing a variation or a mistake to a special cause when in fact the
cause belongs to the system (common causes).

 Initial position of the funnel was A and corresponding variation [i.e. scatter of
marbles] was the region occupied by black marbles.
 But due to lack of concept of variation, one may think, while getting a black marble at
the far left, that the setting is done incorrectly and moves the funnel to one’s right
[position B].
 This position will introduce certain variation. As a result total variation will be
increased.
 Other sources of extra variation
- Outlet diameter of the funnel is more than marble diameter.
- Increased height of putting marbles into the funnel.

ARM Page 6
SQC
Control Charts

The purpose of construction of control chart is to gather information


regarding

 When to leave a process alone

 When to take action.

Table 1 : Causes of Variation

Chance Causes Assignable Causes


Consists of many individual causes. Consists of one or just a few
individual causes.
Any one chance cause results in a Any one assignable cause generally
minimum amount of variation. For results in a large amount of variation.
example, slight variation in For example, a faulty set up, a batch
machines, slight variation in raw of defective or wrong raw material,
materials. operator blunder.
Follow statistical laws of variation. Do not follow statistical law
Chance variation cannot be Can be detected, elimination of these
economically eliminated from a causes is economically justified.
process.
With only random variation the Process is unstable to do so.
process is sufficiently stable to use
sampling procedures to predict the
quality of total production or make
process optimization.
The process is operating at its best If assignable causes of variation are
with minimum, inevitable variation present, the process is not operating
under the given techno-economic set at its best.
up. Such a process is said to be under
a state of Statistical Control or
simply under control.

ARM Page 7
SQC

Vous aimerez peut-être aussi