Effective Exception Reports
Weibull plots provide graphical input to help define clearly the difference between common cause variability and special cause variability. Exception reports or variance reports in production facilities are generally directed toward explaining special causes which have names for the deficiencies in production output. Details are explained in the October 2008 problem. Figure 1 shows a clear example of special causes separated from common causes with the cusp designating the reliability of the process.
For Figure 1:
● the cusp at 91% separates the steep trendline of common cause variability from
the special cause variability with a shallow trendline
● special cause losses (the sum of all gaps between the data points and the
production trendline below 91%) are 10,184 tons/year,
● common cause losses (the sum of all gaps between the nameplate line and the
production line) are 83,133 tons/year (don’t be trapped by the logarithmic
X-axis scale) which totals to a hidden factory of 93,316 tons/ year. Next,
monetize the losses so you speak the language of business.
Suppose for Figure 1 the gross
margin $’s are $500/ton:
● special cause losses are $5,092,000/year, ßProblem #2
● common cause losses are $41,566,500/year, ßProblem #1
● total hidden factory loss is $46,658,000/year problem—this is not a chicken
feed issue! The problem order for solution is based on the Pareto distribution.
Notice that 10.9% of losses are due to special cause and 89.1% of losses are due to common causes. The annual losses are equivalent to 93 days of production not achieved! Compare the % losses to Demings numbers below which are based on his experience.
Most accounting departments will tell you about the $46,658,000 problem, however, they will not usually be able to tell you the specific items needed for correcting the losses nor will they define the skills need for corrective action. The Weibull plot tells the special cause losses need to be corrected by the reliability and maintenance departments. The Weibull plot also tells the common cause losses are squarely in the hands of the black belt problem solvers and directly in the hands of management to resolve.
For Figure 1, the Pareto distribution sets the work priority as first solving the common cause losses and second on solving the special cause losses. Since different skills sets are required, the problems can be approached simultaneously.
For the plant in Figure 1, the original emphasis of the study was on reducing maintenance costs (a typical scapegoat) for the special causes which is not the most important medicine for this facility. The real issue in dire need solving first is reducing production variability to reduce common cause variability. If the campaign to reduce common cause was 50% effective and reduced common cause variability losses to 41,000 tons/year and this would have produced a savings 4 times larger than reducing all special cause events! The point is obvious: you must know where to attack the problem and the real problem may not be your typical scapegoat.
The adage: If you’re a carpenter,
then most of your problems must look like a nail applies to our skill
● if you are a maintenance guru, all your problems will appear as reducing
maintenance efforts (the small portion of the data set in Figure 1),
● if you’re a master black belt problem solver all your problems will appear as
reducing variability in production (the large portion of the data set in
● if you’re the owner/stockholder of the company in Figure 1 you will see all
your problems as low returns on your investment with expectations that
management will make the corrections before you abandon your
investment (or you abandon your management team for more
effective problem resolution) for greener pastures.
Immediate corrective action is required so you achieve bigger bang for your money! Action is needed NOW, you can’t not wait forever for improvements!
What Is The Issue?
Shewhart-the father of statistical process control to reduce variability-
In 1931, Walter A. Shewhart of Bell Telephone Laboratories wrote his important book on the Economic Control of Quality of Manufactured Product. This book established statistical process control in manufacturing (for the purpose of economic control of manufacturing by reducing/eliminating variability). It is still available after more than 80 years.
In chapter 1 and 2, Shewhart defines two categories of variability:
1) Common cause variability occurs from undefined reasons which we have not
yet named and identified for corrective action.
2) Special cause variability occurs from reasons we can name, we can find the
reasons, and we can eliminate the problems.
Characteristics of common cause and special cause variable are detailed in Table 1 of the October 2008 problem of the month.
Now here comes the problem:
● What if your common cause problems are named with special cause events
that are not real?
When this happens reporters of problems are naming “ghost problems” which are not in touch with reality. Those reporting the problem don’t know special cause variability from common cause variability. The result will be as in Aesop’s Fable no one will believe the cry of Wolf and therefore real problems are not solved. Remember the moral of the fable about the Boy Who Cried Wolf :
● “Even when liars tell the truth, they are never believed”.
Therefore problems must be correctly identified for corrective action and you must sell (not just tell) the solution!
father of involving Japanese management to reduce variability-
In 1982, W. Edwards Deming, the technical guiding light for the Japanese quality movement, wrote his important book Out of the Crisis. In chapter 1 Deming makes this point: “Any substantial improvement must come from action on the system, the responsibility of management.” This means management cannot take a Pontius Pilate ritual of washing their hands to show they are not responsible for special causes and common causes—actually, most managers are in over their head with both special cause and common cause problems which cause financial damage to their companies (it’s just convenient for them to point fingers to others as the responsible party).
Management must work to train the workforce in responsible exception reports. Management must train to identify special cause reasons for variances and also to lead the workforce in attacking common cause variability. The “medicines” for solving these two different problems are different.
Likewise internal consultants such as six sigma experts and maintenance/reliability experts much also carefully separate the common cause issues from the special cause issues to make sure the variance reports are correct. The same guidelines for correctly categorizing and naming variances apply to outside consultants for cost effective and timely resolution of problems.
The motivation for reducing/eliminating special causes is to reduce known problems which reduce productive output and increase operating costs these issues are part of the hidden factory robbing profits. The motivation for reducing/eliminating common causes is to reduce unknown and unidentified problems that add variability to productive output which increase operating costs. The common cause problems do not have names and thus are truly part of the hidden factory; and when the problems are named, they become special causes.
Misreporting of common cause
variability as special causes has roots in two of the Seven Deadly Sins of
Management which are:
● Failure to train (train to insure variance reports are correct) and
● Letting the job grow like Topsy (train to insure the workforce
reports deviations in a standard format)
Both deadly sins of management require managerial discipline and leadership to achieve good results. Deming also points out in Chapter 2: “There is no substitute for teamwork and good leaders of teams to bring consistency of effort, along with knowledge.”
No medical illness is corrected or avoided until the illness is properly named—the same is true for manufacturing! Most manufacturing processes do not match the requirements for a world class process, thus the illnesses must be properly named for resolution. Misnaming the illness in manufacturing dilutes your problem solving efforts with waste and ineffectiveness.
If you have variation reports quantifying the amount of money lost, this identification is not effective until the root of the financial variation is named. You must name the problem to focus resources for resolution. If the problem is incorrectly or vaguely identified (i.e., naming a common cause problem with an incorrect name) then effort is wasted. Focus on the illness is mandatory for better financial performance.
In short, Deming requires correctly
identifying the special cause problems with names and the correcting
identifying the common cause problems to convert them into special
causes. In chapter 11 Deming reminds
“The central problem in management and in leadership…is failure to understand
the information in variations.” Deming
● “..the type of action required to reduce special causes of variation is
totally different from the action required to reduce variation and faults
from the system itself [reduce common causes of variation] …”.
● “We speak of faults of the system as common causes of trouble, and
faults from fleeting events as special causes.”
● “Confusion between common causes and special causes leads to
frustration of everyone, and leads to greater variability and to
higher costs, exactly contrary to what is needed.”
Deming further explains “…experience shows 94% of troubles belong to the system which is the responsibility of management [to correct the common cause variability] and 6% of troubles are due to special causes [which are easier to correct].”
“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 causes)”.
2. “Ascribe a variation or a mistake to the system (common causes)
when in fact the cause was special [cause]”.
“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”. “The action required to find and eliminate a special cause is totally different from the action required to improve the process [common cause].” “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.”
How to categorize correctly common causes and special causes:
Process reliability plots of production outputs from a process give a clear roadmap of the issues and magnitude of the problem. They are frequently used with exception reports from the production area. Here is what you should expect to see for special causes on a Weibull probability plot for process reliability issues in Figure 2 with all special causes identified below the cusp.
Figure 2 shows the proper designation of special cause losses below the point estimate of the production system reliability point where common causes cease and special cause begin. Of course you don’t know the reliability point until all data are in hand.
Figure 2: Special Cause Losses
As a rule of thumb, report all
special causes where production is smaller than the nameplate at 99.9% reliability
as shown in Figure 3—of course if the reliability for the process is poor (say
50% or lower reliability for the process) this rule of thumb is inadequate and
thus choose a point on the nameplate line at a smaller reliability value to
catch the special cause events. The
production line with line slope, b, at
5.5 is truly an inferior process with great variability. Likewise the nameplate line slope, b, at 12.5 is also nothing to write home about
but to the owners of the process the nameplate line (established by judgment)
will look like attempting to
By the way, if you are a theoretical person, you will argue the nameplate line should be vertical, however if you establish the theoretical vertical line, experience teaches the production team will consider it impossible and thus they will not rise to the challenge of improvement—it’s the carrot or stick challenge. In Figure 3, all the data points have labels which are a feature of SuperSMITH™ Weibull and Visual software.
If exceptions are reported for the common cause portion of the line, then retraining is required to correctly understand variance definitions and what is causing the variance. Otherwise the cry of wolf will occur to frequently and the variance report will be dismissed as fiction for too many reports in the common cause section of the curve as it will not represent reality.
Unfortunately, Figure 4 is a more typical pattern of reported variances. Notice the larger number of variance reasons and how they are reported up and down the common cause trendline which is a false explanation of reasons for the variability. If you can name the reason for variability, it is NOT common cause variability and thus a false report in the section of the Weibull plot for common cause variability. In Figure 4, for SuperSMITH Weibull version 5.0BI and beyond, each label is turned on individually (go to the magnifying glass icon, select option for point symbol display, activate the type of point as label, and then a separate menu appears to select all or by individual label items with the labeled items appearing one at a time).
Figure 4: Reasons Reported For Common Cause Variability
Generally speaking, the reported reasons for the common cause variability in Figure 4 are not actionable items for the reducing common cause variability. The management of this operation are the key for finding and resolving the common cause variability and it is not an easy task to correct because the system is not well disciplined. In short, the management of this operation must put “spit and polish” into the operation to reduce the variability and this usually means institutionally the discipline of operating by the numbers/procedures to avoid the high variability of a Keystone Kops effort of being extremely busy but not productive.
Other Process Reliability References:
You can download other articles from this site concerning process reliability:
· Summary of Process Reliability, June 2008
· Process Reliability Punch List March 2005
· Process Reliability Line Segments April 2004
· Nameplate Capacity March 1998
· Coefficient of Variation February 1998
· Six Sigma January 1998
Production Output/Problems May 1997
Papers On Process Reliability As PDF Files For
--Process and Equipment Reliability May 2004
--Process Reliability: Do You Have It?—What’s It Worth To Your Plant To Get It? March 2002
--Process Reliability December 2001
--New Reliability Tool for the Millennium: Weibull Analysis of Production Data October 2000
--Process Reliability and Six-Sigma March 2000
Refer to the caveats on the Problem Of The Month Page about the limitations of the above solution. Maybe you have a better idea on how to solve the problem. Maybe you will find that I’ve screwed-up the solution and you can point out my errors as you check my calculations. E-mail your comments, criticism, and corrections to Paul Barringer.
March 19, 2011 – Revised
April 21, 2011
© Barringer & Associates, Inc., 2011