# DPMO

Viewing 40 posts - 1 through 40 (of 40 total)
• Author
Posts
• #45475

Ropp
Participant

150 units are produced and 12 defects are found. There are 2 opportunities for a defect on each unit. What is the DPMO?. What is the sigma value for this process assuming the dat collected represents long term variation?

0
#148722

Participant

Just  go  to  the  formula and  calculate  the DPMO,then  go  to  the  table  check  for  the  Sigma??

0
#148728

Geri
Participant

Use the DPMO calculator on the website

0
#148740

Prof. Paul Swanson
Participant

Paul and Gery are quite wrong.
12 defects in 150 is equivalent to 80,000 defects in 1,000,000.
Nothing more can be said.

0
#148748

Participant

Then  you  have  to  find  out  the  sigma  score  ,using  the  conversion table?

0
#148751

Prof. Paul Swanson
Participant

Forget about six sigma tables.  The data in these tables means nothing.  It is based on the false assumption of normality and 1.5 sigma drift.
You should never use six sigma tables.

0
#148755

lin
Participant

The professor is generally wrong.  Use the sigma table using DPMO, but recognize the corresponding “sigma” is just an estimate of the short term capability.  If your DPMO is about 6,800 (long-term capability) then your estimate of short-term capability will likely be about 4 sigma.  When using defect data, you can compute your long-term capability fairly well, but without switching to continunuous data, you can not estimate short-term capability.  In this case you just add “1.5 sigma” to create an approximation of the short-term capability.  If you have access to continuous data, then use it and compute the actual short-term capability.  The 1.5 sigma shift is just a best-guess model you can use if no other information is available.  The 1.5 shift is just a way to remove assignable causes from the long-term defect data so you can “guess” what the short-term capability could be.

0
#148756

Mikel
Member

Hey Prof (FYI – everyone on here is really impressed when people identify themselves as professors or PhD’s – we assume what follows is theoretical and meaningless, so keep on flagging your “status”),
Let’s see, it is identified as an assumption, how can that be false?

0
#148757

Ken Feldman
Participant

Hey Dave,
Why don’t you just post the entire exam instead of posting one question at a time.  Surely there are posters who have some burning need to spoon feed students too lazy to do their own work.  What you are posting are elementary and basic questions that you should have learned in GB training.  Evidently you didn’t listen and are too lazy to do your own research for the answers.  But, as long as there are posters willing to enable you why not keep posting.  Possibly one of them would be happy to take the test for you.  Maybe this why the field is full of hacks happy with just memorizing things instead of actually learning what LSS is all about.

0
#148764

Paul’s Mom
Participant

Professor Paul J. Swanson, Jr.
Visiting Professor
Trinity College
Dublin 2, Ireland
Room : 419, Aras an Phiarsaigh
Phone : +353-1-608 3705
e-mail: [email protected]
Associate Professor
Department of Finance
University of Cincinnati
Cincinnati, Ohio, 45221-0195
B.S. (Accountancy) University of Illinois, U.S.A. (1959)
B.S. (Economics & Finance) University of Illinois, U.S.A. (1960)
M.S. (Finance) University of Illinois, U.S.A. (1962)
Ph.D. (Finance) University of Illinois, U.S.A. (1966)
Dr Swanson is Associate Professor of Finance, University of Cincinnati and holds the
professional designation of Chartered Financial Analyst (C.F.A.).
Research Interests:
 Financial Management (Forecasting, Investment, International Finance)
 Clinical Mindfulness
Teaching Responsibilities in the School of Business Studies (Trinity):
 JS Corporate Financial Management (2003-2004)
 JF Introduction to Organization and Management (Section Leader, 2004-
2005)
Positions:
BACKGROUND OF MINDFULNESS
Dr. Swanson began his academic study of psychology in 1974 while a Visiting Associate
Professor in the Department of Psychological Sciences, Purdue University. Although his
major interest was in psychometrics and multidimensional scaling, he was introduced into
a field called Attribution Theory. His interest at the time was in a new field in Finance
termed ‘Behavioral Finance’. Dr. Swanson began to study Buddhist concepts to be used
in his courses on cultural and business differences in international financial management.
This has led over the past 8 years to deeper studies of Buddhist practice on a more
personal basis with Dr Leonard Rand, a clinical psychologist. He was most impressed by
a discussion with a Buddhist monk, Ajahn Charles, while leading a student group in
Thailand. Dr. Swanson has lectured at the Christ Church Cathedral on several occasions
on the subject of the procedures of mindfulness during 2004 and 2005. In May, 2004, he
lectured before the postgraduate medical students attending a class in Cognitive
Psychology of Dr. Tony Bates at the Department of Psychiatry, School of Medicine,
Trinity College, Dublin, Ireland. Dr. Swanson presented the keynote speech, on
November 24, 2004, to celebrate the start of the North Wales Clinical School organized
by the Centre for Mindfulness Research and Practice, Institute of Medical and Social
Care Research, University of Wales, Bangor. He has also participated in several courses
in Mindfulness given by the Centre for Mindfulness Research and Practice during 2005.
BACKGROUND OF FINANCIAL MANAGEMENT
Dr. Swanson has received six post-doctorate awards from the University of Cincinnati
for short-term studies at the University of Chicago (1972), Carnegie Institute of
Technology (1973), Massachusetts Institute of Technology (1967), Kelley School of
International Management (1998), and with several major United States corporations
in the area of financial forecasting.
He has held the rank of Visiting Associate Professor at Yale University (Economics,
1969), Purdue University (Psychological Sciences, 1975), and Stanford University
(Applied Mathematical Group, Stanford Linear Accelerator, 1979). He has also held
the rank of Visiting Professor of Finance at the University of Chicago (Graduate
School of Business, 1986), Oxford University (England, 1976), Trinity College
(Dublin, Ireland, 2002,2003-5), University of Warwick (England, 1975), the
University of Jordan (Amman, Jordan,1990), and as Professor-in-Residence,
Columbia Gas System Service Corporation (Wilmington, DEL, 1971),. He has served
as a Visiting Lecturer at the University of Hawaii (1984), the Tianjin Institute of
Textile Science (China,1988), and the Tokyo Institute of Technology (Graduate
School of Science and Engineering at Nagatsuta, 1988).
Dr. Swanson has held the following positions in academics, government and
business: Chairman, Finance Behavioral Research Group, University of Cincinnati,
from 1975 to 1981; President, Miami Valley Chapter (Dayton, Cincinnati Area), The
Institute of Management Science, 1970-71; Member, Advisory Board and Consultant
in Finance and Behavioral Studies, The Institute of Metropolitan Studies, University
of Cincinnati, Cincinnati, Ohio, 1968-78; Member, Board of Directors and Treasurer,
Consortium of Endowed Episcopal Parishes, Inc., Indianapolis IN, 1985-1989;
Consultant in Forecasting and Causal Studies, Stockpile Management, Resource
Preparedness Office, Federal Emergency Management Agency, Washington, DC,
1975-1980; Consultant in Finance, Dayton Region, United States Department of
Justice, Dayton, OH, 1985-1987; Member, National Defense Executive Reserve,
Washington, DC, 1970-1988; Consultant (unpaid) in Finance, Office of the Executive
Vice President, Finance and Investment, Royal Jordanian Airlines, Amman, Jordan,
1990.
Publications, Articles:
1. Solt, M. E. and P. Swanson Jr., “On the Efficiency of the Markets for Gold and
Silver,” Journal of Business (University of Chicago), Vol. 54, No. 3 (July,
1981), pp. 453-478.
2. 2. Swanson, Jr., P., “The Silver Market and the Hunts: A Preliminary Report
on Co-Integrated Vector ARIMA Methods,” Cheng Weimin, ed., Systems
Science and Engineering, ICSSE ’88, Beijing, China (Beijing, Peoples Republic
of China; International Academic Publishers – Pergamon Press: 1988), pp.
603-607.
Conferences and Seminars:
Dr. Swanson has presented thirty-six refereed papers at academic conferences and
conferences in the United States and overseas. He has presented over seventeen
seminars on ARIMA models and other statistical methodologies along with seven
other research lectures as a consultant or as tutorials at professional conferences
throughout the United States including the Lawrence Livermore Laboratories
(Chemical Engineering Division, University of California, and Berkeley, CA). Dr.
Swanson has served as a reviewer for the National Science Foundation (Finance,
Statistics), and is currently a reviewer in Finance and Statistics for the Journal of
Applied Business Research, and Decision Science. He is also a Charter Member of
the Institute of Decision Science.
Memberships:
His has served for over 21 years as Administrator for the Cincinnati and Tri-state
Region, The Certified Financial Analysts Examinations, an international organization.
He has been the recipient of the Outstanding Service Award from the College of
Evening and Continuing Education for two years (1993, 1994).
He is very active in the Cincinnati Chapter of the American Red Cross serving on the
First Aid Committee, Queen City Emergency Net (Amateur Radio) and Weather Alert
Amateur Radio Net. He is also a member of the Big Brothers, Big Sisters of
Cincinnati, Inc. for the past 15 years and has participated in the National Ski Patrol
System (Perfect North Slope Ski Patrol, Lawrenceburg, IN) for twenty years retiring
in 2001.

0
#148766

Ken Feldman
Participant

Can’t be the same guy.  He would be toooo busy to be putzing around on this Forum dealing with the issue of the 1.5 shift…don’t ya think???

0
#148767

Ken Feldman
Participant

Here is the real Prof. Paul Swanson.
http://www.nanzan-u.ac.jp/~pswanson/

0
#148768

Keith
Participant

People who are new to this forum should take careful note of the approach taken by six sigma consultants here.  Rather than discuss the content of any attack on six sigma, they attack the person who delivers the information.  This has been repeated dozens of times here.
The reason ? Simply because six sigma is 99% snake oil.  Six sigma consultants know they are conning the public. They can only attack the individual when six sigma is threatened.  They know that it is useless to try to defend six sigma’s basis of bulls**t.

0
#148771

Craig
Participant

Keith,
Is that your best shot? Maybe you can post a more meaningful challenge to six sigma methodology than your snake oil comment. What on earth does that mean anyway?

0
#148772

Keith
Participant

I assume that you believe that you have 3.4 living neurones per million dead ones ?

0
#148775

Anonymous
Participant

How did you arrive at such a stupid conclusion?

0
#148782

Chen
Participant

How do you know the assignable causes weren’t present in the short-term estimate?
Doesn’t your argument have an implicit asssumption – the more runs you have the greater the potential for an assignable cause? Is this a reasonable assumption?
I claim this is only a reasonable assumption after assignable causes have been removed from the short-term estimate, which is the same assumption used with Shewhart Charts. In other words, before estimating process capability one should first establish the ‘real’ process stability. Otherwise every run could exhibit an assignable cause and your assumption would be quite wrong.
If we now ask what happens after establishing a short-term stability, the presence of future assignable causes must be due to a lack of process control. Therefore, we have to ask whether defending against a possible future lack of control using a ‘guardband’ is a good strategy?
I claim it is not a good strategy since it would be better to defend against future assignable causes using process desenstization, rather than through the use of a ‘guardband.’
In other words, the presence of a guardband by itself does not guarantee a robust product or process!
Now that I’ve argued the concept of a shift offers no advantage – what is a disadvantage?
The disadvantage is it detracts people’s attention away from one of the most important principles of process engineering – process desensitization. In other words, Y = F(X) is insufficient knowledge by itself because the derivative of Y is also an important consideration, Y’ = F'(X). Once the derivative is known, a suitable set-point can be chosen to minimise the effect of any variation in X.

0
#148788

Antoine
Participant

Just for information :
DPMO = (Number total of defects * 1 million) /(Number of units * Number of opportunités by units)
In this case, you have :
DPMO = (12 * 1000000) / (150 * 2) = 40000

0
#148789

Ken Feldman
Participant

Wow Dave, an incredibly witty retort.  No wonder you have to post your silly questions on this Forum instead of spending the time figuring it out yourself.  And yes, you have me pegged accurately.  You must be as brilliant as a judge of character as you are at dealing with the most fundamental issues of LSS.  Good luck on your test.  I hope one of the many posters who have helped you will be in attendance as well otherwise you appear to be in trouble.

0
#148802

Craig
Participant

Come on Keithy boy…can’t you do better than that? Do you really think Six Sigma is about 3.4 PPM? Can’t you offer some better challenges to six sigma than that?
Keith’s caution to newcomers:
Rather than discuss the content of any attack on six sigma, they attack the person who delivers the information.  This has been repeated dozens of times here.
Keith’s hypocritical response:
I assume that you believe that you have 3.4 living neurones per million dead ones ?

0
#148804

Airedale
Participant

I have recently done some investigation into DPMO/DPU to Sigma conversions. From what I have found I have to agree some what with the Prof. here. While I will not say some one is quite wrong, I will say that the DPMO to sigma level conversion is a big approximation. In my personal research I found that both DPU and DPMO are both Poisson distributions, though very different ones. With DPU Lamda is a whole number but with DPMO it is a fraction. When lambda approaches 10 (in other words when you quality sucks, at least where I work it would) the Poisson is a close approximation to the normal distribution used for Z. Since Lamda is a fraction for DPMO, and it is often less than .05 then it is actually a close approximation to a Binary distribution. I am not aware of any point where the binary distribution approximates the normal distribution and there for any conversion from DPMO to Z should be approached with much salt. If there is someone out there that knows a viable mathematical conversion then pleas post it. As Six Sigma being 99% snake oil, I would dare say that statement applies to any filed of endeavor, IT, Medicine, Engineering or what ever, the application of technology or theory is ONLY as good as the person applying it.

0
#148805

Markert
Participant

I also agree that SS being 99% snake oil is an exaggeration.
It is closer to 20% snake oil.  Mikel Harry says that six sigma is 80% TQM – the 20% that Harry has added is snake oil.

0
#148822

Airedale
Participant

Phil, OK, I’ll by that. I think of the commercial with the consultant telling people in the the training session “you have to change the Pair-A-Dig-um”.

0
#148823

Mikel
Member

That is so 60’s.

0
#148831

Monk
Participant

Hi Phil,
I have been personally involved in six-sigma as well as lean manufacturing, since Y2000 and found it very useful both at personal as well as professional level.
Can you share with the forum, any experiences that you would like to share? I understand there must be some truth in what you are saying, but it is better to share the facts, instead of just giving opinion.
Monk

0
#148834

Markert
Participant

Monk,
Does this mean that you believe in Harry’s 1.5 shift nonsense and the resulting meaningless DPMO calculations ?

0
#148835

Monk
Participant

Phil,
I have to say that, I am not fully convinced with the explaination for 1.5 shift, as I have not seen the data, based on which this has been concluded.
I have used DPMO calculation in all my improvement projects and found it very useful as a universal tool to quantify any prolem.
What do you have to say?
Monk

0
#148836

Markert
Participant

Have you read Prof Swanson’s explanation of why the 1.5 is rubbish ?

0
#148837

Monk
Participant

0
#148838

Markert
Participant

“The origins of six sigma’s numbers will help people understand how the many error’s have been created. A brief summary follows:

Bill Smith, a Motorola engineer claims that for uncontrolled processes “batch to batch variation can be as much as +/-1.5 sigma off target.” He gives no references or justification for this. In reality there is no limit to how much that uncontrolled processes may vary.
At that time Motorola used Cp=1. Bill Smith suggested “Another way to improve yield is to increase the design specification width.” He broadens specification limits to Cp=2.
Mikel Harry derives +/-1.5 as a theoretical “shift” in the process mean, based on tolerances in stacks of disks. Stacks of disks obviously bear no relation to process.
Harry names his Z shift. The Z shift makes a number of additional errors: his derivation dispenses with time yet he refers to time periods; he claims a “short term” and “long term” yet data for both are taken over the same time period.
Harry realises his error in the 1.5 and says it “is not needed”.
Harry in about 2003 makes a new derivation of 1.5 based on errors in the estimation of sigma from sample standard deviations. For a special case of 30 points, p=.95 he multiplies Chi square factor by 3, subtracts 3 and gets “1.5”. The actual value ranges from 0 to 50+. He calls this a “correction”, not a shift.
Harry’s partner Reigle Stewart adds a new calculation he calls a “dynamic mean off-set.”: 3 / sqrt( n ) where 3 is the value for control limits and n is the subgroup size. For n=4 he gets “1.5”. Reigle says “This means that the classic Xbar chart can only detect a 1.5 sigma shift (or larger) in the process mean when subgroup size is 4″Reigle is quite incorrect. Such data is readily available from ARL (Average Run Length) plots.
Furthermore six sigma tables are based on the assumption of normal distributions.  No data will ever exactly follow a normal distribution to the extreme six sigma tail.
In summary, the 1.5 does not exist, despite the many attempts to prop it up. Calculations involving 1.5 are hence meaningless. That is, 3.4 dpmo is meaningless. Six sigma tables are meaningless. Sigma levels are meaningless.
DO NOT USE SIX SIGMA TABLES.”

0
#148840

Monk
Participant

Phil,
Thanks for sharing the information.
What other metric would you like to suggest instead of ‘DPMO’ ?
So how do we benchmark a characteristic at a universla level? Any thoughts on this !
Monk

0
#148842

Chen
Participant

There are other reasons for not using DPMO.
The DPMO procedure involves counting defects within a single unit. For example, four transistors in a simple circuit. We can neither assume defects are randomly distributed across these four transistors, or assume the four transistors are independent. Anyone who has made substantial yield improvement would soon realise the distribution of apple sizes in an orchard is a different proposition to transistors in a circuit!
In fact this is one of the reasons why Motorola made so much progress in a short space of time – a recognition that most defects are correlated, and mostly occur at the edge of wafers. There is a good reason for this – it depends on how rapidly wafers are extracted from a furnace and the time taken to cool down!!!
If the shift is wrong, and DPMO is wrong – how much else is wrong?
Like the other poster asserted – the 20% contributed by Dr. Harry is wrong!

0
#148844

Participant

Sorry?I  can’t  believe  that,it  is  not  convinving  enough,please  elaborate  more,thanks

0
#148847

Mikel
Member

Monk,
You have been involved in Six Sigma for 6 years. You should not need to see anyone’s data as you have your own. Does the 1.5 match your experience?
What I have seen –
1) For those that really have all the disciplines of  Lean in place, process movement is small < .5 sigma assuming suppliers are on board.
2) For those who have identified their critical  processes and have a valid statistical method for control, process movement is small. Any movement of > 1 sigma would be immediately detected.
3) For those who are not doing 1 and/or 2, process excursions of 1, 2, 3 or more sigma are quite possible. I do work with a few companies who are getting “savings” for material supplied from China. Their batch to batch receipts are in the neighborhood of 5 sigma.
The idea of DPMO is useful only to management and only in a narrow sense. If we can get a number that is not being gamed, DPMO is useful for comparing similar processes, It is better to go learn from people who know how to significantly run a process similar to yours better than you. Why work a project if you can learn more from others?
The idea of building the 1.5 shift permanently into the tables of DPMO to sigma level is pure crap.

0
#148860

Airedale
Participant

Monk,
I would not say that using DPMO to internal measure your process is wrong. I would however say that making the universal goal of 3.4 DPMO the standard to live by because it supposedly equates to 6 sigma and that makes you great if you achieve it, would be wrong. I am actually coming to the conclusion that the best way to measure our process is DPMO. But is also has to be very judiciously implemented. One has to be very careful not to mix opportunities with causes and I am still working that out. However, one also (IMHO) has to be aware that simple process/products will naturally have a lower DPMO than more complex ones. Example, I have found that the DPMO for building parts for our space line is much higher than for lines that make parts for AFVs and the reason being that the tolerance for FOD is much higher for Bradelys than for and Atlas rocket. Thus we can not compare the two lines directly. I like DPMO because (in my mind) it is direct measurement of the process where DPU, yield and first pass yield are inferences of process performance based on the final product. I am looking at it as layers of an onion where yield is the first layer. Any comment on this anyone?

0
#148873

Markert
Participant

Monk,
My apologies … I thought you were another one of the six sigma consultants whose livelihood depended on supporting the nonsense of six sigma tables.
The only really useful metric I’ve seen is MSD – Mean square deviation about target. It gives a measure of how well a process is performing “on target with minimum variance”.
Even this should be always be used in conjunction with control charts and histograms.

0
#148874

Markert
Participant

Counting defects is a poor approach to quality. (Defects other than attributes that is).  Defects relate to the specification, not to the process.
As Bill Smith himself said:  “Another way to improve yield is to increase the design specification width. This influences the quality of product as much as the control of process variation does”.
The only measure of good quality is on target with minimum variance.

0
#148892

Monk
Participant

Phil,
Thanks for the information.
Yes, you are right in saying that my livelihood is dependent on the six-sigma as I consider myself a six-sigma practioner. At the same time, I am aware that nothing is perfect in this world and ‘evolution’ is an on-going process. By that I mean that just like the world is going after ‘six-sigma’….I would not be surprised if the same happens with ‘MSD’.
Anyways, can you share more information and insight into ‘MSD’ and how it works. Pl. share the same on my e-mail address – [email protected].
Monk

0
#148934

Markert
Participant

Monk,
MSD = sigma(X)^2 + (Xbarbar – T)^2
where sigma(X) and Xbarbar come from the control chart for X.
Note that the above is only meaningful for a predictable process (in control). Similarly, Cp is intended to be a measure of how capable a process is of meeting specification.  This can only apply if a process is predictable.  Using single numbers as a measure of how a process is performing may be misleading.

0
#149433

Karimkonda
Participant

defects:12
units 150
no of opps: 2DPO DPMO/PPM Sigma Yield Cpk
0.04 40000 3.25 0.9599409 1.0833
email
[email protected]

0
Viewing 40 posts - 1 through 40 (of 40 total)

The forum ‘General’ is closed to new topics and replies.