The Elsmar Cove Business Systems and Standards Discussion Forums Ppk vs Cpk - A Good, clear explanation and How Mini-Tab Handles Certain Statistics
 Forum User Name Keep Me Logged In Password
 Register Photo Albums Blogs FAQ Registered Visitors Social Groups Calendar Search Today's Posts Mark Forums Read

 Elsmar Cove Forum Visitor Notice(s)

# Ppk vs Cpk - A Good, clear explanation and How Mini-Tab Handles Certain Statistics - Page 5

 Elsmar XML RSS Feed Monitor the Elsmar Forum Sponsor Links Courtesy Quick Links Links Elsmar Cove visitors will find useful in the quest for knowledge and support: International Standards Bodies - World Wide Standards Bodies International Organization for Standardization - ISO Standards and Information Quality Digest

Post Number #33
22nd May 2001, 06:14 PM
 lkickflip99@aol.com Total Posts: n/a

Cpk suffers from three maladies 1. lack of economic consideration in the metric 2. A numerator which has questionable limits (and process for determining them) and 3. The denominator which does not consider off center and mean shift separately.

Consider the following: If the cost of scrapping poor products is the same as the loss incurred by sending the part (or assembly or product to the next ) then Cpk is probably ok. If, however, the loss exceeds the cost, which is mostly the case, then higher Cpk numbers should be required to balance the \$ differential.
my 2 cents.

Post Number #34
30th October 2002, 06:16 PM
 Marc Total Posts: 26,354
Have The Definitions Changed?

I received this interesting e-mail today. I wonder if any of you folks have any comments about the statement that the definitions of Cpk and Ppk have 'flipped'. I am not posting his e-mail address because I'm not sure if he wants this to be public, it being in an e-mail to me rather than a post in the forums where he could explain.
Quote:
 On 10/30/02 at 1:36 PM EST, T wrote: > Marc: > I've read several of the postings about the confusion of Cpk and Ppk. > > First - Who am I? > XXX is the world's largest manufacturer of XXX. > > Many of our employees are receiving training is the six sigma methodology. > Within Six Sigma applications you will discover that the terms Cpk and Ppk > have been changed. It use to be that Cpk meant Long Term Process Capability, > and Ppk was Short Term Process Capability. Six Sigma has reversed these > terms. Why and how this has happened is a long explanation. I just wanted to > mention it since I did not notice reference to this in any of the postings > on the Forum. This may add to some of the confusion people have. > > Second, I wanted to pass along a simple SPC program I developed in Excel. > You may make it available to your readers. > It will calculate Cp, Cpk (R-Bar/d2 method); Pp, Ppk (n-1 method) and PPM. > It can be used with either Bilateral or Unilateral > tolerances. > > Let me know if this is of any help to your readers; > > Best Regards; > T
Now - that said I wrote back:
Quote:
Tim,

I appreciate your e-mail. However, I am not the expert. The experts are those in the forums so I will defer to them to debate.

No - I must admit, I interpret Cpk and Ppk as defined earlier in this thread. I know six sigma enthusiasts claim quite a lot. Six sigma its self has been debated rather severely in the forums (see Six Sigma - Statistical Tools - Valid or Hype? Value? Can a CQE do the same? ) as to whether it's Motion Lotion that works or totally over hyped sales indoctrination.

I must admit that I believe a lot of the Six Sigma is sales. I was privileged to be trusted by the Motorola Semi-Conductor Sector (or what was) with internal documents (the old 'You tell and we shoot you') and my impression was, and is, that this is nothing more than understanding statistics and using data to manage. From one document:
Quote:
 In early 1987 the concept and goal of Six Sigma capability was introduced.
There was nothing really 'secret' in the documents, but rather techniques and theories.

My background was biology with chemistry and anthropology as minors in college. I dealt with processes on the level of atomic / molecular reactions. So when I got into 'product' manufacturing I was amazed at how few companies used statistics to manage their processes. In biology I saw the Cause and Effect thing up close. Root Cause in manufacturing, which few really understand, was part and parcel of a biological system and understanding failure modes at the molecular level. This was particularly helpful to me when I had to work in micro-electronics where the cause of a component failure was often an upstream component 'event' - hard to diagnose. Bottom line is we're talking complex systems.

Cpk and Ppk are essentially unitless numbers used to indicate whether a process is 'in control' or not using distribution, location and probability as a base.

I would very much appreciate your discussion on this in the forums. You state:
Quote:
 Why and how this has happened is a long explanation.
We would all really like to hear the long version. The more specifics the better.

With respect to your Excel spreadsheet offer, I can bet over 500 downloads if you post it as an attachment (Which means you agree that it is acceptable per, as a minimum, the constraints as defined at: http://www.opensource.org/licenses/gpl-license.php ). Or, if it too large to attach to your post, send it to me as an e-mail attachment and I'll put it in the 'Free Files' (Free Fils directory ) for you.

There is no doubt in my mind that your contributions - both your reasoned, detailed explanation of how and why Cpk and Ppk definitions have switched due to Six Sigma - and your gracious offer to share your spreadsheet - will be appreciated by folks around the world!

My Regards and My Thanks!

Marc T. Smith
What say ye, Elsmar Cove Forum Folks?
Post Number #35
31st October 2002, 03:29 AM
 Ravi Khare Total Posts: 70
Yes, the terminologies of 'Long Term' and 'Short Term' have flipped somewhere along the way.

Cpk could be called as 'long term' due to its predictive nature, since d2 used to predict std deviation is an imperical constant.

Cpk could also be called 'short term' since we evaluate the std. deviation on the basis of limited range values selected over a specific horizon. We also do not take cognizance of long term drifts, and supress the effects of outliers by averaging the range values.

Your definition depends on which view you take.

In fact the Long/Short terms cause so much confusion that I have stopped using these terminologies . I simply refer to them as Cp/Cpk and Pp/Ppk. The formulae for these still remain the same, and lead to no ambiguity.
Post Number #36
31st October 2002, 06:43 AM
 lee01 Total Posts: 89
This is all good statistical fun at its best! (err, hmmm, well. . . . )

The way I have always used ppk/ cpk is this:
Cpk determines your potential process capability

Therefore, in my opinion, when trying to determine the performance of say a manufacturing process, I would and do use Ppk. But when trying to determine the potential capability of a manufacturing process I would/ do use Cpk.

But, and it’s a very BIG BUT! The hierarchical authority within businesses that I have worked in seems only to grasp the very basic of ‘idea’ of Cpk! And the idea they consider Cpk is incorrect anyway! Most people see Cpk as Ppk but no nothing of Ppk?

I’m all in a muddle now with p-p-p-k-c-c-k-p-c-k-p coming out of all orifices!

Lee01
Post Number #37
31st October 2002, 09:59 AM
 Craig H. Total Posts: 2,048
Hi all

I just stumbled on this thread, and have read (and learned)with great interest. There is one thing missing, I think, though I have been wrong before. Don't these metrics "assume" normality? If so, than it seems to me that the use of z or t to estimate the % nonconforming would be somewhat dangerous unless very large, very near normal sample sizes are used. Right??

I would be very interested if the mystery emailer would share more of their story.

The only regret I have about these forums is that it took me so long to find them...Thanks Marc
Post Number #38
31st October 2002, 01:16 PM
 Ravi Khare Total Posts: 70
Well said Craig! Step no further unless Normality is confirmed.
Post Number #39
1st November 2002, 11:53 AM
 Tim K Total Posts: 7
Hi:
This is my first time posting anything. I just found the site. It looks like a fantastic forum for all kinds of Quality related issues.

Yes, the terms have been reversed. The explanation I have is: Although the big 3 developed the Statistical Guidelines as published by AIAG, Chrysler actually never agreed with the terminology. Dr. Mikel Barry (of Motorola fame) used the Chrysler definitions; Cpk = short term, Ppk = long term. Dr. Barry was/is involved in the Six Sigma Academy and ASQ. They adopted the same use of the terms. MiniTab software used this same interpretation. Now we are all stuck trying to make sense out of it.

To add further to the problem, in Ford’s specific guidelines for ISO/TS 16949, they refer to Cpk as “the Historical Process Capability” ( Pg 72). See their guidelines posted at: www.ioab.org

As we all know, the calculations using R-bar/d2 and n-1 are used to develop the Cpk and Ppk indices, respectively. To use these, the process must be stable and in control. To use Cpk the samples within the sub-group must be taken sequentially and each sub-group taken consecutively over time. Since Ppk is based on the population, the samples do not need to sequential. Cpk averages the effects of variation, Ppk does not. These are simple but important concepts. The factor that must be understood is time. Was the data gathered in a short period? I fully agree with earlier postings that suggest Ppk be used when reporting short term data, and Cpk is used when reporting long term data. Our company suggests that when reporting process capability, it should be stated that the index is the result of a short term or long term study.

This all becomes very important when reporting capability after several set-ups. Example: a process is set-up 3 times: 1 – at the mean, 2 – slightly below mean (but capable), 3 – slightly above mean (but capable). If Ppk is used, it would be an average of the entire spread of each set-up. Not Good. Cpk would be an average of each sub-group and would provide a better picture of the process. ( Of course short run SPC methods would be better)

Also, in response to requests for help calculating PPM, I have attached (hopefully I succeed) an Excel program I developed for capability studies. It can be used for bilateral or unilateral specifications; and with sub-groups of 2, 3, 4, or 5. I removed our corporate name and logo; you can add your own.

I have revised the attachment - see below tim

Last edited by Tim K; 6th November 2002 at 02:01 PM.
Post Number #40
2nd November 2002, 04:31 AM
 Marc Total Posts: 26,354

Welcome and thanks for the input and for the spreadsheet contribution.

Cp and Pp have come up a lot recently where I am working. I have a twist to this, however.

Different companies have different volumes. As an example, let's say at a company a good run is 1200 parts over 36 hours. Other runs may be as high as 10,000, but some are as low as 100 over 24 hours. One part is simple and run at rate is 1800 and hour.

My question is: How does volume affect using these different indices?

 The Elsmar Cove Business Systems and Standards Discussion Forums Ppk vs Cpk - A Good, clear explanation and How Mini-Tab Handles Certain Statistics

 Bookmarks

 Visitors Currently Viewing this Thread: 1 (0 Registered Visitors (Members) and 1 Unregistered Guest Visitors)

 Forum Posting Settings You may not post new threads You may not post replies You may not post attachments You may not edit your posts BB code is On Emoticons are On [IMG] code is On HTML code is Off Forum Rules

 Similar Discussion Threads Discussion Thread Title Thread Starter Forum Replies Last Post or Poll Vote saint77 Environmental Related Standards 11 5th March 2011 02:15 AM Pythagore Japan Medical Device Regulations 8 21st September 2010 02:23 PM amanSGS Six Sigma 114 24th June 2009 07:38 AM gard2372 Statistical Analysis Tools, Techniques and SPC 1 20th February 2005 10:17 PM Charmed News! 14 11th September 2004 06:53 AM

The time now is 12:00 AM. All times are GMT -4.