MONDAY, JULY 28, 2014
Font Size
Topic DPU Vs DPMO

DPU Vs DPMO

Home Forums Old Forums General DPU Vs DPMO

This topic contains 68 replies, has 1 voice, and was last updated by Avatar of Darth Darth 10 years, 7 months ago.

Viewing 69 posts - 1 through 69 (of 69 total)
  • Author
    Posts
  • #50764

    Normally DPMO is used in the BPO industry to calculate sigma values. I am facing a challenge and I see subjectivity when we move from DPU(One opportunity ) to DPMO ( multi opportunity ).
    At times there are instances when our guys feed in 5 fields and at times they feed in 60 fields. Calculation of six sigma is purely based on the types of request we get.(Fields vary based on the request and they cannot be segmented as there are many requests).
    Is it that we have to capture sigma values requestwise ??
    Suggestions please ??
    Jai

    #50775

    Here’s my take on this:
    What’s most important here is to determine what your most common and significant defects are and what inputs are contributing to them.  If you use more than one opportunity when calculating the sigma level of your process, it’ll be higher than if you consider your unit one opportunity (how much higher depends on your number of opportunities).  Manipulating the numbers to get a higher sigma level isn’t improving the process. 
    Just based on what you said it sounds like you should go with DPU because your opportunities fluctuate.  If you do go with DPU your sigma level will probably be negative.  But that’s okay in a situation like this because it’s more important to make sure that your data is an accurate depiction of your process.  Whatever your sigma level is now  – drive it up from there!
    Good luck! 

    #50778

    Thanks for that. If a process follows DPU at times it results in killing itself and if we need to move to DPMO – issues on manipulating numbers come in. Again we if go by analysing the defects – it is subjective and may be we need to put a pareto and deceide on it. But the excerise would be cumbersome when the number of fields to be updated is 80 to 100. Capturing information field wise is challenging and may require more resources. Let me put up a business case and take it up..
    Jai

    #51173
    Avatar of Eleodor Sotropa
    Eleodor Sotropa
    Participant
    Reputation - 0
    Rank - Aluminum

    We use the following process for calculating the Six Sigma value:
    1. Identify the critical fields for input. Typically these are fields that have an impact on the customer (upstream process customer), or the data integrity for these fields ensure that the end-to-end process meets customer requirements.
    2. Use this number as the opportunities for defects, and in calculating the 6S metric.For example, in Accounts Payable Invoice Processing, the Purchase order based invoices will have far less opportunities for error than a non-PO based invoice, as most of the data for a PO based invoice has already been setup during the PO creation – currency, account, expense code, cost center, tax. For the non-PO based invoice, you can identify all these fields as opportunities for error – amount, currency, vendor, approver, expense code, cost center etc.
    From my experience, the PO based invoice will have 5 to 6 opportunities for eror, while a non-PO based will have 9-10. The only way to compare the quality for these processes is via DPMO.Eleodor

    #51181

    I agree with all that you say except it is not opportunities for error. It is the opportunity to do correct value added things. There is a possibility for more things to go wrong than the things that need to go right. You don’t get credit for how many ways we can mess things up.

    #51190

    Semantics….
    A characteristic can fail in many ways once it fails it is a defect.  For each defect there are many possible ways to fail but it is still a defect. If you count the number of characteristics, or solder joints, or components..it doesn’t matter, what matters is consistancy. Pick a method document the method be consistant..that is what matters not splitting hairs over ways to count it.

    #51203

    Wrong.

    #51206

    Here we go again.
    First of all, if a single step in a process can be counted unambiguously, then it can be considered an opportunity. If it can fail 2, 5, or 50 ways, it still is one opportunity.
    If counting opportunities is a hassle, don’t do it. You have less need to count opportunities than you do defects. In fact, counting opportunities might well distract you from the ultimate goal of finding out how to prevent the defects from occurring.
    The only thing we need the denominator term to be is valid and repeatable. Staying with units often achieves this just fine.

    #51209

    Ron,
    The difference between opportunities to do it correctly and brainstorming everypossible way you can screw something up is not splitting hairs. The count difference is a function of a persons immagination – that is a long way from rational.
    Your advice to stay consistent is good as long as you aren’t staying consistently screwed up.

    #51210

    I am not sure why Stan does not agree with Ron.  In my opinion, it is more important to drive improvement than to go “by the book” on counting opportunities.  Use what makes the most economic sense.
    Stan–can you elaborate more on why you disagree.
    I have worked for a company that have tried to count EVERY opportunity.  I felt like they were spending more time counting opportunities than working to improve their process.
    My recommendation is that if you have hundreds or thousands of opportunities per unit, then you should use the Pareto principal and reduce the scope of your project.  For example count the number of opportunities for the problem operations.
    As I stated above, use what makes the most economic sense.  Suppose that a drill breaks and makes five holes oversized resulting in scrap.  If it is five parts and the Machinist had a chance to stop it after the first part, then you will want to see this.  If it was a single part where the Machinist could not catch it before all five holes were drilled, then it can potentially mask bigger issues.  Ask the production manager if scrapping five parts should count the same as scrapping one part with five oversized holes caused by a broken drill.  In this application, I think it makes sense to use DPU. 
    However, if there is a reasonable chance that each hole is a non-conforming for different reasons: depth, diameter, location, then DPMO is more relevant.  Also if the cost to rework five parts is the same as five features then DPMO makes sense.
    Again, the most important part of the metric is to drive true process improvement.
    Tom

    #51214

    TomF,
    This is actually pretty simple. If I am soldering a chip resistor in place I get 1 count for getting the correct part and 1 count for each end of the resistor because there is one opportunity to make a solder joint correctly. Total count of 3.
    The other method says I get one count for the part. Now I get to let my immagination run wild. I can have a blow hole, pinhole, pit, void, insufficent solder, excess solder, de-wetting, non-wetting, grainy and frosty. Times 2 (like Lupienski negotiating the price of jogging suits in Taiwan). That is the limit of how many ways I can conjure up screwing up a solder joint without the help of some Rum induced fog. I have two of them. Total count 21.
    That is just a simple example and I am not all that creative.
    This is like the next generation active and passive hallucination that we had to endure about 8 years ago. It is like Six Sigma gout that flairs up everytime someones bonus is in jeopardy and they have been dis-engaged (PC word for screwing off) all year.
    Lets see the “denominator shuffle” (you remember that dance with Jim McMahon – we ain’t here to cause you no trouble, we’re just here to do the denominator shuffel or something similar) is back on the top ten list with a bullet - it must be December,
    Good luck.

    #51221

    Sometimes two person disagree because they are seeing one thing at different side. It’s like a cylindrical can, if you look at it at the top, it’s a circle and if you look at it at the side it’s a rectangle. It depends on what view we are looking at or what question we want to answer.
    In the case of this discussion, we want to know how to count the opportunity, the correct way ( by the book ) is Stan’s definition in which during improvement would do his best to have a process that will deliver the best result ( in mike’s example of resistor = best soldering result ). But you cannot achieve this ” best soldering result” if you don’t list all possible failure modes and do your best to avoid these failure modes or “opportunities for defects .”
    I think you would agree with me that the best way to “perfect” something is to avoid all possible ways of impurities or defects. This is where the other side of the cylindrical can comes out. Though we should not equate it to opportunities, it is even more important than the metrics because metrics will not improve any process, we can only see the status of a process while defects reduction or even defects elimination would surely improve any process.
    So, in summary, If we want to count the Opportunity we use Stan’s definition to be used for metrics ( top view of the can ) and use Ron’s defects counting for actual process improvement ( side view of the can ).
    I’m trying to mediate between the two, I hope this would help rather than confusing you.
    AbetF

    #51222

    AbetF–I am confused.  Perhaps because it is late here. 
    AbetF says “( in mike’s example of resistor = best soldering result ). But you cannot achieve this ” best soldering result” if you don’t list all possible failure modes and do your best to avoid these failure modes or “opportunities for defects .”  However, I read Mike’s example as counting only one defect per solder join–not any opportunity for each distinct failure mode (pin hole, void, etc).
    Mike–you’re right.  You can tell its December.  I think there is a misunderstanding.  My post was not about how to count opportunities.  My post was about whether DPMO or DPU is better.  I believe that it depends upon the application.  I agree that the application you used as an example, DPMO may be the best answer.  However, there may be other applications where DPU is the best answer. 
    Respectfully,  Tom

    #51226

    TomF,
    I was reading all the mails on this subject ( DPU vs DPMO ) from Sotropa, Stan, Ron, Jonathon, and Mike. They always disagree on what Opportunity is, especially Ron and Stan. I didn’t notice that I was reading your mail when I replied.
    You’re right it’s already December and we’re thinking already of holidays and parties.
    Merry Christmas,
    AbetF

    #51248

    From yours and other posts, it seems to be an agreement in that the importance of how “opportunities” is defined is relative, as long as this definition is clear, allways the same, documented, as free of subjrctiveness as possible, etc. As you said “…it doesn’t matter, what matters is consistancy”.
    Then, What is the added value of counting opportunities? (would somebody be so kind to answer that question?) Why on earth would we bother about “opportunities” at all in the first place?
    For the sake of consistency, use the following definition:
    Opportunities: One
    The thing is either right or wrong! I don’t see that clear the definition of “opoertunities of doing it right” either. To have a good solder joint you need (I guess) a good positioning of the elements, the right temperature, the right soldering elemnt (Sn), not to mention that you must be joining the right parts that must also be good parts, etc, etc, etc… At the end there is only ONE opportunity to do it right. Everything must be done right the first time. It does not matter if you fail in the selection of the component, in one of the solder joints, etc. You spoil one of them, and you have screwed up it all.
    Opportunities=1. And don’t tell me it is not repeatable.
    Reduce the number of defects, and you will be improving. Note that a reduction of the number of defects that is achived by the reduction in the complexity of the product or process would not be taken as an improvement from the DPMO perspective (unless you use a fixed number, such as 1). And it is the best of the improvements: The best component is the one that does not exist. The best operation is the one that is not realized. You can never screw them up, so it is the best poka-yoke.
    (I admit my lack of experience and knowledge in the field of opportunities counting, so you can take this as an outsider’s view)

    #51258

    TomF,
    If you read your last response it was about opportunities and economic value – which has no bearing on it. It is as simple as “am I trying to do it correctly?” or do I want to see how many ways I can screw this up.
    The posts are indented so you know which post is being responded to.
    You will have trouble using DPMO (back to your original post) without deciding how to count opportunities – that is what the O is. It is like any other metric – why use it. The O portion is only useful for complexity. When you are working a project where is the advantage in adding a complexity factor? You are not comparing it with anything else – which is why you need complexity (comparing a Jet engine at 1 dpu to a light bulb at 1 dpu doesn’t make any sense – but using DPMO does – that is a management type perspective). dpu is a very straightforward calculation that is easy for everyone on a team to understand. The minute you throw opportunities into the equation it adds confusion and is its own complexity factor. What advantage do you gain from DPMO?
    Good luck.

    #51270

    “comparing a Jet engine at 1 dpu to a light bulb at 1 dpu doesn’t make any sense – but using DPMO does”
    In which wahy does it make sense to compare a jet engine with a light bulb using DPMO?
    Don’t know you, but I would feel much more comfortable using a 1000 DPMO light bulb that flying a 0.1 DPMO airliner.

    #51274

    Consider our product, which is a radiator.
    We make several different sizes (ie number of channels) for different applications. From a customer perspective (and from ours too, since we don’t ship leaking parts) it doesn’t really matter if a given radiator leaks only from one place in one channel, from many places in the same channel or from one place in each channel…it still leaks. We consider this to be a single opportunity, even though there are in fact multiple places the leak COULD be on a single unit. Whether or not there are multiple leaks in a given unit is of no importance to us for a first pass. It does allow us to have the simple, in-your-face kind of numbers that everyone can understand.
    On the other hand, when we are working on process improvement, it helps us to understand the problem if we begin to stratify and determine leak frequency by location within the channel, product family, number of channels per radiator etc. and at this stage we often move to DPMO to aid our analysis.
    We currently have over 300 potential defects (including assembly types) that a single unit might experience. We could delude ourselves into thinking we have a 6 Sigma process for each of these defect types, especially if we used DPMO (by channel quantity for example), yet common sense tells us that if 5% of our product is going into the red bucket we can hardly claim to have a world class process.

    #51275

    You are talking about comparing apples to oranges (or light bulbs to jet engines). 
    Consider that a typical NBA players makes between 90% to 70% of their free throws.  A NFL quarterback completes between 70% to 40% of their passes.  And the best major league baseball players gets a hit approaching about 40% of the time.
    By these numbers, we can clearly conclude the NBA players are more athletic than NFL quarterbacks, who in turn are better athletes than major league baseball players.  Obviously this is garbage!!!  They are completely different process having differing rates of success.  You need to know the variation and trending of each process.
    This is a clear abuse of metrics.  It is a big cause of the demoninator shuffle, especially in December when bonus time rolls around. 
    Tom

    #51277

    Let me clarify a few things by putting the concepts in bullets:
     

    There is a famous quote by George Box:  “No model is right, some are useful.”  There are disadvantages in using both DPMO and DPU.  You need to be smart in choosing which one to use.  Consider which of these metrics will help you make the greater amount of economic gain.
    A disadvantage of using DPMO is that you can create bureaucracy in collecting the opportunities across an entire company or department.  You are better off using DPMO on smaller scale projects.
    Another disadvantage of using DPMO is that you can have one cause that is replicated numerous times without a reasonable chance to remedy the situation (this is my broken drill during CNC program example).  In this situation, you will be swamping the signal for significant improvement with noise.  Note:  I understand that there will be people that can debate that the Machinist could have heard the drill break and had a chance to fix the cause before it was replicated, but I consider that not to be a reasonable chance to remedy the situation until after the CNC program is finished.
    A disadvantage of using DPU is when you have multiple causes affecting one part.  You will want to identify and fix all of these multiple causes.  Using Mike’s solder example (assuming that this is a manual solder and not wave soldering), I believe that the operator has a chance to remedy the situation after placing the resistor and making each solder joint.   In this situation, using DPU will understate the potential process improvements (resistor placement and repeating a bad solder joint when you could have fixed it after the first solder joint was made).
    Lastly, I do not believe that Ron was advocating counting a characteristic multiple times due to different possible defects.  He says “count the number of characteristics, or solder joints, or components.”  I believe he is being somewhat glib in saying “it doesn’t matter” in how you count opportunities.
     
    Hope this clears things up and helps.
     Respectfully, Tom

    #51281

    TomF,
    The point was that once you include a complexity factor you can compare apples and oranges. This may come as a shock to you but that is the value of the sigma calculation.
    As far as not making any sense to compare them – it may not make any sense to compare them at the project level which is why the use of DPMO presents no advantage to the project level. When you are a CEO and your company makes light bulbs and jet engines (this is GE by the way) and you are looking for that common metric that lets you gage the performance level of your organization. You use the metric because it is your job to find a way to comapre apples to oranges.
    So if I had to work on your project with football, baseball, and basketball players I would work the project with each one at a dpu level. If I owned all 3 teams and I was setting up budgets based on who needed the most work – I would use DPMO.
    As far as your conclusion that the metric makes them “more athletic.” That would be a stupid conclusion because there isn’t a relationship between free throws, passes, hitting and athleticism.
    Let me quote Stan “wrong.”

    #51282

    aj,
    We worked the exact same process in Mexicali except on Charge Air Coolers wich is just a giant radiator. We ran it on dpu just like we always do. Explain the advantage you have to dpmo?
    As far as working it by location dpmo doesn’t have anything to do with that. You map your defects – that is location.

    #51283

    Gabriel,
    If you read the other post you will see who needs to compare the Jet engine to the light bulb. Actually if you are comparing dissimilar process you should understand this as well.
    In you last sentence you have confused risk with defect level. Two separate issues.
    Good luck.

    #51288

    Mike,This may come as a shock to you but there is as much of a relationship between free throws, passing, hitting and athleticism as there is a relationship between light bulbs and jet engines.  Namely, professional sport teams may share the same name.  The plants making the light bulbs and jet engines share the same name.Tom

    #51290

    This may come as a shock to you but just because CEO’s compare apples to oranges that does not make it a valid use of metrics.
    Indeed, W. Edwards Deming talks about this misuse of metrics in two of his seven deadly diseases. 
    Evaluation of performance, merit rating, and annul review.  To quote Mary Walton’s excellent book “The Deming Management Method”: 
    Performance evaluations encourage short-term performance at the expense of long-term planning.  They discourage risk-taking, build fear, undermine teamwork, and pit people against each other for the same rewards….The result is a compony composed of prima donnas, of sparring fiefdoms.”
    “Running a company of visible figures alone.  The most important figures are unknown and unknowable–the multiplier effect of a happy customer, for example.”
    Tom

    #51291

    TomF,
    You have thrown up a lot of smoke and wraped yourself in selective interpretation of Demings principles which means you have no more understanding of Deming than you do the basic underlying properties of Six Sigma.
    One last time. Oportunities for project work are irrelevant. dpu works becuse it is a simple ration of the number of defects to the number of units produced.
    DPMO can be used when comparing dissimilar products because the opportunity count levels the playing field in terms of complexity.
    That is a fact. You can pull quotes all day long and the math is the math.

    #51320
    Avatar of Andy Urquhart
    Andy Urquhart
    Participant
    Reputation - 0
    Rank - Aluminum

    Gabriel,
    You are quite right; it is bad enough having to count defects, let alone having to count ‘opportunties’ as well. More seriously, how do we know whether opportunities are random or not. This can be an important when considering causes: in the same way that it is important in control chart theory, where ‘special causes’ tend to be random and localised and ‘common causes’ tend to be systematic and non-localised.
    In the case of PCB manufacture, the solderability of a Via during a wave solder process can depend on the amount of heat ‘sunk’ by a track, electrical shield, or a component. How do we know if this is an opportunity or not? Is it a random or non random defect? What is the fix? My contention is that this is a systematic defect and that the layout is the true cause. Perhaps it ought to have been caught in a design rule check. I hope you agree.
    Anyway, I know from previous work that I can calculate both DPU and DPMO from yield results – a procedure known as spatial yield analysis; so I would always advocate doing some electrical testing on a test vehicle to ‘calibrate’ any defect or opportunity counting.
    Andy

    #51324

    “In you last sentence you have confused risk with defect level”
    I know. It was intended. Comparing in “defects” in different contexts is nonsese, because not all defects have the same weight in terms of the impact on customer satisfaction, cost, safety, ethics, public image, etc.

    #51326

    Gabriel,
    It may apper nonsense in the context of what you are doing but understanding performance across a variety of businesses, processes, etc and making decisions about resources has to be someones responsibility. People have spent years perfecting the art of hiding bad performance behind the excuse “You are comparing apples to oranges.” The sigma value has taken that away from them.
    If you build jet engines or medical equipment you live in a 6 sigma world. There are two issues 1. what do you build and 2 what do you deliver. If you don’t build SS but you deliver SS then the only way you get there is by inspection and test. Therefor you have opportunity.
    If you build light bulbs you probably live in a 3.5-4 sigma world. If you build 3 sigma you have less opportunity.
    Nobody said that the DPMO was the only criteria but a first pass is I can compare the two. It is not nonsense. It is how you determine where and how much you deploy resources. That is why you use DPMO in the strategic operations and keep life simple at the operations level – reducing complexity where it isn’t needed is as basic to process improvement as addition and subtraction. There is absolutely no advantage to a project by using DPMO. dpu will do everything I need it to do in terms of evaluating any improvement I make. The calculation of a ratio of defects to the number produced is something most can do with a pencil and piece of paper – it is a clear metric people understand. I don’t add the controversy of what is and is not an opportunity and I don’t lose any team members who can spin off at the sight of some number that isn’t relative to their world – comments like “we didn’t build a million of those” etc. You don’t add complexity where it isn’t needed.
    Good luck.

    #51333

    Mike,
    First, I must say that I do not appreciate the demeaning manner in which you reply to others’ posts.  I would hope that people using this site would respect each others opinions and offer constructive comments.
    If you disagree with my interpretation of the Deming quotes, please let me know why you disagree with them.  Summarily dismissing my interpretation is not constructive.
    As far as comparing DPMO for dissimiliar products, the math is the math.  It is the interpretation that is wrong.
    If a NBA player’s free throw percentage is 80%, a NFL quarterback’s completion rate is 50% and a Major League Baseball players batting average is 30%, you cannot compare them to against each other as far as performance is concerned.  Note that each of these is a DPMO metric.
    Each is a unique process with its own natural variation.  If you do not understand the natural variation of your processes, then you will be mislead into a futile process improvement efforts.
    Tom
     

    #51335

    Tom,
    If you don’t understand that dpu and dpmo are performance type metrics and not the statistics you use to perform analysis you really need to do some reading before you get into a string like this.
    Good luck. 

    #51336

    Mike,
    I had challenged you to give me a constructive counter-arguement to my interpretation of Deming’s quotes.  You have responded with an insult.  My challenge for you to reply with something constructive still stands. 
    By the way, the issue is not whether DPU and DPMO are performance type metrics.
    Tom

    #51338

    Tom,
    You probably wouldn’t be so easily insulted if had any clue what you talking about. You pulled in two points from the 7 Deadly diseases because thy seemed to justify some inane opinion. The first was on merit reviews which have never been part of the discussion. The second was about running the company on visible figures alone which was never part of the discussion. That is interpretation for convenience. You wrap the Deming philosophy around yourself assuming that everyone will immediately stand up and salute regardless of the lack of applicability.
    Point 10 of Demings 14 points addresses his position on metrics:
    10. Eliminate slogans, exhortations, and targets for the work force that ask for zero defects and new levels of productivity.
    In this point, Deming attacks, without naming it, Crosby’s Zero Defects Program. Crosby stressed the role of worker attitude as being critical to a quality effort. As such, Crosby suggested workers be given quality targets of zero defects. Once a year, Crosby advised there be a “Quality Day” reminding everyone of the importance of quality efforts.
    Deming believed that poor worker attitudes are symptoms of supervisor inability to lead. He stressed again and again that it is the system which produces errors, not people. As such, it isn’t the work force that needs attention, it’s the controllers of the system, management.
    Slogans like “Produce zero defects” and “Do it right the first time” are quite common. But Deming stressed, they are also quite meaningless. At best, they are ignored. At worst, they infuriate people who understand the system causes errors not workers.
    I pulled this off the Endsoftheearth.com website. Deming was opposed to setting goals (particularly idiotic ones such as Zero Defects) without having a methodology to achieving those goals. Floor mats and posters didn’t do it.
    What is next?

    #51342

    Hi Mike,
    Still waiting for you to call be back…
    Patrick

    #51343

    Tom,
    Mike only seems to be demeaning when responding to you, so I don’t understand the “others” you are talking about. Are you one of those that talk about yourself in the third person?
    Your understanding of Deming does seem a little weak. Why don’t you take some time to read and reflect on Deming’s writings, not Mary Walton’s, and come back here – say in about 5 years.
    If you think you can absorb Deming or Six Sigma in a short amount of time, you are wrong. Your idea of defect opportunities lacks that understanding.

    #51345

    Mike:
    Please note that I agree with you about all the reasons why not to use DPMO in some contexcts (process improvement). What I don’t get (probably my fault) is the reason why DPMO (or a sigma metric based on this metric) is of any value in any other context, such as strategic decisions about resources and the like.
    Let’s use the definition of “opportunities” as “parts + added value operations”. A jet engine has hundreds of vanes, each vane has several value added operations where a mistake can lead to a broken vane, and also a munting operation. let’s say you have 5 opportunities per vane, you have thousands of opportunities per engine in vanes alone. Just to put a number, let’s say that there are 10,000 opportunities of doing it right that, if not, the engine will be lost (not just any CTQ, I am speaking of a lost engine, and probably during the take of that is where the engine is under maximum stress). Multiply it by 4 in a Boing 747 (4 engines). And 1 out of 25 Jumbos would have an engine that WILL be eventually lost. (well, with this level of DPU may be it is a little less than 25 because some Jumbos will have more than one thefect, either in the same or in different engines).
    Now, what do you want to compare “this” six sigma level with?
    There was a looooooong thread (the longest I’ve ever seen in this or any other forum) that you missed (probly in one of those trips). Click here and here to see a couple of messages from myself (the last one includes a DOE like comparison of how the sigma level is affected by the shift and the opportunities count). If you want, you can read the rest of the thread. It is long, but I think you will enjoy the messages from Reigle. ;-)

    #51346

    Patch,
    Sorry about that. trying to get a house closed before the end of the year.
    I get you tomorrow.

    #51347

    Gabriel,
    I am not sure how to convey it without just rehashing the same things we have been over time and time again on this thread. It is not about engineering. It is about management. Whether you or anyone else likes it or not there is someone who makes decision about various types of resource allocation on a daily basis. They don’t care about opportunity counts on vanes and they don’t care about the black ends on a flourescent light bulb. They care about placing the correct resources in the correct place. That means you either find a common metric to use to help you understand your business or else you listen to a load of crap about fruit (apples and oranges). Once you have a measure of complexity – you level to playing field and it gives you a measure that lets them understand how much of a change needs to be made and where that change resides (shifting from 1 sigma to 3 is much different than shifting from 4 sigma to 5).
    That is the type of measure that a C level person makes in a large corporation but it shouldn’t be that much of a foreign concept at plant level. If you run multiple products inside a factory – how do you make decisions about the levels of quality between different products? If someone isn’t making those decisions then you have to many resources available (which I rarely see) or it is a political battle which is completely stupid to manage a SS program with out using data.
    Regards,
    Mike
     

    #51348

    Stan,
    Thanks for the help with the TomF.
    The upside is I hadn’t read throught the 14 points in years.

    #51425

    I am not a manger, but what about this metric?
    $$$
    Eli says:
    Tell me a change in this process is an opportunity to make a delta of X$ in the throughput of the whole organization, Y$ in the inventory of the whole organization, and of Z$ in the operating expenses of the whole organization. These are the metrics. These are universal. These are the ones that should lead your resorces allocation.
    And “sigma value” and DPMO are not linked with $$$.
    As allways, that’s just my opinion, and I can be wrong. (I probably am in this case, because it is a opinion based on “feelings”, not data or actual experience)

    #51448

    Gabriel,
    And “sigma value” and DPMO are not linked with $$$.
    Who ever told you that? There is a cost associated with defects. It may be indigenous but as a manager I would hope you realize that. Where did you think the dollar figures came from?
    As far as speaking in dollars – it is always a good idea to be able to put you issues in dollars because management has made sure that message was clear. What language do the LEADERS speak? 
    Let’s take a different approach. When I go South America I try to speak Spanish as much as I can (it is poor on a good day but I am in a Spanish speaking country – at the very least it is respectful). Graciously they speak English – which is generally much better than my Spanish. How do you feel when someone shows up in your country and makes no effort to speak Spanish – they expect you to speak their language? Great foundation for a relationship.
    Managements language may be dollars but that does not exhonorate them from learning them from understanding the technical issues. If you don’t have two people in the conversation all that is left is Cool Hand Luke.
    Just my opinion.
    Merry Christmas.

    #51460

    What is “Cool Hand Luke”?
    “There is a cost associated with defects.” I could not agree more on this. But not all defects have the same impact on cost. It depends on the type of product, the type of defect, and how far did it go until it was detected. For PPM and DPU a defect is just a defect. For DPMO you add even more uncertainty with the opportunities (a given defect can be as costly in a complex system as in a simple one, but the DPMO will be different even if the opportunities are counted in the right way). Add a sigma shift to make things even less clear, and you have the sigma level.
    If there is any link between DPMO or Sigma level and $$$, I don’t see it. And don’t tell me that, for comparable process, products and defects, that link exist. Because I already had that link (and much more clear) with the PPM or DPU alone. Remember we were talking about the value of DPMO and sigma level as a sort of “universal” metric that managers use to prioritize the allocation of resources for improvement among different processes / products.
    Again, just an opinion based on “feelings”.

    #51461

    What is “Cool Hand Luke”?
    Gabriel,
    Cool Hand Luke is a cultural thing here in EEUU.  There was a famous movie in which a director of a prison had problems with at least one ‘unruly’ prisoner.  He said “What we have here is a failure to communicate.”  This movie became very popular and entered our ‘pop’ culture.  The ‘Cool Hand Luke’ effect has come to mean that people aren’t communicating effectively, often times the problem is that people might being speaking different languages (six sigma people speaking in six sigma terms and managers speaking in financial terms or some people speaking English and others Spanish).
    Ojala que te ayude.
    Saludos,
    Faceman

    #51463

    Gabriel,
    Cool hand Luke is a movie with Paul Newman. It is where the line “What we got here is a failure to communicate” came from.
    My post spoke about indigenous cost which should indicate that it is not a universal measurement. If you go back far enough in the posts you will find a response that talks about understanding what level you are trying to from and to. There is also a discussion around being a particular level in a particular industry and being a particular level and shipping a particular level. If you believe (or you believe that I believe) that there is this mindless stand alone metric that people can use so that they can abdicate decision making responsibility that does not exist. That is not what management is about. It is about being able to put together a story from data and make decisions that are the best decision for the overall business. The differences are not generally so tight that absolute precision is required. Calculate the rty on a jet engine and decide what happens if you put that in the field – then you understand the level of test and inspection necessary to stay in the business – the difference is a couple sigma. How close does that relationship have to be to make a decision?
    Sigma and DPMO are the metrics that are a piece of the management picture. A CEO of even a midsize company thay is managing at the dpu level is managing several levels below their responsibility and probably not doing the job of CEO effectively. It is no different than a CEO that managed the business using EBIT or EBITDA.
    Just my opinion.

    #51465

    Faceman,
    Thanks.
    Great explanation.
    Merry Christmas.

    #51467

    Mike, I am really trying to get your point, but I can’t. I am not saying that you are wrong, only that I don’t understand. I think an example may help.
    Would you tell me (may be invent for me) an example of a decision making case where DPMO or sigma level is used as a part of the picture with significant specific weight in the decision that is being taken, and were DPU or PPM would not have been so appropiate for a decision in that level?
    I would really appreciate that.

    #51479

    Hello Gabriel:
    It appears that question is about DPU vs. DPMO, or even why DPMO? Let me take a crack at it.
    DPU is a process output that is a discrete measure of the deliverable. While DPMO is a measure of complexity of the process that produces the DPMO. For example, in a same manufacturing factory, one may manufacture cell phone as well as the base station. DPU of both units may not be a true representation of their performance. Instead, the DPMO will represent the reproducibility the units better. In other words, the defect rate is directly proportional to the opportunity for making mistakes.
    In order to count opportunities, one must look into the items as well as related actions that go into the process. However, one must be careful not to go microscopic in counting opportunities beyond the practical level.
    I hope it helps.
    Praveen

    #51494

    Sorry Praveen. I still don’t get it (my fault).
    Lets say that the cell phone has several times the opportunities of the base unit. Let’s also say that I have more defects in the cell phone line, but because of the more opportunities, the DPMO is lower and the sigma level is higher. Just for this example, let’s say that the cost of any defect is the same. That means that I am loosing more money in form of COPQ in the cell phones line. If what I am mesauring defects in ready to shipment units, that would mean that I am delivering more defective cell phones to the customers. What does the better DPMO and sigma level of the cell phones line tells me? That, taking into account the complexity, the cell phones line performs better, so let’s put our resources in improving the base units line, even when I am loosing more money and dissapointing more customers with the cell phones?

    #51496

    Mike:
     
    You said:
    “It is not about engineering. It is about management….a first pass is I can compare the two. It is not nonsense. It is how you determine where and how much you deploy resources. That is why you use DPMO in the strategic operations”
     
    Ok, play the role of my consultant for a moment. 
     
    I have three different divisions which produce three different types of products going to three distinct markets.  We have mapped the manufacturing process of the four and have used your prescribed method for counting defects and opportunities for the same time period on all three.  We have determined the following:
     
    Process 1:  DPMO = 160,000 Sigma level = 2.50
    Process 2:  DPMO =   66,800 Sigma level = 3.00
    Process 3:  DPMO =   62,000 Sigma level = 4.00
     
    With this information, can you tell me the following?
     
    Ø      Which process has the most complexity?
    Ø      Which process has the greatest opportunity ($s to the bottom line) for improvement?
    Ø      Which process produces products with the highest level of customer satisfaction?
    Ø      How should I deploy my improvement resources?
    Ø      If I was to sell off one of these businesses and use the profit to invest in another, which one should I sell and which one should I invest in?
     
    If the DPMO metric can tell me this, I will agree that it is a valuable metric for making strategic business decisions.  Otherwise, I will have to conclude that it does not correlate to my primary business metrics and has no value to my strategic decisions.
     
    Statman

    #51558

    Did Mike respond to this example or Gabriel’s?
    Just curious.

    #51565

    Statman,
    Sorry about the delay. I am in a new town and was a little distracted finding a dentist that takes new patients (customers). We may be in the wrong business.
    I will give you my take on the issue but like I said the DPMO is only a piece of the information that management needs.
    I don’t have to worry about complexity when I am using DPMO the metric normalizes the data. Basically it gives me a common metric that will let me decide at a macro level how easy a shift I have to make. Process1 (in the absence of a technology issue) should be able to improve easier because it is at a lower level. The improvement of Process 3 will be a much more difficult issue because most of the noise is out of the system and it leaves me with in most probability technical issue as opposed to something like Standardized Work that I cam probably get some improvement with in Process 1.
    I stated in an earlier post that $ were indigenous to the process so in the absence of cost data that is an issue.
    I would also want cycle time data to make a decision like this. If process 3 was the bottle neck and process 1 was still out running it then process 3 needs more resources.
    The point of using DPMO is that it takes away the issue of complexity. If I put products with your data rather tham processes and Process 1 is light bulbs, process 2 is appliances, and process 3 is aircraft engines. Engines gets more resources. They exist in a 6 sigma+ world and I know I have to ship at that level so I have to test and inspect to get it out the door. Even though light bulbs are at a lower level than appliances my cost is less and there isn’t any rework. The appliance process will draw rework costs were light bulbs will draw scrap or discounted prices of lower cost materials. I can throw away a lot of light bulbs for what rework or field failures will cost on an appliance line.
    The technology shift in Jet engines will also take a different level of expertise. Because of the issue with qualifying an engine it will be more of an issue with increasing process capability to maximize what I get out of the current qualified design. There is design opportunities in appliances and light bulbs because I am dealing with a less regulated industry and trying for a lower level shift.
    As far as what you sell that is more determined by a company strategy. If you compare Allied and GE in the late 90′s both were in the engine business. Allied had a 3 legged stool with Aerospace, Engineered Materials, and Automotive. GE had a 13 legged stool. Allied trimmed back Automotive and looked for aquisitions (like Honeywell). GE is always moving in and out of businesses but you don’t see them selling off lighting (a traditional business) or Engines (provides long term stability) even though Capital is really where most of the revenue is generated. If you make the strategic decisions for other reasons the DPMO or sigma level gives you the constant measurement system in a business like Engines that lets you constantly monitor how you are closing the gap between what you build and what you ship (how did the GE 90 engine compare to the older product line – am I manufacturing an engine that requires less test and inspection to get it out the door and still be a viable vendor in the Engine business).
    For some reason we don’t seem to be getting the idea that there is no single measurement system that gets you through this. I worked in a facility that had bad light bulbs (dashboard types) at over 500,000 ppm and another line with a bad ASIC at just under 500,000 ppm. The bulb was less than $0.001 USD per and the ASIC was about $8.00 USD per. I could get SQA resources on the ASIC because SQA was measured strictly on the defect levels of incoming material. Nobody is saying that one system does it all but it is used as part of the process.
    Just my opion.

    #51566

    sqe,
    Thank you for the follow up. I missed Statman’s post.
    Regards,
    Mike

    #51569

    Hi Mike,
     
    Great response!  A very well thought out response as well.
     
    I think the key to this is as you say: “For some reason we don’t seem to be getting the idea that there is no single measurement system that gets you through this”.
     
    The fact is, no business leader can make strategic decisions based on a single metric as there is no such thing as a single metric that can be used to compare performance across different products and processes.  My goal (or bias if you will) is not to discredit the DPMO metric, as it has its place and value, but to point out the fallacy that the DPMO and the resulting Sigma metric is the “be all end all” metric for process performance.   Strategic decisions like resource planning, customer value and satisfaction, make vs. buy, product line complexity, growth and revenue improvement, profitability, etc. are too important of challenges to be cheapened by the broad brush Sigma level generalities.  Fortunately, our brains are advanced enough to handle more than one metric when making strategic decisions.
     
    The critical question for Six Sigma practitioners is what is the proper application of DPMO?
     
    Does it measure complexity?  No, as you stated, it is normalized (actually I prefer standardized as it does not co-notate a normalizing transformation) across processes.  Process 1 having 2.5 times the DPMO of process 3 does not necessarily have more complexity.  There may be far fewer opportunities but higher occurance of defects.
     
    Does it measure opportunity for improvement ($’s to the bottom line)?  Not necessarily.  Process 3 may have achieved the level of 4 sigma by actually increasing cost rather than reducing cost.  For example, Process 3 could have had two independent black belt projects working on two sequential sub-processes.  The first project increased the batch size of sub-process one to reduce the defects due to change over, resulting in higher inventory at sub-process two.  Project two reduced the exit rate of parts from sub-process two by slowing down the cycle time to reduce defects at sub-process two. The result of the two projects: significant decrease in defect rate, significant decrease in DPMO, and a huge increase in WIP creating a huge opportunity for improvement even though the DPMO improved.  Therefore, I would say that it is related to but not a direct measure of opportunity for improvement.
     
    Does it measure customer satisfaction?  Once again, not necessarily.  There are two reasons it is not a direct measure.  First, as Gabriel has pointed out, not all defects have the same severity of impact on the customer.  The defects in DPMO are not weighted in terms of severity.  There are two aspects of risk, occurrence and severity.  DPMO only captures one.  Secondly, as Gabriel also pointed out in his spreadsheet.  There is not a one to one relationship between DPMO and PPM defective.  Once again, it is related to but not a direct measure of customer satisfaction.
     
    So what is the application of DPMO?  In my opinion, it works well for measuring and tracking improvement for a particular project or process with all other considerations being equal.
     
    Statman

    #51570

    Hello Gabriel:
    I think we were mixing Base Station (5000+ parts) with a Base Unit (20-100 parts).

    We should be looking at the measurements in context. We first use DPU that normalizes the defect rate to the unit. So, if one produces different number of units on different days, one can see the trend in the quaility of operations based on the DPU. Ultimately, the customer wants units of products or services with ’0′ defect per unit.
    Now, suppose in the same manufacturing facility, one builds two products. The base station that has about 5000 components, and a cell phone that has about 500 components. Suppose they happen to have the same DPU level. Can one say that the manufacturability or operational quality is same for both the products. The answer is ‘no’ as the base station has more opportunities, just base on parts count, for things to go wrong. This means a next level of details needs be looked into.
    Next we look into opportunities. How do we define an opportunity. Any item or activity during the process that can go wrong or produce a defect will be considered an opportunity. If there is a system that has 100 washers and they never failed, i.e., failure-free, there is no opportunity for error, so we should not count them. The reason for looking at opportunity is because that’s where defects lie. In order to remedy opportunities for defects, we establish a measurement that is based on opportunity. So, DPMO is needed to expand the DPU measurement. The quality of a product is directly related to the number of opportunities, therefore, DPMO.
    Now let’s take ur example:
     
    Lets say that the cell phone has several times the opportunities of the base unit. Let’s also say that I have more defects in the cell phone line, but because of the more opportunities, the DPMO is lower and the sigma level is higher.
     
    [Most likely will not be true. Because, base unit will have fewer defects proportionate to the related opportunities. So, it will most likely have a higher sigma. Opportunities do not just count alone in determining sigma level, instead they may cause more defects too.]
     
    Note: When I mention base station, it is the larger network center that routes and connects calls, not the charger type device.
     
    For improving the process, DPU and DPMO are good enough because they provide necessary information leading to process improvement.
     
    Why Sigma level then? Well, it creates a simpler and focused measurement tool that requires lots of effort to improve. A fraction improvement in sigma level could be a significant reduction in DPMO level. For example, going from 3 sigma to 4, we need to reduce DPMO by about 10 times,…
     
    So, when we move from DPU to DPMO, we go to a more detailed analysis of a process. When we go from DPMO to Sigma level, we get a much higher level (less detailed) information. Hierarchically, Sigma, DPU and DPMO is the order of measurements. It appears this way.
     
    Praveen

    #51571

    Hello Statman:
    I agree that DPMO or DPU are not the measurement to make some strategic decisions. And a set of measurements need to be considered to set some strategic direction. That’s why Balanced Scorecard and Six Sigma Business Scorecard have been developed to make strategic decisions.
    Kaplan and Norton have called their Balanced Scorecard system the Strategic Management System. 
    Regards,
    Praveen
     

    #51573

    Hi Praveen,
     
    That’s what I’m talkin’ about!
     
    There is this perception that if you don’t accept the premises of the Sigma/DPMO metrics then you are anti Six Sigma and are blind to the benefits of application of Six Sigma (statistical thinking) in strategic business planning.
     
    In my case anyway, this could be no further from the truth.  Metrics drive action and the application of statistical thinking and a Balanced Scorecard/Six Sigma Business Scorecard approach along with a Y =f(x) cascading to projects level metrics will provide the proper use of metrics to drive the right actions.
     
    Cheers,
     
    Statman

    #51579

    Hello Statman:
    You are absolutely right about having the right set of metrics for right actions. Six Sigma is more than metrics. I believe it consists of four things: Strategy (intent), methodology, measurements and tools. If we just focus on measurements, we become manipulative, if we focus on methodology, we may become bored with six sigma, if we use the tools only we may get partial results, and if we use only strategy part of it, we just talk about it. Therefore, we gotta use all four aspects of six sigma. The Six Sigma Thinking alone can lead to more benefits than any components of the Six Sigma methodology.
    Regards,
     
    praveen.

    #51580

    Mike and Statman:
    These two post are among the best things I have seen in this forum for a while. Very eye-openner. Thanks.
    Statman:
    “So what is the application of DPMO?  In my opinion, it works well for measuring and tracking improvement for a particular project or process with all other considerations being equal.”
    In that case, what is the added value of DPMO over DPU or PPM that pays for the added complexity and uncertainty of the measurment?
    Mike:
    I liked very much your explanation that, in other words, the DPMO or sigma level are an indicators about if going for improvement means going to the low-hanging fruit or to the fruit at the top of the tree. That gives you a first idea about how much effort may it take to make an incremental improvement. Of course, as you said, that is only one side of the equation. The other side would be what such an improvement is worth of (is this correctly said?). Is “that improvement” worth “that effort”? I thought that the sigma level was onn the side of the “value” of the improvemnt (How good would moveing from 4 sigma to 5 sigma be?) Now I see it is on the side of the “cost” of the improvement (How costly would be to meve from 4 sigma to 5 sigma?). So when I say “process A is 3 sigma” and “process B is 5 sigma”, I am not saying “process B delivers better customer satisfaction”, or “process B has a lower COPQ”, etc. What I am saying is “process B is more dificult to improve”. Wether it is a good decision or not to improve it depends on this but also on other factors. Did I finally get it? More or less at least?

    #51582

    Hi Gabriel,
     
    You asked:
    In that case, what is the added value of DPMO over DPU or PPM that pays for the added complexity and uncertainty of the measurement?
     
    It really depends on the objective and the approach to the project.  Lets say we have the following information on 5 sub-processes within an overall process:
     
                             Sub A           Sub B          Sub C             Sub D        Sub E       Final
                
    # of Defects:        500             25                300                 50              25                200
        OPU                500           150                180                100           200              1130
    # of Units             300           250                200                150           125                125
    Defectives              80            50                  30                  20              15                10
    # of Opp        250,000      37,500          36,000          15,000       25,000           141,250
    DPU                    1.67           0.10               1.50                0.33         0.20             1.60
    DPMO                2,000      666.67         8,333.33        3,333.33    1,000.00          1,416
    PPM               266,666     200,000        150,000         133,333      120,000          80,000
     
    Ø      If the objective was to reduce defect rate, you would focus on Sub C and use DPMO as your primary metric. The relatively high number of defects and low opportunity count makes it a candidate for improvement
    Ø      If the objective of the project was to reduce complexity, you would focus on Sub A and use OPU as the primary metric
    Ø      If the objective was to improve yield, you would focus on Sub A or Sub B (depending on cost of scrap which will probably increase as you move down the process) and use PPM as your primary metric
    Ø      If the objective was to reduce repair cost at final, you would Pareto Defects at final to determine the area for improvement and use DPU at final as your primary metric.
    Ø      Etc.
     
    I could go on and on.  There are many other primary metrics that could be used for various projects ( e.g. cycle time, WIP, Cpk).
     
    There are two important points from this example.  First, there needs to be congruency between the problem statement, objective, and primary metric for a BB project.  Secondly, these metrics are of course related but not one to one.  An improvement in one may not correlate to an improvement in another.
     
    Hope this helps.
     
    Statman

    #51583

    Repost with the table set up better
     
     
    Hi Gabriel,
     
    You asked:
    In that case, what is the added value of DPMO over DPU or PPM that pays for the added complexity and uncertainty of the measurement?
     
    It really depends on the objective and the approach to the project.  Lets say we have the following information on 5 sub-processes within an overall process:
     
                             Sub A           Sub B          Sub C             Sub D        Sub E       Final
                
    # of Defects:        500             25                300                 50              25                200
        OPU                500           150                180                100           200              1130
    # of Units             300           250                200                150           125                125
    Defectives              80            50                  30                  20              15                10
    # of Opp        250,000      37,500          36,000          15,000       25,000           141,250
    DPU                    1.67           0.10               1.50                0.33         0.20             1.60
    DPMO                2,000      666.67         8,333.33        3,333.33    1,000.00          1,416
    PPM               266,666     200,000        150,000         133,333      120,000          80,000
     
    Ø      If the objective was to reduce defect rate, you would focus on Sub C and use DPMO as your primary metric. The relatively high number of defects and low opportunity count makes it a candidate for improvement
    Ø      If the objective of the project was to reduce complexity, you would focus on Sub A and use OPU as the primary metric
    Ø      If the objective was to improve yield, you would focus on Sub A or Sub B (depending on cost of scrap which will probably increase as you move down the process) and use PPM as your primary metric
    Ø      If the objective was to reduce repair cost at final, you would Pareto Defects at final to determine the area for improvement and use DPU at final as your primary metric.
    Ø      Etc.
     
    I could go on and on.  There are many other primary metrics that could be used for various projects ( e.g. cycle time, WIP, Cpk).
     
    There are two important points from this example.  First, there needs to be congruency between the problem statement, objective, and primary metric for a BB project.  Secondly, these metrics are of course related but not one to one.  An improvement in one may not correlate to an improvement in another.
     
    Hope this helps.
     
    Statman
     

    #51584

    Gabriel,
    You’re right, about the two post of statman and mike. Maybe because it’s Christmas, a time to give and understanding.
    AbetF

    #51591

    DPU, DPMO, Y(RT), Ynorm, etc. all are important metrics and should be appropriately used. Fine. All explanation done. True six sigma practitioners are well verse with these.What about our old friend Dr. Mikel Harry, P. hd? Have you really forgot about the postings on the Zshift? Has Dr. corrected the mistakes in his book? I have not bought it yet, though wanted to buy sometime back. I still wish to buy his other books for which I saved pennies but only after…I think six sigma community should be made clear about the mistakes in the book or Dr. Harry should post an article on iSix Sigma giving explanations to this. This could be the best option if no going totally public.If Statman is really true then lets except that shift is 1.5 sigma only when Zst = 4.5 and Zlt = (2/3)Zst. Or is it that Dr. Harry has secretly called up Statman to close this issue?Friends, those who followed this debate closely should be eager to know this. I am surprised. Thousands go through messages daily posted and no one really thought of demanding an explanation.Better late than never for Six Sigma!hestatis

    #51597

    Statman,
    There is always a new learing opportunity at this forum. The latest post on DPU vs DPMO is great. I see myself aligning with you (quite a lot)on your thinking about Six Sigma in general and application of the same.
    Great job.
    PB
    PS – In your post, for Sub A the calculated value for  # of Opp  is 250,000 and looking at the calculations on the other values this should be 150,000, DPMO goes up to 3333.33. However, I do not see it changing anything in your presentation of your thoughts. 
    To All – Have great Christmas and safe holidays.

    #51601

    Statman – great example, thank you.  I have just one question, if you could clarify.  I was confused as to how you arrived at your PPM figures.  I noticed you used the “defective” count to calculate the PPM.  The reason this threw me off was because I understood PPM to be (dpu * 1,000,000) – “defects”, not “defectives”.  If you, or anyone could clarify my thinking I would appreciate it.
    Thanks.

    #51602

    Hi PB,
    Thank you for the correction and the compliment.
    I Should have used excel when I made the table rather than calculating it in tabular form. 
    Cheers,
    Statman

    #51606

    Hi Sqe,
     
    I should have been clearer in this post and defined PPM as PPM defective.  The reason for the clarity is that PPM is often used without the definition of what one is counting.  I have seen PPM and DPMO used interchangeably by some authors with out explanation.
     
    My definitions will probably not agree with others and I run the risk of creating a whole new string on defining PPM but here is the context:
     
    DPU = (# of defects)/(# of units processed)
     
    PPM (defective) = 1,000,000* (# of Defective units)/(# of units processed)
     
    Where # of Defective units is the number of units that do not meet acceptance criteria
     
    The reason DPU*1,000,000 does not necessarily equal PPM (defective) is that the distribution of defects across units may not be uniform.  In other words, all defects may be on one or two units and the other units have no defects.  If you had 20 defects and 20 total units, 12 defects on unit one and 8 on unit two.  1,000,000*DPU = 20/20 = 1,000,000  and PPM = 100,000.
     
    Statman
     

    #51607

    Thanks Statman.  I did some digging and discovered much the same as you mentioned.
    Again – nice example.

    #51610

    Hi Statman,
    Thanks for the quick response. Your response to Sqe has me thinking and maybe you can (or Gabriel, Mike C) can shed some light on this.
    We typically identify defects under a quality plan for each stage of the product. In order to find these defects (lead to being parts that are defective) we do random sampling with a certain confidence level. (Also, you inspect against an AQL with AQLs being tighter for some products – example medical device). Therefore, in your table you have # of units, # of defects, etc. The # of units processed you talk about – Are these the units that are the inspected 100% (to the quality plan) from which you derive the statistics (per your table) or the actual # of units processed? Should you actually not process 1MM pieces to find the PPM defective?
    Are theses defects the ‘likely defects’ or ‘actual defects’?
    If one has similar test criteria but different parts (made in different cells), does one combine the data to get the DPMO?
    How does one factor in customer return? If my customer returns 100,000 parts back because they found 10 defective (based on their control plan), what DPMO should I figure? When I shipped these parts, I counted on no defective being shipped.
    Your answers are appreciated.
    PB

    #51617

    While we are on the subject what can you tell me about Mama Juana?   Have you done any hypothesis tests concerning the before use and after use results?

Viewing 69 posts - 1 through 69 (of 69 total)

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

Login Form