iSixSigma

DPMO

Viewing 28 posts - 1 through 28 (of 28 total)
  • Author
    Posts
  • #52698

    Schuette
    Participant

    My company is considering changing from the RTY metric to the DPMO metric in four of its plants to provide a “common” measure of performance across the plants for upper management.  Our plants are job shop and batch process in form.  We do not mass produce the same products.  We manufacture and assemble smaller (<10) quantities of products that we sell to our customers.  Many of our products have special features or options ordered by our customers.  Is the DPMO metric the best metric for measuring product quality in a job shop / custom build environment?  What metric would be better and why?
    Feedback is appreciated.

    0
    #185727

    Zak Roy Yoballa
    Member

    I think that the DPMO calc would be great for your situation because you deal with such small and special orders.  At the plant I work at, we produce approx 85,000 cases a day where one or two mistakes doesn’t even make a ripple.  In your case of 10 items, if you have to rework 1, that’s huge!  I would imagine that your profit margins would reflect that as well.  If you use the calculator on this site it also gives you a sigma value if you are uncomfortable looking at DPMO.  I suggest you play around with it, just for grins…
     
    ZRY

    0
    #185730

    Schuette
    Participant

    Thanks for your feedback.  To make sure I am clear about our products, I will expand on the description more.
    For example, one product line may make only 10 units.  Some product lines make up to 70 units per shift.  However, each product or unit contains many parts (up to 100 parts per unit, some units take more parts – these are refrigeration cooling systems / units). 
    For calculating the DPMO, is it best to say each unit has a total of 100 opportunities (each part installed in a unit equals one failure opportunity as a whole) or should all the opportunities for failure for each part be rollled up for the calculation (i.e., 100 parts in a unit with 10 opportunities for failure each = 1000 opportunities for failure in the unit built)?

    0
    #185732

    Eric davis
    Participant

    DPMO is the worst possible measure. 3.4 dpmo is utter nonsense.Use control charts to monitor quality.

    0
    #185733

    newbie
    Participant

    I don’t think you will find a concensus as to which capability metric is “the best” but there are situations where one might be a better fit than another.
    With DPMO, how you define your opportunties will make all the difference.  DPU is often quite useful in that it is a bit more responsive to effects than DPMO, can be charted (ie U Chart),  and you can take the inverse natural log of it to show RTY (or the convert it the other way around).  Whatever you use, trying to capture every potential failure for your DPMO calculation would be an error….do you really have the time and resources to validate 1000 different potential failures in a single part?  If it has failed in the past and hasnt been mistake proofed, include it.  If it has never failed or hasnt failed in the last 12 months (for example), why consume resources to prove what is already known?
    I would say the metric isnt as important as picking one, standardizing its operational definitions and methodology for a given location or process, and then showing improvement in the metric and within that value stream over time.   Good luck.

    0
    #185737

    Szentannai
    Member

    Hi,
    I guess it will all come down to the definition of the opportunities. I would consider each manufactured part as 1 opportunity and each defective part as 1 defect. That would give you a picture of the quality of the production in each plant, that is BTW very similar to Yield.The problem IMHO with DPMO to measure the production process, where you count all the opportunities to make a defective part (O > 1 per part) is that it risks to provide a picture which says, that “we produce many defectives, but it is OK because our product is sooo complex”. 1 per part) is that it risks to provide a picture which says, that “we produce many defectives, but it is OK because our product is sooo complex”. 1 per part) is that it risks to provide a picture which says, that “we produce many defectives, but it is OK because our product is sooo complex”. Also it will make the inspection process very cumbersome . E.g. if you define 100 opportunities for defects per part and by inspection you find a defect after checking say the first 10 opportunities, you can not just stop and put the part away as common sense would require. Instead, you will have to continue inspecting until all your opportunities are accounted for, in order to get the correct defect count.Regards
    Sandor

    0
    #185739

    Mikel
    Member

    What nonsense.

    0
    #185740

    Szentannai
    Member

    Care to be more specific?

    0
    #185741

    Mikel
    Member

    The objective is not to monitor quality, it is to improve it.But I agree, DPMO only for the purpose of having a common measure
    is stupid.

    0
    #185742

    Mikel
    Member

    Your conclusions don’t make sense.The one of the finding a defect and having to keep going because
    of a metric. I know of no place or business where this is the case. If the product is repairable, all the defects will need to be found
    and repaired prior to shipment. Records of this should be kept. If the product is not repaired, there should be a failure analysis
    strategy for at least a portion of the failed product. Records should
    be kept of this.The records of either should be analyzed and acted on.This does not have anything to do with SS, it is common sense.And your first point of opportunities masking problems with
    products, just the opposite is true in practice. The understanding
    of complexity is to stop the excuses of complexity. Which would
    you rather have – someone who says “I understand the complexity
    of my product and I know how good my processes must be to
    protect my customer” or someone who says “oh, you just don’t
    understand, my products are so complex”?And to those of you who say 3.4 is nonsense – we all use products
    daily that are that good or better. It is a matter of cost and
    reliability. Think of all your electronic gadgetry – all of the memory
    and microprocessors are that good and usually the assembly
    process as well. How do you think we get to 16 gig memory sticks
    that cost $30?

    0
    #185744

    IW
    Participant

    Make them in China, Taiwan, or Korea.

    0
    #185746

    Mikel
    Member

    What nonsense. There is as much crap made in those countries as anywhere else.There are also companies that make very good products there, just
    as there are in the US, Mexico, Europe, the middle east, …The only thing those countries had is cheap labor and an educated
    work force. The cheap labor is going fast – on to Africa (it’s
    happening).People will be saying the same thing about Cuba a decade from
    now.

    0
    #185748

    Szentannai
    Member

    Hi Stan,
    I am talking in the context of using DPMO as a quality metrics – the way the question was raised in this thread.You say:”there should be a failure analysis strategy for at least a portion of the failed product. Records should be kept of this. “What I mean is that “at least a portion” will not be sufficient if you want to measure the DPMO – it will lead to counting only a portion of all the possible Ds while keeping the O at the previously defined high level.
    I think, to measure the DPMO correctly, you will have to check every opportunity, which will be quite involved if you define say 1000 different opportunities. If the product is repairable, I agree, this is a must, but if not? Can/will people have the necessary time, during normal production? “And your first point of opportunities masking problems with products, just the opposite is true in practice. “Well, if DPMO is a metric to judge the performance of a plant by, I guess the temptation will be considerable to use this argument. But maybe I’ve just seen more of this sort of number games as normal ? Regards
    Sandor

    0
    #185760

    Severino
    Participant

    In my opinion (and it is only an opinion), the simplest metric is the best.  Measure your defective parts per million.  The only time you should be going through all the opportunities is if you are looking to simplify your design and/or processes or to introduce robust controls for each opportunity.  Your customer doesn’t care how many opportunities you have to screw up.

    0
    #185761

    Spectacle
    Member

    This is the same method we use when we glaze spectacles. We take the average size of a scratch and divide the area of the lens by the scatch size to calculate the number of opportunties. With our new equipment we think we can achieve 13 sigma perfomance  :-)

    0
    #185762

    Mikel
    Member

    Gaming is a problem. Those who claim to be the best (statistically different) have to show
    everyone their process for the purpose of learning. Those that show big jumps in performance have to show everyone
    their process for the purpose of learning.Those gaming will be found out rapidly

    0
    #185764

    N.Ganesan
    Participant

    Best in a manufacturing of even varying batches to stiffle others into attention first I found is to calculate OEE and then process sigma for problems identified. It has a real magic effort to get started and people also co-operate at all levels. You can do that for different batches of various sizes and evaluate your process operating efficiency to target your precious customers and improve your processes along with it to Six Sigma. 

    0
    #185766

    anupam
    Participant

    Hi Jim
    Answer:
    All the opportunities for failure for each part be rollled up for the calculation (i.e., 100 parts in a unit with 10 opportunities for failure each = 1000 opportunities for failure in the unit built).

    0
    #185767

    Chocolateur
    Participant

    I agree. Consider two chocolate products – one solid and one with indentations for breaking the chocolate bar into several pieces. It would be nonsense to suggest one of the bars had more opportunities than the other.Previous discussions have convinced me the DPMO metric is seriously flawed because it cannot be applied to all types of processes and should be ‘thrown in the trash’ along with the 1.5 sigma shift.

    0
    #185769

    Szentannai
    Member

    Hi,
    at the risk of being called an idiot :) I would like to raise some concerns I always had with DPMO, in the hope of getting some good answers to them.My problem is with the case where we have a complex process, consisting of many steps. If I count the opportunities, as the steps adding value, I will have a large(ish) number of Os. 1. Counting the defects will be interesting in this case – to stay consistent I`ll have to insist that on each inspected part each opportunity should be inspected, otherwise my DPMO measurement will underestimate the real number. IMHO this is quite difficult to maintain in a normal operational context (I mean after the Control phase.)2. The other problem is about the significance of this number. I think that the number in itself gives no relevant information about where the problems in the process are, moreover it will cloud the problems rather then making them apparent. Yield per process step, or equivalently the number of defectives per step would be much better measurements IMHO.This is not to say that we shouldn`t consider the opportunities of making a defect per process step and indeed buils Paretos with them to see the improvement opportunities better. I only mean that when we aggregate this information into a DPMO number we lose all the valuable information.So, anybody cares to comment?Regards
    Sandor

    0
    #185770

    Layman
    Participant

    Sandor,
    Just because someone might call you an idiot in this forum does not imply you are. There are mean spirited people everywhere these days – I think some call it self-expression, so you should expect someone to project one of their inner personalities on to you, especially in an open forum. (Here the arguments reign supreme – not personalities!)
    Several posters have already suggested using an operational definition for opportunties (Mach’s) If one really has to report a DPMO this might be a way forward.
    For example, in the case of the die there are about 25 ways to put a dot on a cube, so one dots would represent 1/25 opportunity and a six would represent a 6/25 opportuntiy since all dots are indistinguishable. If they were not the situation would be different. Therefore, one could report a DPO for cubes, dimples, and painted dots.
    This approach would allow you to determine what you can make well and what you can’t.
    Taken all things into consideration, I don’t believe DPO or DPMO can be used as a standard or as a goal. Others have already commented on the need to eliminate defects which I also believe is the correct approach.
    Personally I always recommend using Yield (100-%Defective) because in my field it is difficult to convert area, or line defects into individual defects.

    0
    #185771

    Mikel
    Member

    It’s not opportunities for failure. It is the number of value added
    operations, i.e. the number of things that must be done right.

    0
    #185772

    Mikel
    Member

    Sandor,The idea of having to inspect every opportunity is incorrect.Consider two operations. One has been studied, an accurate FMEA
    is in place, every input is appropriately controlled and it is known
    that the capability of every critical input and every customer critical
    output has a Cpm > 2. Customer complaints are monitored and
    field failures are monitored – both metrics are best in class and no
    unexpected failure modes have been detected in more than a year.
    The other operation is run by an expert businessman who contends
    there is an optimal level of quality and his process is achieving it
    even though failures are rampant both in his factory and the field.
    His QA people are constantly tied up firefighting customer
    complaints and new and unexpected failure modes come up
    monthly. 2. Customer complaints are monitored and
    field failures are monitored – both metrics are best in class and no
    unexpected failure modes have been detected in more than a year.
    The other operation is run by an expert businessman who contends
    there is an optimal level of quality and his process is achieving it
    even though failures are rampant both in his factory and the field.
    His QA people are constantly tied up firefighting customer
    complaints and new and unexpected failure modes come up
    monthly. 2. Customer complaints are monitored and
    field failures are monitored – both metrics are best in class and no
    unexpected failure modes have been detected in more than a year.
    The other operation is run by an expert businessman who contends
    there is an optimal level of quality and his process is achieving it
    even though failures are rampant both in his factory and the field.
    His QA people are constantly tied up firefighting customer
    complaints and new and unexpected failure modes come up
    monthly.Two questions -1) Which operation needs the most inspection and test?
    2) Which one is approaching the idea of 6 sigma?

    0
    #185773

    Mikel
    Member

    Considering all else was equal, the chocolate bar with the
    indentations would have one more opportunity than the solid one. the
    step where the indentations are formed is one opportunity.

    0
    #185774

    Szentannai
    Member

    Hi Stan,let me make this more concrete: suppose we want to measure the process of filling out a form, for example. The form has 20 different fields each one of them can be filled correctly or not. The team defines the number of opportunities as 20, which I would have to agree to, right?
    Now, during the inspection, the first field is the address of the customer. If this is incorrect, whatever else stays in the form, it is completely useless as the form will never reach the customer and so it can be thrown away. Still, to measure the DPMO with 20 opportunities, I would have to go on and check every field for correctness. I think, that during normal monitoring this will be very difficult to achieve, as the testers already know that they have scrap, so why should they bother ?The second point is actually more disturbing for me- whatinsight do I gain with the DPMO number into this process, that is more helpful as the Yield?

    0
    #185775

    Severino
    Participant

    I think the overall question is not so much, “How does one determine an opportunity” (although I realize that is central to the debate about gaming and has been asked), but rather “For the purposes of metric reporting, why do I care?”

    0
    #185776

    Chocolateur
    Participant

    Exactement …

    0
    #185777

    Chocolateur
    Participant

    We do not consider changing the set-up as an extra step. For the planar bar we use one mould, for another with indentations we use a difference mould – one with the indentations.
    Bonne chance

    0
Viewing 28 posts - 1 through 28 (of 28 total)

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