iSixSigma

Not all Defects are Created Equal

Six Sigma – iSixSigma Forums Old Forums General Not all Defects are Created Equal

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • #27655

    Michael Kalman
    Participant

    I’m new at understanding the SS process. Going through your 5 step process to determining a process sigma value it occured to me that defects and opportunities should be weighted (as in a neural network structure). An example of why can be found in extending the Motorola pager example. If a pager doesn’t function between the hours of 2-3AM the defect may be less important to the client base versus a malfunction bewteen 2-3PM. Accordingly, the opportunites over the duration of a day may not be valued symetrically by the client base.
    In this case is it better to slice up processes into finer detail that more accurately identify true sigma values?

    0
    #68050

    Anna
    Participant

    Michael – While it’s true that not all defects are equally agravating to the particular customer to whom they happen, at the time when they happen, it is estremely difficult to predict for any given customer / defect experience the “value” of that defect, to that customer at the time when it occurs. And, if the defect is sufficiently minor or transient, the customer may never even tell the manufacturer. He or she just doesn’t come back. So the 6 Sigma philosophy is to reduce all perceptible defects to a very low level, and to treat every customer-perceptable defect you find seriously. If you really want to measure the “degree” of defectiveness in some instance, look into the Taguchi loss function approach. This approach, to oversimplify a bit, computes the monetary loss to society (customer, manufacturer, everyone) of the variation of a product parameter from the ideal or target specification. It’s not incompatible with 6 Sigma per se, and it isn’t all that much harder to use. American Supplier Institute teaches and publishes a lot of Taguchi material.

    0
    #68063

    john beaudoin
    Participant

    By definition of 6-Sigma, your sigma level will never be weighted.  But you do have a point that not all defects are created equal.  When you evaluate your process, you will need to perform a Failure Mode Effect Analysis (FMEA).  There is a good version available in SPC KISS software (macro software for Microsoft Excel), which is available through the Air Academy of 6 Sigma Training or from a book “Basic Statistics Tools for Continuous Improvement”.  An example of how FMEA works is that if you were an automotive company making cars, there would be several opportunities of defects in your vehicle.  If you make 200,000 units, and 1 out of 1000 has a braking problem and 100 out of 1000 have a turn signal problem, then the FMEA will tell you which item you should concentrate on next.  By looking at the detectability of the flaw, the rate of occurrence of the flaw, and the severity of the flaw, you will get a number that will help rank the importance of the problems.  In this example, since the repercussions of bad breaks is much more severe than that of a bad turn signal (accidental death possible), the value on an FMEA for the breaks should show that although the rate of occurrence for bad breaks is much smaller, it still should be the primary focus for improvement.  Your sigma level on the turn signal will still be poor, but remember the goal of your projects is to reduce the COPQ (Cost of Poor Quality) and you will want to go after the big hitters first.  In the example with Motorola Pagers not working between 1am and 2 am, the Detectability will be low, but the rate of occurrence would be fairly high, and the severity of the flaw could also be high (A doctor being paged for an emergency).  Although there is a low chance someone will experience the problem, the other 2 factors will give you a high RPN number (formed by multiplying the ratings of 1-5 with 5 being severe of Detectability, Rate of Occurrence, and Severity).  In this case: 1 x 5 x 5 = 25.  If the problem was the light would not come on, you would have a 3 for detectability (only would notice at night or in dark room), and it happened on 1 out of 10 pagers, you would have a 3 x 3 x 2 = 18.  (the 2 was for the fact that you could find another light source to see the page, but it would be inconvenient.  You can see that the 1st problem is more important than the 2nd.
    Hope this helps.

    0
    #68065

    Joe Burke
    Participant

    Thanks for the great and clear discussion about the “weighing” of defects. But I need some help on the “detectabilty” factor. Could any of you provide a practical example of how that would be measured. I work in the health care area and would like to understand how that factor would be quantified in the clinical care realm of a hospital.
     

    0
    #68069

    Dewayne
    Participant

     Michael, Anna makes some good points in that each identified problem should be taken seriously, especially in that you may not get feedback by the party involved, they just don’t return. I fully support John’s recommendation of FMEA usage, as it provides the mechanism to evaluate all the identified facts involved, therefore giving you a method of incorporating the weighting factor you are looking for, not by addressing the sigma but by allowing you to prioritize your planning in resolving the known problems.
    I assume your organization is registered to ISO 9001:2000, or seeking registration to the new specification (or the QS system), and the new spec concentrates much more heavily on evaluating customer inputs and subsequent feedback to them. If you create a process based on FMEA, I suggest that you feed your developments up through the system to your ISO development/maintenance group so that they can be sure and use reference to the technique you create as a means of addressing the new spec requirements for customer feedback when they are audited.  This method of addressing your customer’s needs can also be a valuable source of corrective action feedback to them, as they then know you are using the well know FMEA methodology for evaluating their input and addressing their (your) areas of concern.    – Dewayne  

    0
Viewing 5 posts - 1 through 5 (of 5 total)

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