iSixSigma

Normality, Out of Control Cp/Cpk

Six Sigma – iSixSigma Forums Old Forums General Normality, Out of Control Cp/Cpk

Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
    Posts
  • #37608

    Shawn Struzik
    Member

    Is it bad to have a process that is out of control(IMR Chart), Non-Normal data distribution and a Cp/Cpk of 3.9? How can one explain to a customer this is not a bad thing?

    0
    #111112

    Bob J
    Participant

    Shawn,
    It is bad… 
    The fact that you have a process that is not in control means exactly that there are factors that are significantly impacting your outputs that are unknown, unmonitored and unmanaged.  That being the case you can never count on any consistant performance level or drive any significant improvement….
    Best Regards,
    Bob J

    0
    #111123

    beverly daniels
    Participant

    worst case – the 3.9 is a result of a very ‘targeted’ sample not representative of the process. 
    However:  assuming that you do have a representative sample, the ‘out-of’control I, MR chart could simply be due subgrouping errors or you have selected the wrong chart for your statistic. 
     
    Although the traditional conversion from Ppk to ppm RELIES on a Normal distribtuion, there is NO requirement that processes be Normal.  Most aren’t.  There is no reason to achieve a Normal process.    A good process is one that is stable, predictable and capable.  Normality has nothing to do with that.
     

    0
    #111125

    Shawn Struzik
    Member

    Can you be stable and out of control at the same time?

    0
    #111126

    beverly daniels
    Participant

    Noooooo.  Stable = in control
    UNLESS you are charting (subgrouping) incorrectly. 

    0
    #111127

    Shawn Struzik
    Member

    I’m using Minitab, capability 6 pack. My subgroup size is one. Machine process, 40 consecutive pieces with no offsets.

    0
    #111130

    beverly daniels
    Participant

    well, there you go.  you have not gotten a representative sample.  In fact is is probably ‘best case’.  no offsets, no operator to operator, set-up to set-up or raw material batch variation…
    Without seeing the data I might think that the only active variation is tool wear and that would give you an out-of-control I, MR chart with a high Cpk.  (if you’re using minitab – you are probly also looking at Cpk which is the within sugroup capability and can be quite ‘good’ looking…be sure to use Ppk) 
     
    Are you experiencing defects from this process?  if so, then your study is way too compressed in time.  Try taking subgroups of 3 every 2 hours for a couple of weeks…

    0
Viewing 7 posts - 1 through 7 (of 7 total)

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