iSixSigma

UCL and LCL

Six Sigma – iSixSigma Forums Old Forums General UCL and LCL

Viewing 8 posts - 1 through 8 (of 8 total)
  • Author
    Posts
  • #31681

    Rj
    Member

    I have to explain to some outside audience what exactly UCL and LCL means.  My difficulty is putting it in terms they will understand.  Does anyone have a good explaination for a average person?

    0
    #83689

    Swaggerty
    Participant

    Control LimitsControl limits define the area three standard deviations on either side of the centerline, or mean, of data plotted on a control chart. Do not confuse control limits with specification limits. Control limits reflect the expected variation in the data

    0
    #83690

    Gabriel
    Participant

    Control limits define the zone where the process parameter is expected to be when the process behaves as it usually does.
    There is strong tendency to confuse or relate control limits with specification limits, even when explained.
    That’s why I found usefull to use some figures for which there is no specification.
    For example, you can use body temperture. Nobody ever specified where they wanted the temperature to be. It is what it is, and we should ask God or Mother Naature if this is the temperature they expected or specified.
    So, the body temperature is between 35ºC and 37º. We don’t know why, but when the body is behaving as it ussually does it is in that range. When we find the temperature beyond that range, the body is not behaving as it ussually does and then there must be a special cause of variation (like n infection) that we should identify and eliminate (take medicines to kill the virus). Once it is done, the body temperature returns to the normal range. Look that the way to recover the normal temperature was not acting on the temperature itself but on the special causes that made it go beyond the normal values.
    Another nice example was using the arrival time to work of three employees (the data was invented, but it could be taken from acrual records). Then the arrival time was plotted by day for each employee, and it was nice to look for “abnormal” times, trends, etc, and imagine what could have caused that “special variation”. It was clear that there were differences in the average and variation of each emplyee, and it was also evident that there were some emplyees that had a very “random” pattern (moving consitently arround a certine range limited by some “control limits”) and some with a lot of “special causes” (like points clearly far from the random distribution). Finally, I asked which of them was on time more often. It was a tricky question, because I used times close to the arrival time at our factory, but I didn’t say that it was a case from our factory. So when they started guessing, I asked: “How do you know if you don’t know the specified arrival time? It could be what you say, but they three can be allways in time or allways late”. That was good to introduce variation, stability, specification and capability.
    Hope this helps.

    0
    #83761

    Erik L
    Participant

    RJ,
    The simplest explanation that one can give to the control limits is that the represent ‘action limits’.  Managers are always looking for ways to ensure that their limited resources of people, time, and money are used most efficiently and against truly salient issues.  The control chart limits trigger a reaction from management to isolate and eliminate events that are degrading performance and triggering customer dissatisfaction.
    Regards,
    Erik

    0
    #84070

    Hersey
    Participant

    Ditto what the other guys said pretty much…. but also when done correctly and if all points are in control you have essentially defined the boundaries of the “normal” in-control process where we’re sure 99.97% of the measurements will be.  Therefore–when outside of the boundaries–(given a good measurement system) we’re highly confident that something has changed. Whether you react or not is another decision and that depends on what you’re monitoring.  Example:  Sometimes you can go past the control limits-but in a very positive direction–ie. You’re monitoring body fat measurements and you have broken the lower control limit–your body fat has been reduced and something changed to cause this-i’t s NOT just a good day on the scale-what you’re doing is working.
    A final note:  While the control limits represent the “normal” process– don’t make the mistake to assume that it is normal to be in-control.  More often than not, the process will be out of control when you first measure it and this is actually normal (a state of chaos is a normal reality).  Deming as well as others have writings on this phenomena as well–noting that it takes “work” to bring them into control.  Don’t underestimate this.  From my experience he is very correct.  Good luck.     

    0
    #179401

    sharples
    Participant

    The # of sigmas from the mean does not have to be 3 though.  The control limits simply define the outer limits of spec variability.  A very high-quality process may require 10 sigmas, while a lower quality system may be fine with 1 sigma.

    0
    #179404

    Mikel
    Member

    Great Graham, not only an old post, but nonsense for advice.

    0
    #179421

    Brian M
    Participant

    LOL

    0
Viewing 8 posts - 1 through 8 (of 8 total)

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