iSixSigma

2 sets of control limits

Six Sigma – iSixSigma Forums Old Forums General 2 sets of control limits

Viewing 17 posts - 1 through 17 (of 17 total)
  • Author
    Posts
  • #47121

    pilot1012000
    Participant

    Hi all – first time user / asking.
    I was asked to produce a chart showing information on incident rates, using the standard upper and lower control limits.  The person asking for the chart then said “can we also show the second / additional set of upper and lower control limts?”
    Now, I am very, very new to Six Sigma, and did not want to let the person know that, and I have asked around everyone I know in our workplace, but no one has any idea what he was asking for, so that is why I am here….
    Any ideas?
    Thanks!

    0
    #156764

    DaveS
    Participant

    pilot1012000
    “Oh, what a tangled web we weave…..when first we practice to deceive.”
    Sorry, but your ego has gotten you into a tough spot.
    The “standard” control limits are generally taken at +/- 3 sigma. “Second/additional” limits has no accepted definition. Practicioners hav for 80 plus years attenmpted various other schemes (+/- 2 sigma, acceptance limits, modified limits, zone charts, precontrol limits), but none have been generally acknowledged as superior to Shewharts scheme and been considered second/addditional limits by the community at large.  
    So you really have only one recourse. Return to the original requester and ask “WTF do you mean?” There may be a local interpretation, perhaps only in his mind.
    If this guy is really sharp, it might have been a trick question to see if you knew your stuff or were a poser. If so, you’ve answered it for him.
    I’d be curious as to what he meant also, if you find out; post it.  

    0
    #156765

    Rupesh
    Member

    Hi.. I think the person means the standard limits on the basis of +/- 3 Sigma lines as well as the lines on the basis of +/- 1,2 sigma. This you can do by putting your data in Minitab. You have to use the option of drawing the lines for different sigma limits.

    0
    #156783

    ShootMeNow
    Member

    There is only one set of control limits.  Dont guess. Go ask.  Get used to not knowing and asking questions.  You wont know anything for along time.  Good luck.

    0
    #156786

    BTDT
    Participant

    pilot1012000:I agree with Daves about this one. There is a bit of ambiguity in the question you have been asked.- the +-3 standard deviation limits have been accepted for decades. Please don’t change them.- the “second” set of upper and lower control limits is probably a way of asking, “Can we add the specification limits to the same chart?” This is not really accepted for control charts, but I can see why someone might ask this question.Talk to this person and ask them,”Do you want me to add a second set of ‘control limits’ to see if the process is within customer specification? If so, we should not call them, ‘control limits’.”Cheers, BTDT

    0
    #156787

    pilot1012000
    Participant

    Thanks all for the info / answers.
    I checked back with the customer, and it seems he wants to have the +/- 1 and +/- 2 limits added, in order to more easily identify when a process is in or out of control (his words).
    So, ego has been pushed to the back of the office, and he was cool with the question – i just posed it as a ‘singing off of the same hymn sheet’ question.
    Thanks again all!  Really appreciate the help – i iwll definately come here again for answers!

    0
    #156789

    BTDT
    Participant

    pilot1012000:The logic behind the +-3 standard deviations is to balance the number of false alarms against detecting a real change. If you change them to +-1 standard deviations, then you will have to put up with about one out of twenty signals being a false alarm.It will NOT result in a process which is more pro-active in detecting changes.There is some research that shows you will create more variation in a system by reacting too soon. See what you can find about the Deming ball and funnel experiment.Cheers, BTDT

    0
    #156791

    Dillon
    Participant

    BTDT,
    I don’t disagree with what you’re saying but I took the request to mean they just wanted to more easily identify violations of the other seven tests Minitab conducts.
     
    Regards,
    Doug

    0
    #156792

    walden
    Participant

    The customer may want the +/- 1 and +/- 2 sigma limits for using the Western Electric rules for interpretting control. See the below link for more information.
    http://www.answers.com/topic/western-electric-rules

    0
    #156793

    BTDT
    Participant

    Doug:That would be an interesting exercise: to determine the changes to the 7 tests such that they all give an elevated alpha risk.Hmm, I’ll leave that for another day.Cheers, BTDT

    0
    #156794

    Craig
    Participant

    My guess is that the customer wanted a depiction  of zones A, B, and C in the chart by adding the 1 sigma and 2 sigma lines. Asking for the other set of limits was a poor way to word the question (If my assumption is correct!)
     

    0
    #157113

    sathish chandran
    Member

    Hi,
    From the detials you have given it seems like there could be two possibilities
    Case #1 – The person who asked you the question might be testing your knowledge on SPC
    Case # 2 – The person might be asking if the control chart can be drawn in such a way that there is pre control limits to warn us early of any changes in process (may be having a green zone, yellow zone and a red zone above and below the mean).
    You need to check with the person on this and take the appropiate action on the charts and calculate the revised control limits and draw the chart to monitor the process
    with regards
    sathish chandran

    0
    #157144

    Va
    Member

    Hi Pilot,
    The first that you should do is to plot the incident rates using the std.UCL and LCL and from there you can analyze the need to show the second set of limits.  The spread of the process will tell you how you will handle the situation.  I presumed that your attribute control chart(%) will give you an idea whether it will fall into 1,2,3….6 sigma process and discuss it with him/her and before you do that try to study patterns of variation and the meaning of 1,2…6 sigma.  The person was asking you the possibility of having second set of limits but not telling you to have one.
    I know that you are smart enough to handle this inquiry.  The company will not put you there if you are not capable.
    Good luck….Va

    0
    #157146

    Va
    Member

    Correction:
    Hi Pilot,
    The first thing that you should do is to plot the incident rates using the std.UCL and LCL and from there you can analyze the need to show the second set of limits.  The spread of the process will tell you how you will handle the situation.  I presumed that your attribute control chart(%) will give you an idea whether it will fall into 1,2,3….6 sigma process and discuss it with him/her and before you do that try to study patterns of variation and the meaning of 1,2…6 sigma.  The person was asking you the possibility of having second set of limits but not telling you to have one.
    I know that you are smart enough to handle this inquiry.  The company will not put you there if you are not capable.
    Good luck….Va

    0
    #157219

    Bryan penn
    Participant

    If im understanding you correctly, there asking you if the product is undercontrol.  You can run a control chart. Of course you have to do several runs of the product whether it weights or measurements. Once you find that out you can put that into a probability chart using your upper can lower control, you would also have to know you target. If your product shifts that would tell you what your new upper an lower specs would be. You cant change your target. You may also  only have to change one spec depends on which way your process shifts. Your new spec would be what makes you product closer to target also giving you a better bell curve.

    0
    #157223

    Gary J.
    Participant

    Is it possible he/she wanted to see the control limits based on historical data from the process, rather than current production, to get an idea of progress?  Or maybe some points were deleted from the set due to the finding of an assignable cause in process variation (that can inflate the estimate of the process standard deviation and thus the control limits of an SPC chart), and he/she just wanted to see the new limits with the elimination of said data?  Just wondering…
     

    0
    #157238

    SixSigmaGuy
    Participant

    i agree with gary.. or maybe control limits before and after improvement? Jus a thought..
    Regards,
    Six Sigma Guy

    0
Viewing 17 posts - 1 through 17 (of 17 total)

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