iSixSigma

Spec Limits Outside of Control Limits on X-bar/R

Six Sigma – iSixSigma Forums Old Forums General Spec Limits Outside of Control Limits on X-bar/R

Viewing 17 posts - 1 through 17 (of 17 total)
  • Author
    Posts
  • #43172

    KKN
    Participant

    As we are implementing control charts to monitor daily production metrics, we are running into quite a few where the CL’s are outside the SL’s. Other than the obvious answer that we need to work to reduce the variation in the process (already working projects to do that), does anyone have advice on how use control charts in this type of envirioment.

    0
    #136651

    Craig
    Participant

    Make you are going to assure that non-conforming product gets scrapped or reworked. An Xbar/R chart would present great risk since you are plotting the mean of the sample.  If you want to use a variables SPC chart, use an individuals chart.  If you are stuck on XBar/R, research “Pre-control” and this might shed some light.
    I would start doing mulitvari analysis now and determine where the sources of variation are.
    Good luck

    0
    #136653

    TwoCents
    Member

    Is the process stable?  Just curious…
    It sounds to me like you’re using them exactly the way they should be used…you’ve identified the fact that your process has too much variation, and are therefore attempting to figure out why.
    Sounds pretty good to me!

    0
    #136668

    KKN
    Participant

    I mis-typed. I’m actually using an IMR chart. My real question is not around capability and stability. It’s more on how to coach a group that is new to SPC and SS to switch from a run chart with spec limits to a control chart when the control limits are greater than the spec limits (As the process bounces up and down, often crossing the specs but staying inside the controls, they are going to react to the out of spec conditions. I’m having a hard time to with the value proposition of switching to control charts)

    0
    #136669

    Anonymous
    Participant

    Explain special cause versus common cause variation. You simply cannot address special cause variation with spec limits. This holds true even if the CLs are greater than then SLs.  So address the special causes with the control charts (and definitely do not place the spec limits on the control chart). Then tackle common cause variation with Multi-Vari as mentioned by the previous poster (along with the other DMAIC tools).
    To explain the value prop use Deming’s funnel experiment to show how responding to common cause variation increases overall variation.  Another good demo (if you can get a sampling bowl) are the red beads where you promote the low defect operators and demote the high defect operators.

    0
    #136684

    R.M.Parkhi2003
    Participant

    Instead of wasting your time on conventional statistics, go for Shainin Bhote technique. Multi vari analysis will help you identifying the causes of variation. You may kindly refer to the book ‘ World Class Quality ‘ by Mr. Keki R. Bhote publishers – American Management Association.
    Regards,
    R.M.Parkhi

    0
    #136685

    Bhardwaj
    Participant

    In Xbar-R control chart or any other cotrol chart Specification Limits ant Control limits are entierly different things Control limit is just for the process capability in which 99.7% of the times your process is control if all values fall between control limits. But in case of specification limits it is derived from customers ends.
    Awanish

    0
    #136692

    Anonymous
    Participant

    Hear Ye, Hear Ye, Thou disciples of Shainin. Come listen to the Word of Shainin as delivered by Keki Bhote. Bow down at the feet of Shainin for there is no other.
    Thou shalt not worship other foreign gods or use their methods. These include classical DOE methods by Box, and robust techniques by Taguchi.  The lowest of these foreign gods are Shewhart and Deming who espouse classical Statistical Process Control.
    Only Shainin’s Pre-Control is worthy. Only Red X is worthy. Only Component Swap is worthy.  All other techniques developed by Non-Shainins are sinful.
    References to years of research in journal articles  and proven methods are not necessary for we have the Word of Shainin as delivered by the Prophet Bhote.
     

    0
    #136765

    KKN
    Participant

    I mis-typed. I’m actually using an IMR chart. My real question is not around capability and stability. It’s more on how to coach a group that is new to SPC and SS to switch from a run chart with spec limits to a control chart when the control limits are greater than the spec limits (As the process bounces up and down, often crossing the specs but staying inside the controls, they are going to react to the out of spec conditions. I’m having a hard time to with the value proposition of switching to control charts)

    0
    #136766

    KKN
    Participant

    I mis-typed. I’m actually using an IMR chart. My real question is not around capability and stability. It’s more on how to coach a group that is new to SPC and SS to switch from a run chart with spec limits to a control chart when the control limits are greater than the spec limits (As the process bounces up and down, often crossing the specs but staying inside the controls, they are going to react to the out of spec conditions. I’m having a hard time to with the value proposition of switching to control charts)

    0
    #136770

    Anonymous
    Participant

    “Instead of wasting your time on conventional statistics…”
    You’re joking, right?
    Shainin = “DOE for Dummies”…take it from someone who wasted their time with it for a while…

    0
    #136773

    anonymous1
    Participant

    Shaininites never joke about their beliefs.  They believe in the divinely inspired Word of Shainin and therefore all other methods are a waste of time.
     

    0
    #136907

    AggieIE
    Participant

    Exactly!  You have demonstrated the variability in the process exceeds the specs.  As you eliminate the causes of variability, you will see the control limits shrink until they are within spec.  Remember to recalculate control limits only when the process has improved.

    0
    #136916

    Geri Malott
    Participant

    Send me your email and I will send you a presentation on SPC for 1st time users.

    0
    #137017

    AJ
    Participant

    Dear KKn,
    Your main aim is to reduce the variability. Please keep the focus on that. There are a lots of tools that will help you do that. One of them is Taguchi Online Control. Please provide me the process details and I will help you in applying the same.
    Taguchi Online control creates a ready reference for everyone to reduce the variation in real time.
    Regards
    AJ
     

    0
    #137019

    Anonymous
    Guest

    AJ,
    To date, as far as I’m aware, you’re the first person in this forum to mention Taguchi Online Control.
    Could you explain the difference Taguchi On line control between the conventional method. (I’m sure many people are familiar with Offline, robust design.)
    Cheers,
    Andy

    0
    #137052

    MrKid
    Participant

    If you are reacting to an out-of-spec result that is within control limits (I assume that when you say reacting, you are talking about making some process adjustment) then you are treating common cause variation as special cause variation.  The danger is in overadjustment.  If your output is below spec and you make some adjustment to bring it up, due to common cause variation you will eventually find yourself above the upper spec (and wanting to make the adjustment back to where you were before). 
    Are you really gaining anything by making these adjustments?  Is the process just as likely to come back into spec on its own without making the adjustment?  If you are making all of these adjustments, is the process really stable?  Do you know what the common cause variation is with the process if you don’t make any adjustments?  We have had that experience, where we didn’t really know what the common cause variation was because a special cause (operator adjusts the machine) was interfering with the data.
    Is there an understanding of the critical inputs and how they affect the outputs?  Are they being addressed?
     

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

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