iSixSigma

SPC Coaching

Six Sigma – iSixSigma Forums Old Forums General SPC Coaching

Viewing 23 posts - 1 through 23 (of 23 total)
  • Author
    Posts
  • #38700

    aa
    Participant

    I am in charge of implementing an SPC sysetm and so have to give a brief overview  to the operators. One of the questions which has already come up is this;
    Q. We are making 20 parts per half hour out of which we measure three. in other words our subgroup size is 3 and we are making 20 every half hour which we inspect. We are constructing XbarR charts here. Since we supply parts to the medical profession, and we measure only 3 out of every 20, we are bound to miss some bad ones. How can that be allowed in a medical profession?
    I can statistically prove it to them but is there a way i can explain the answer in laymans terms so all can understand?
    Thanks in advance

    0
    #116265

    Dog Sxxt
    Participant

    Shift control limit to 2-sigma instead of 3-sigma.

    0
    #116268

    Andy P
    Participant

    If your process is in control & capable, why would you want to 100% inspect anyway?
    100% inspection is only about 80% effective in the long term.
     

    0
    #116283

    Schuette
    Participant

    Yo Dog – aren’t you confusing spec limits with control limits?  Why would you say switch to 2-sigma limits, when he hasn’t given any statement regarding current capability?  He may already be at 6-sigma.

    0
    #116287

    Thai
    Participant

    Read Jim’s response…Don’t confuse SPC and acceptance sampling.  SPC is based on the voice of the process, not the customer meaning that the control limits are set on historical process variation.  As long as you have a “capable” process with a Cpk>1.33 and your process remains in control, then there will be a very low probability of producing out of spec products. 

    0
    #116290

    Dog Sxxt
    Participant

    When I mention spec. limit in my posting? Put these words into your mouth.
    Control limit is nothing to with your 6-sigma capability. You are the one confused with spec and control limit.
     

    0
    #116291

    Dog Sxxt
    Participant

    Learn more what is power and sensitivity of a control chart. Learn the basic first!

    0
    #116292

    Dog Sxxt
    Participant

    Control limit is nothing to do with your 6-sigma capability. You are the one confused with spec and control limit.

    0
    #116293

    Thai
    Participant

    I agree that by changing the control limits from 3 to 2 sigma will cause more “out of control” parts to be captured and may require the operator to correct the process more, but this could cause over control resulting in more process variation that if the process was allowed to run in control (against control limits of 3 stdev).
    My understanding of the original question was to do with capturing out of spec products which is more of a sampling plan question, not a SPC question.
    If I am incorrect, please enlighten…

    0
    #116295

    Dog Sxxt
    Participant

    How can Xbar-R chart as mentioned by the original poaster can capture out of spec product?
    If your product is critical to human life, increase beta risk and  provide more protection to your customer is a practical and viable option.
    Which guru told you setting the control limit to 2-sigma will increase your process variation? It has nothing to do with “tempering” kinda stuff. 
     

    0
    #116296

    Schuette
    Participant

    Down doggy,down!  Didn’t you read the question? He was asking about the control chart and how it relates to good and bad parts – which it doesn’t (in control).  You responded by saying “switch to 2-sigma limits”.  What does that have to do with good/bad parts doggy? He wants to show his operators how he can sample production and still keep bad parts from getting to the customer – that requires knowledge of the specification doggy.

    0
    #116298

    Thai
    Participant

    Again, you are bringing specifications into SPC which is not correct…that being said, if the process does not show good capability where the spec limits fall within 3-sigma, then I agree with what you are saying, however if the process shows good capability (and if it is critical to human life I hope it does show good capability otherwise I hope they proof test 100%) then tightening the control limits to 2-sigma will do nothing for you.
    As far as over control is concerned…by definition, there is a 4.6% probability of a part falling outside the 2-sigma limit with a STABLE process.  If you treat any point that falls outside a 2-sigma limit as being ‘out of control’ and therefore would require an adjustment be made to the process eventhough it may be due to normal process variation.  This can cause over control in some cases.
    Kirk

    0
    #116299

    Dog Sxxt
    Participant

    You are totally clueless about how to re-set control limit, but still want to show off  like a guru.
    A famous slogan for you, if you cannot attack a person’s idea, then the best thing you can do is personal attack the person.  

    0
    #116300

    Dog Sxxt
    Participant

    “Again, you are bringing specifications into SPC which is not correct..”
    Amazing, these people tried hard to put something not from me into my mouth.  When I relates specification into SPC?
    I have to surrender because no way I can continue this nonsense debate.
     

    0
    #116301

    Thai
    Participant

    Have a good weekend…

    0
    #116308

    Ken Feldman
    Participant

    This has been a confusing thread to me.  The original post said, “We are making 20 parts per half hour out of which we measure three. in other words our subgroup size is 3 and we are making 20 every half hour which we inspect.” 
    My first question is what is the difference between measuring three and inspecting 20?  What do you measure that is different than what you inspect?  The poster needs to explain this first before we start arguing about control limits and specs and 2 s.d. limits.
    We all know that inspecting for compliance is different than sampling for process monitoring.  I have no idea, based on the post, what this guy is doing.

    0
    #116309

    Adam Bowden
    Participant

    Darth,
    I think what they are getting at is that they are using go/ nogo gauges for the 20 for acceptance purposes but measure the 3 parts so that they can adjust the machine / tools – well that was my experience when I used to set up & run production machines.
    Best regards,
    Adam

    0
    #116317

    Ken Feldman
    Participant

    Hey Adam, hope all is well.  If that is what the poster is doing….. In lieu of Dog’s suggestion to increase Type 1 error by going to 2 s.d., why not just increase the sample size a little, to say, 5 and pick up a little more comfort with being able to make judgements about the process?

    0
    #116321

    Plasic
    Participant

    This the original poster said their product has something to do with the medical field, maybe they should contact the HMO’s and ask them.  They seem to know all the answers anyway.  Of course medical devices could mean “Bedpans”.
    Plastic

    0
    #116757

    Sujit R. Kapse
    Member

    Your question is very valid to convience in layman langauge to your customer. Look 100 % is also not effective in long term that you have convience and conduct the Hypothesis testing with the 99 % Confidence limit in place of 95 % and explain to the customer in suitable langauge which he can understand better.
    I hope with this method you can convience to your customer.

    0
    #116840

    Mike Carnell
    Participant

    AA,
    There is absolutely nothing in a control chart that will guarantee you ship good product. You can be “in control” and ship 100% bad product (your control limits can be located outside the spec limits). There is nothing that makes your control limits occur inside the specification. You can also have a distribution so tight inside the spec limits you could have a >6 sigma process.
    One of your earlier post discussed using Capability Analysis to explain the probability of defects that is probably the easiest way to explain it. If you are using Minitab the numbers that occur outside the spect limits are at the bottom of the page. If your people believe that 100% inspection (which is actually sorting not inspection) will do better run a Attribute gage study with them.
    Do yourself a favor – let go of the medical device stuff. The aircraft engine guys have a similar issue and so do the nuclear energy manufactureres (anyone for some defective fuel rods?) how about people who build bridges, cars, bomb fuses, etc. There are a lot of industries that have life threatening effects from defects.
    Just my opinion.
    Good luck

    0
    #116859

    Sinnicks
    Participant

    Something else you might consider, putting control charts on your inputs, all that Y is a function of X stuff.

    0
    #117148

    marklamfu
    Participant

    Hi:
    I believe we need to know the process capability of the control chart establishment.
    If the control chart is generated based on a capable process (e.g. Cpk>1.33), the in-control plot-point can ensure good shipment, because the possible escaped rate under this situation is less than the escaped rate of performing normal sample plan inspection(e.g. mill-std-105).
    If the control chart is generated under a incapable process (e.g. CKP<1) , It is difficulty to explain the question, at that time, "in-control chart" only state the process is stable, but, can not ensure outgoing product quality.
    Thanks!
     

    0
Viewing 23 posts - 1 through 23 (of 23 total)

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