iSixSigma

Attribute Charts

Six Sigma – iSixSigma Forums Old Forums General Attribute Charts

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • #40117

    Tierradentro
    Participant

    Hi All,
    i’m new to this site so please be patient. 
    My question is regards Attribute charting, i am the first to admit its not my strongest subject. 
    My company manufacture “cans” for the packaging industry, approx 4.5 million per day.  The can has a label printed to it.  Prior to the label the can is bare metal.  This “can” can have a number of different defects on it.  Currently we have QC’s who 3 times per 12 hour shift randomly look at cans for defects and notify the area responisible accordingly. 
    This does not tell me whether we are in control, does not show me trends or whether the “special” cause defects are being resolved.  Also there is no traceability.
    So my questions are:
    how frequently should checks be done?
    how many cans should be inspected?
    what sort of chart should be used?
    Any help is appreciated
    John

    0
    #123589

    Ken Feldman
    Participant

    You are correct in noting that the current system provides very little useful information. The appropriate chart to use and the frequency of sampling depends on the answers to a few questions. First, are you interested in the different types of defects or merely that the can is defective? Will a Pareto Chart be useful in classifying the different types of defects? If defects you can use a c or u chart. If defectives you can use a p or np chart. Defect/defective charts will require sampling a larger number than if you were doing dimensional checks. How frequently you sample will depend on how quickly the system can degrade. Does it take an hour for noticeable change or 1 minute? Here is a simple approach as a starting point. Much more info is required before providing a definitive answer.
    1. Every two hours, grab 50 cans. Inspect and record the total number of defects and then record the number of specific type of defect (use this later for a Pareto Chart). Also determine how many of the 50 had one or more defects and call that a defective can.
    2. For fun, start a np chart using the total number of bad cans out of the 50 (regardless of how many defects it has)defined as a can with one or more defects.
    3. Start a c chart using the total number of defects (all types)for the 50 cans. Try this for a week and see if it tells you anything.

    0
    #123590

    Tierradentro
    Participant

    Thanks for that Darth, i will give it a go and feed back the results.
    john

    0
    #123592

    Jaybee
    Participant

    Hi John,
    Number of different defects on an individual can = U chart, this is an attribute chart for poisson data (no. of defects per sample)
    i.e 1 can = 3 indent/ 2 scratches etc.
    P chart is binomial attribute (good can / bad can) i.e. 125 checked for defective cans, 22 ng from your first inspector(cant be bothered to work out the Proportion defective for that, but its looks about 18%), then the next inspector goes & checks 200 cans (he’s not as lazy as the 1st guy) & gets 17 defective & so on for each inspector, then chart as P.
    (the control limits will automatically be worked out for all charts using Minitab)
    NP = same principal, but constant sample size – 3 QC inspectors check exactly 100 (or whatever) cans for defective ones.(the lazy guy has to check the same amount as the rest of the inspectors)
    you mention that the QC ‘notify the area responsible’ for creating the defective cans, so you know which area are causing these defects?
    you need to find out roughly what % failure you have on your cans before thinking about how many cans should be inspected using a sampling plan, but i guess that it will be a real significant amount.
    If your cans go through various process steps, you might want to think about rolled throughput yield & Defect Per Million Opportunities – just a thought. (there are some real good links on this site for more info on those 2 points)
    you could even Pareto by defect i.e. dents/scratches or whatever defect you are seeing.
    Maybe best for you to put a proportion defect on the forum for more significant feedback regarding sampling plan
    I am also worried about traceability? – do you mean the cans themselves have no traceability? or the defects.
    post more info, as i know Darth has a super dooper easy to use sampling plan…….
    Jaybee (formerly known as another John)

    0
    #123593

    Ray The real deal
    Participant

    John,BTW, make sure you perform the Attribute Rnr since you’re at it. Good Luck.

    0
    #123594

    Jaybee
    Participant

    John / Darth,
    I was writing the last post as Darth flew in with his reply! very similar (& i like the idea of the np for a week) apart from i forgot to mention c chart……tut tut, having one of them bad hair days
    best of luck in your trial…
    Jaybee

    0
    #123604

    Ken Feldman
    Participant

    That’s why I am Dr. Darth and you are merely “John with a bunch of screen names”.  I like the fact that you are enthusiastic and a contributor rather than a Lurker or Cheapshot Expert.  Keep ’em coming.  And, at least you have enough hair to have a bad hair day.  Note, I can’t have a bad hair day anymore.

    0
    #123605

    Dayton
    Member

    John, et al
     
    At the risk of appearing a lurker or a cheap shot artist, a few of us do regret the several weeks of painful and patient instruction on our part that it took to get Darth up to speed on posting a picture.  
     
    I had a feeling that it was one of those things that would eventually come home to roost.   
     Vinny

    0
    #123606

    Ken Feldman
    Participant

    Dr. Vinny,
    I hoped you have recovered from your mathematical hiatus and are ready to post more accurate computations. Babe is out there waiting for you. On the aforementioned topic of posting, please be aware that copying and posting directly from the Web was never an issue. My problem then and still now, is how to copy and paste personal pics directly from my C drive. Did I miss that training class or has that little problem not been solved yet? Thanks in advance.

    0
    #123607

    Dayton
    Member

    Dr. Darth,
     
    At the risk of actually offending you, I was referring to those many days of your trying to post any type of a picture, graphic, caricature, or cartoon – regardless of source.   Many of us sort-of-old-timers on the forum remember the painful attempts on your part to post other than blank spots, followed-up with comments like “Darn, what’s the deal???”  
     
    Thank you for your concern regarding my ongoing struggle with applied mathematics.  I am feeling much better now and promise in the future to re-read and recalculate before posting.  It’s just so difficult and a bit of a backwards process to post and then re-read and try to recover the posting from the non-relenting ethersphere.   But I sense that you know of what I speak/type there. 
     
     Vinny

    0
Viewing 10 posts - 1 through 10 (of 10 total)

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