iSixSigma

Frequency of quality checks….

Six Sigma – iSixSigma Forums Old Forums General Frequency of quality checks….

Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
    Posts
  • #41438

    Da Agent
    Participant

    Does anyone have any good information on what is the best way to setup the frequency of checks in a production environment given different targets per value stream?
    For instance where I currently work we ask our operators to do a quality inspection every 15 minutes. This number was used based on a containment plan versus some statistical test that showed how many parts should we truly sample over an 8 hour work day.
    Any thoughts?

    0
    #130087

    crobertsjr
    Participant

    It may help for you to get a couple of basic stats textbooks. I am not the best to explain the numbers. I focus on the softside of human activity systems. The number to be sampled comes from calcuations based on the number of items produced durring a stated time. The interval for inspection is random and not specific in ‘normal’ distirbutions.
    One of the easier books that I have read is Jaeger, R. M. (1990). Statistics: A spectator sport. Sage Publication, Inc., Thousand Oaks, CA . Another good text that is a more recent publication date is Triola, M. F. (2004). Elementary Statistics (9th ed.). Addison-Wesley: Boston. The second one is easy to follow stats textbook.
    The normal distribution ‘or bell curve’ is based on a representative sample of a given population. If the popoulation has normal characteristics, then the distribution of the presence or absence of a defined identifier will fall into a bell curve.
    Regards, Curtis

    0
    #130089

    Sami Lehtonen
    Member

    Hi,
    how have you calculated how much costs quality control? Now you are checking quality every 15 minutes so there is possibility to produce that 15 minutes out of specs between quality check. How much this 15 minutes out of specs can costs and how much costs if you are checking quality for example every 5 minutes? Also, I have noticed in my projects calculating for example control and specs limits, that there is no short cut, you have to check and measure every product comes out line, when calculating limits. It is very important to understand the difference between within and overall stdev and after you have measured “natural changing”, how the process varietes, it will be much more easier to find out checking intervall. Hope this help something and sorry about my English :)
    BW, Sami Lehtonen

    0
    #130090

    Sami Lehtonen
    Member

    And little add forgot my previous post, how much is DPMO and CP/PP now? In which intervalls failures/out of specs comes and how much failures have maximaly and minimaly been in one intervall?
    BW, Sami Lehtonen

    0
    #130139

    Jonathon Andell
    Participant

    SInce your process is unique, we cannot just give you a number. If you want an intensely technical treatment, Deming’s book “Out of the Crisis” has a chapter devoted to the topic. In principle, you want to sample as often as it takes to “stay ahead” of time-related sources of variation, but no more.

    0
    #130163

    R.M.Parkhi
    Participant

    Dear Sir,
    I suggest that using DOE [ either Shainn-Bhote technique or Taguchi ],
    you improve CPk value of the process beyond 1.75.Then tell your operator to fill the precontrol chart once an hour.The details of the run chart are given in book ‘World Class Quality’ by Mr. Keki Bhote Publishers American management Association. As you improve process capability, you can further increase the interval. However control of process parameters is a must. This is called PosiTrol [Positive Control ].
    Pl feel free to ask, if you have any query.
    With regards,
    R.M.Parkhi

    0
    #130171

    mark lamfu
    Participant

    The frequency of quality checking should consider the process capability( CPK, Yield, etc.), for capable process, the intervel time between 2 checkings amy more long, otherwise, the intervel should be short.
    In addition, the output qty. of production line per hour also should be considered, when you perform the checking, basic output qty. is necessary. if the output is too less, the sampling result is not accuacy
    In fact, I suggest you to study P- chart theory, perhaps, you have new ideas

    0
Viewing 7 posts - 1 through 7 (of 7 total)

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