Dismiss Notice
You must be a registered member in order to post messages and view/download attached files in this forum.
Click here to register.

Judging supplier MSA

Discussion in 'Gage R&R and MSA - Measurement Systems Analysis' started by AxelR, Nov 26, 2019.

  1. AxelR

    AxelR New Member

    Joined:
    Jan 16, 2018
    Messages:
    2
    Likes Received:
    0
    Trophy Points:
    1
    Location:
    The Wetlands
    Hi Everyone,

    We ask for MSA's / PPAP's of our suppliers because our own customers insist on receiving the files for their administration. I have a fourth edition MSA book, but i'm hesitant to simply follow up on the information in the book.

    Can someone explain to me what key factors are the most important when it comes to judging a supplier's MSA?

    See the picture below, which we could use as an example.

    upload_2019-11-26_14-15-16.png
     
  2. Gejmet

    Gejmet Member

    Joined:
    Jul 23, 2019
    Messages:
    8
    Likes Received:
    3
    Trophy Points:
    2
    I am sure that you are aware that in order to answer this question will require your organisation to adequately understand their own requirements from measurement processes. It would be very tempting and easy to pick a few highlights from the book you quoted and throw them back at you but in the long term it is always better to get back to first principles and be able to stand behind statements.

    What I am about to quote inevitably relies on a level of understanding of Measurement Systems Analysis, some of it you may get some detail of from the book you quoted and some elsewhere, here goes;

    Firstly, all measurement study data needs to display consistency, without this you don't have measurements you have a random number generator. For more detail please read EMP III by Donald J Wheeler or go to the reading room on his website www.spcpress.com

    Assuming that the study has been set up properly, depending on what you want to uncover you should be able to extract some behaviour from the graph or graphs. For an EMP Study (see ref above) you will be able to answer most pressing questions about the process by just interpreting the graphs.

    If you have to follow the requirements of the book you have quoted, usually the vanilla statements required are;

    - % EV - Referring to percentage of Equipment Variation consumed from the specification
    - % AV - Referring to percentage of Appraiser Variation consumed from the specification
    - % R&r - Referring to a combination of EV and AV Variation consumed from the specification
    -ndc - Referring to the number of distinct categories a product can be divided into

    For general target numbers for each of these please see the book or ask your customer.

    Please be reminded that there are issues with all these numbers from how they are calculated, what they are trying to describe and the targets given but there just isn't enough room here to get into this, a taster is available on Donald J Wheeler's website www.spcpress.com from a paper entitled "An Honest Gage R&R Study".

    I hope this helps you.
     
  3. AxelR

    AxelR New Member

    Joined:
    Jan 16, 2018
    Messages:
    2
    Likes Received:
    0
    Trophy Points:
    1
    Location:
    The Wetlands
    Hi Gejmet, thank you for your reply.

    I'm aware of the fact that its not simply a number crunching method. The literature relates that any outcome of variation below 10% would or could be considered to be 'stable'. The graphs appear to fall within tolerance limits. Unfortunately I can't add the entire document here as it includes supplier specifics. I'm currently reading on the link you provided. I'll try to find the EMP III, by Wheeler and report back.
     
  4. Gejmet

    Gejmet Member

    Joined:
    Jul 23, 2019
    Messages:
    8
    Likes Received:
    3
    Trophy Points:
    2
    Hi AxelR,

    Yes, the common approach is to apply percentages as a form of an insurance policy to provide confidence for the ability of the measurement process to sentence parts, there are many issues with this, not least the fact that the method over estimates the effect of measurement variation. For some people this is comforting and of course if the percentage falls within the magical guidelines benign neglect will ensue.
    If the consumption is important to compute then it should at the very least be computed properly and consideration given to guard banding the specifications to allow for measurement variation. If the opportunity is there it's even better to control the manufacturing process so that some elbow room exists, in this way the effects of measurement variation as it interacts with product variation are controlled and monitored.
     
  5. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    457
    Likes Received:
    492
    Trophy Points:
    62
    Location:
    Maine
    The “incomprehensible table of statistical output” is a horrible way to judge any analysis. If you plot your data* you will then have powerful means of analysis (pun intended)


    *Ellis Ott
     

Share This Page