iSixSigma

Understanding variation

Six Sigma – iSixSigma Forums Old Forums General Understanding variation

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • #31331

    Sakiwa
    Participant

    Greetings,
    Can some one help !
    We are having difficulty in measuring variation within our process.Our process is not a simple one. It is a Semi -continous liquid  process with appox. 8 sub processes each with different optimal rates.At the monment we are baselining. Obviously we are collecting data around wastage, operator actvities, machine/ process observations, some critical parameters in certain sub-processes, etc. We trying to paint a picture of the whole process. But unfortunatley nothing is showing on variation.for example. How does sub-process B have an effect on Sub-process D ?
    One of the outputs we want is the impact of this variation to the process. Also want to tease out why is it bad for us & diffcult to quantifiy.
    Regards.
    Saleem (BlackBert)
     
     

    0
    #82464

    Summerfield
    Participant

    A well planned DOE seems to be the answer

    0
    #82466

    A.B
    Participant

    Mohammed ;
    You should be very careful in planning for the experiment. Looks like you’re adjusting levels so to get at the end the results that you really want to get and that’s may be the reason for the failure of your experiments.
    This case is may be typical of mixture experiments.
    What are the sub processes ? Have you defined them ? Which factors are you considering ? How do you define optimal rates ? on which basis ?. If possible start the experiment with 2 sub processes , then 3….etc , then you can set a strategy to conduct experiment with 8 sub process.
    Good Luck.
    A.B
     

    0
    #82474

    Chad
    Participant

    Saleem,
    You have a couple of good suggestions from the other postings.  I would recommend that before you do either of those recommendations, you do a Gage R&R (if you haven’t) to make sure that you are able to measure the variation that you believe exists.
    Good luck,
    Chad

    0
    #82488

    question
    Participant

    Hi Mohammed,
    What is your final product ? “Semi-continous liquid” sounds interesting.
     

    0
    #82490

    Mohammed Saleem
    Participant

    Thanks for the relies!
    Here is some additional information
    Our final product is FOOD powdered Product.
    Process follows from mashing of cereals (biological process), separation, liquid additions, drying at various stages, Cake formation & finally Milling into powder.
    Currently I have recommended for each sub-process generate IPO’s & understand how can we measure some of the outputs. I will also look at process variability on the outputs & quantify the variation ! ( not easy)
    Difficulty will be to understand the size of opportunity. for example say we are running our mashing process at a particular rate, what does that mean for the separation process. Currently operators say no problem. But guess what we have to slow down our drying process.
    Additional work is being carried out on bottleneck Analysis, Capacity constraints etc & operator observations etc.
    At the moment I think I will not able to do a DOE – I don’t have my factors & response variables. Not happy with the way we measure them & not if we can measure them accurately.
    All Thanks for the help – Will keep in touch
    M.Saleem

    0
    #82500

    Erik L
    Participant

    Mohammed,
    The best way to parse a process into its many sources of variation would be through a Components of Variation (COV) study.  If you can stage the factors so that they are all nested within each other, it’s fairly simple to have Minitab calcualte the % impact of those various factors against the total variation of the system.  If you have a mix of crossed and nested factors, a Factor Relationship Diagram (FRD) helps quite well here, you’ll need to use GLM as the analysis technique and it starts to get a little tricky to ensure you’ve specified the correct model for Minitab to analyze.  If set up correctly, the base of the COV analysis can be set up so that you will in effect perform a Gage R&R analysis. 
    Regards,
    Erik

    0
    #82614

    Karel Burgoyne
    Participant

    Hi
    sounds a nice problem to have to work on.
    For me I’d start with the data I have available, and where I believed it was reliable, simply put it in to a historical DOE. Select an appropriate inputs model I think you said 8 inputs, and for many software systems you can live with some 10-15% of the data missing if you don’t have everything available. (well mine certainly does)
    See what that hints at, in particular the possible interactions, and of course the inputs that appear to influence both the final value and the variation around this value.
    Then you, or someone with an understanding of the process, make a more considered judgement on does what this data try to tell us, and ask -does this really make sense what the model predicts.
    Armed with this data you are in better position to look at then doing some trials to more rigourously verify the model.
    Karel

    0
    #82619

    Dr. Steve W.
    Participant

    what is the problem statement? What is it that bothers you so bad that you want to do a 6-sigma project on it? Capacity, cost, inventory or defect rate? What is the metric? If you can answer these questions, I think we have got a starting point. You do not need DOE to baseline. DOE is for optimization purpose.
     

    0
    #82630

    Hemanth
    Participant

    Hi
    I agree with Dr. Steve. Dont bother about data collection at this stage. Talk to your final customer? In your case it could be your dealer chain or retailers. Identify what’s bothering your customer? Is it Cost, Quality or delivery of product? Map your current process. Identify a broad metric for each CTS characteristic like % rejection or first pass yield for Quality, Through put rate for delivery. Keep it simple.
    Develop measurement system for measuring each characteristic (if you dont have one..) and start monitoring.
    Once you know that its either Cost, Quality or delivery then plan ahead. Then identify and prioritise projects..this will give a long lasting and immediate results.
    Another point which I could make out from your message is that you dont
    Would be happy to help you..through this forum..
    Hemanth

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

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