iSixSigma

FPY for iterative processes

Six Sigma – iSixSigma Forums Old Forums General FPY for iterative processes

Viewing 17 posts - 1 through 17 (of 17 total)
  • Author
    Posts
  • #52144

    MCorbo
    Participant

    Does anyone have any insight on how to calculate first pass yield (FPY) for an iterative process, that will also be compatible with non-iterative processes? I am trying to define and calculate the FPY for some of our optical polishing operations – the current process involves polishing then testing the optic until it meets spec. In many instances, it will take multiple polishing runs to complete the optic, therefore it is unreasonable to expect it to meet spec on the “first pass” of the polishing process. Any suggestions or feedback would be appreciated.

    0
    #183007

    GB
    Participant

    Testing until good is a common ailment.   In order to achieve true FPY, Having worked in this industry, we ended up leanflow redesign of the process, to include a total redesign of the polishing process, so that the iterative was mitigated. and specs could be met per run.
    If you cannot get to this point, you may want to quantify “internal specs” for each sub-tier process, in order to make your measures meaningful.   This is not a waste-free process though.
    Until you deal with the “test until good” aspect, your RPY calcs will be meaningless.

    0
    #183024

    Taylor
    Participant

    Although I agree completly with hbgb on this one, I have seen similar iterative process’s calculated such that if at least every part took a certain number of runs to be good, then this would be the benchmark starting point for FPY. I don’t like it, but it does offer a starting point and can at least generate a number to see when the process is changing.
     

    0
    #183026

    GB
    Participant

    Chad, I’ve seen that approach as well…it gives me the heebeegeebees
    too much room for variation and hidden factory, not to mention the waste involved.

    0
    #183029

    MCorbo
    Participant

    I am confident that using a standard number of runs for each part / process will not work for me. I have been thinking of defining a minimum amount of convergence of each run. If a run results in a part that did not meet this minimum convergence (i.e. the part isn’t getting better fast enough) it fails the FPY. I won’t care if a part takes 1 run today and the next part takes runs, as long as a part is getting better with each run. What are your thoughts on that line of thinking?

    0
    #183041

    Taylor
    Participant

    I totally agree, just as long as you know what to do with the data.
    HBGB, totally agree.

    0
    #183044

    Gary Cone
    Participant

    I think as a containment that is okay. Go tell people you will do this
    while working a project to fix it.Take HeeBee’s advice – just because tribal knowledge says the
    process must run this way does not mean that it is correct.I have seen dozens of processes like this that can be radically
    improved to achieve desired results on the first pass.

    0
    #183072

    pise
    Participant

    Is it possible to go to the underlying physics of the optical polishing process?
    What combination of the polishing material and the surface to be polished gives the result with the various process parameters? If we understand the physics and can develop a matrix, we may be closer to reducing this iteration.
    Raj

    0
    #183074

    riyan
    Member

    the idea to develop FPY is to have baseline of process, such that when we apply improvement on the process we will understand the impact of improvement. Iterative process is common in manufacture environment, to set FPY we can use existing process as baseline by establish time required/ iteration required to achieve good unit based on specification. If the number/time of iteration required is less than standard we can call it FPY and vice versa. Once, the baseline data is set, then we could apply six sigma tools to improve the process and quantify the improvement. In future, if FPY reach 100%, we can set new baseline of FPY, that’s why it is called continuous improvement…

    0
    #183076

    Fontanilla
    Participant

    I totally disagree with your statement, “I won’t care if a part takes 1 run today and the next part takes runs, as long as a part is getting better with each run.”  That, my friend, is a wasteful process.  I’ve seen projects where people thought exactly the same thing.  In one case, we had product that sometimes “came in” in 2 tests, sometimes in 60.  By carefully rooting out the key inputs that were varying (widely) in the process, we were able to take this down to essentially first pass, good part (although the really smart guys in sales and marketing told the customer we would ensure 3 “break in” cycles before testing so now we are locked into 4 cycles.  Geniuses.)
    You need to follow your DMAIC steps, paying close attention under “Analyze” and find which inputs are varying causing your output to vary.  Which X’s drive your Y?  There is a very scientific reason why one piece takes 1 pass and others take multiples.  Understand the physics, understand the mechanisms, and you will reduce your variation.
    Oh and one other thing, I’ve assumed your measurement system passes muster.  If not, don’t neglect “M.”
    Good luck.

    0
    #183083

    Jer
    Participant

    As an out side observer to this conversation, and not directly your question, sounds like a great opportunity for a well planned DOE which should result in increased process stability, specs and easier final quality measurement.

    0
    #183084

    Mikel
    Member

    I have a hammer, this looks like a nail.

    0
    #183100

    MCorbo
    Participant

    Ok Dan, I understand your point about our “wasteful processes” and HBGB’s original comment about “tribal knowledge”. Concurrently with this FPY project I am trying to debunk the belief that optics cannot be finished in one pass. But until that day comes I have to work with what I’ve got.Currently, some of the iterative processes we do to bring our optics into spec have many hundreds of competing variables, only a few of which we currently control (to exhibit control over the others requires more money, time and research than we cab afford to spend at this time). Phase 1 of my project is to define FPY within our manufacturing organization and phase 2 will be to improve upon our processes’ yield. Concurrently, I am looking to make other improvements to many of our processes throughout our facility.One example of an iterative process that we are struggling with is the hand-lapping of eight datum surface on a reticle handler. Currently, it takes our optician approximately 7 (+/-2) iterations of hand lapping and metrology to meet the spec. The problem with this feature (or perhaps the process) is that if the optician were to overshoot any one of eight datum surfaces the entire $25,000 part would be scrap. Therefore the optician takes an iterative approach to gradually reach the target without overshooting. In this case, the fewer the iterations performed, the greater the risk to the part. In this particular case I don’t want to (notice I didn’t say “I can’t”) define the “first pass” as a specific number of iterations, giving the optician incentive to put the part at a higher level of risk.Thank you all for your input on this matter.

    0
    #183105

    MrMHead
    Participant

    Does each piece have a different spec to meet?  Or is it a mass production process?
    If it’s different specs, then I can see this being difficult to pin down FPY if you have to change your settings for each run.
    If it is mass production, then your settings shouldn’t have to change from piece to piece.  Do you have a run chart on # of iterations per piece?  Is “tool wear” at play here?
    Do you run all pieces once, measure, and run a second time for those that need it?
    As mentioned in another response, this sounds like a good opprtunity for a DMAIC that includes a DOE.
    The “Test Until Good” didn’t work so well for Peanut Corp of Am. (though they didn’t modify the product between tests, they were just looking for a “Good”)

    0
    #183177

    Avijit Ghosh
    Participant

    What,I can make it from ur communication is the method for calculating the FPY for a coating process.
    The process of Six Sigma talks about first time right of whatever u do.The second & the simultaneous process done is a waste & the variation needs to be reduced.
    It would be FPY= (Total passed first time / Total Input) &
    if u have some simultaneous process,this multiplication continues.
    (Process A x Process B x Process C) = FPY
     
     

    0
    #183184

    MCorbo
    Participant

    MrMHead, Most of the work we do is not mass production. We frequently make small batches of lenses or mirrors for a customer (qty: 5-20). From project to project the specs can vary greatly causing us to slightly modify our “processes”. I use quotes because our process is never truly defined and can change from optic to optic. The set of conditions that produced a great optic today may not work as well tomorrow — yes, I understand this means we don’t have control over our processes and I am trying to deal with that concurrently. Right now, we perform a lot of hand polishing operations which seem to be VERY technique dependent. One optician may be able to quickly converge upon a given spec while another optician may take a little longer and a third optician may downgrade the optic all while performing the same “process”.Typically, we do not do much batch processing; the optics proceed through each process one or two at a time. In some cases multiple optics can be at different stages within the same process and are handled by different opticians.I agree that this sounds like a good opportunity for DMAIC, we just don’t have a lot of experience with six sigma and I currently don’t have the support from management. I am trying to define a FPY metric that can be applied throughout our manufacturing facility that accurately reflects what we do.

    0
    #183186

    Taylor
    Participant

    Mcorbo
    I sense a little frustration in your last post. I fully understand you wanting to use FPY as a metric, and with good cause. I don’t have simple answer for you moving forward. Instead I want to offer some advice that may help just a little.
    OK, this is what I read that caught my eye: I use quotes because our process is never truly defined and can change from optic to optic. The set of conditions that produced a great optic today may not work as well tomorrow . Now I know you said you don’t have control over you process,  For one to have control over the process and be able to make meaningful adjustments to the process, one must first Understand the process.  This what I think you should do, and maybe you have already done this exercise, But I think you should try to perform a Root Cause analysis of why this particular set up will not work today, when it worked yesterday. There has to be an underlying issue that is to the heart of the FPY. Understanding the process in its intirity is a must. You can gather FPY data all day long and it will not help you understand the process.
    Just My Opinion
     

    0
Viewing 17 posts - 1 through 17 (of 17 total)

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