iSixSigma

shainin methodologies

Six Sigma – iSixSigma Forums Old Forums General shainin methodologies

Viewing 40 posts - 1 through 40 (of 40 total)
  • Author
    Posts
  • #31902

    Picklyk
    Participant

    hi guys.  we ship a product to an OEM.  the OEM assemblies it to their product.  there is an issue with the finished assembly.  the OEM has just  conducted a shainin study where they took a BOB (best of the best) part and a WOW (worst of the worst part).  When swapping our part, the BOB turns into a WOW (not as bad, but bad) and the WOW trends toward a BOB.  obviously, this points to our part. 
    the question is, how can i disprove these results?  obviously, the shainin study captures zero interactions, that’s my first course of defense. 
    thanks guys!

    0
    #84560

    Charles H
    Participant

    Jay,There are three options, as I see it.  “Defend” with “data”, hoping to back them down and prove your customer wrong; do nothing and keep quiet, hoping another fire will take this one’s place and everyone will forget about it; or get to the root cause and actually fix the problem.  Two of the could lose your customer for you.  Concentrate on finding the root cause and fixing the problem – if the problem lies with the customer or with you, it’s a win/win. Just my two cents worth.  I must say, I do find your post interesting.Take care,Charles H

    0
    #84561

    Mike Carnell
    Participant

    Jay,
    Why would you set out to prove it is wrong? Wouldn’t it make more sense to understand what they did and why? You may have an opportunity to deliver a better product.
    There is another tool called failure analysis. I would take your failed module and start a failure analysis (since at this point they have more data than you that you delivered a defect) and replicate the first analysis. You will not go to statistical jail for running a Shanin study.
    The technique they used is called “Component Search.” I like it and use it when it is the correct application. It does ignore interactions is taking the good one apart and reassembling it and the bad one apart and reassembling it. No swapping. It is a way of splitting the problem into catagories such as “is it an assembly issue?” or “is it a component (subassembly) issue?”
    That sounds like a hypothesis test used to gain knowledge about the x’s. Sounds like the Analysis Phase. It is the thought process, application and resulting information that define the analysis phase – not the tool used to get it. Most of the hypothesis tests also ignore interactions. The next time someone runs a homgeneity of variance test or a t test or a Chi Square test are you going to try to prove that what they did was incorrect? There are a lot more tools being used out there that ignore interactions than don’t ignore them.
    Having the name Shainin attached to it does not make it incorrect either.
    He used exactly the same technique to fix a problem (in 2 days) that our engineers had struggled with for 19 years.
    Just because someone is doing something outside your frame of reference doesn’t make it wrong. It makes it an opportunity for you to understand what they did. Then you have the opportunity to either teach them something or for you to learn something.
    Good luck. 

    0
    #84587

    Mikel
    Member

    Jay,
    If the problem follows your part, its you part. Fix it.
    Shainin methods are valid.

    0
    #84833

    Fontanilla
    Participant

    Jay,
    The component search method can be very effective in understanding assembly problems – specifically what component/s are responsible. I am an MBB but prefer to use Shainin methods for this purpose. The way to understand whether interactions are present is to increase the sample size from 1. (Typically 5-8) Assuming the failure mode of the WOW parts are all the same, then ask “does the same thing happen with each BOB-WOW pair”.
    Regardless, there IS an issue with your part if the assembly/disassembly stage passes as suggested in an earlier post.
    Generally though your comments are of concern. Why would you want to disprove the customer versus understanding the problem.

    0
    #85078

    Markert
    Participant

    One issue with the so called BOB and WOW strategy is that it ignores the issue of special and common cause variation.  Often the selected BOB or WOW are nothing more than a set of parts that do not differ in distribution; i.e.; there is no special cause that separates their performance.  If your process is stable or approximately stable, then the BOB and WOW are just two random observations from one continuous process (e.g., normal) and as such there is no real difference in their performance. I have seen a number of scenarious where the BOB and WOW strategy led to conclusions that were not reproducible.  For an unstable process the concept might work.  However, I personally think failure analysis and designed experimentation are far more powerful tools — you are correct that the Shainin approach to DOE does not explicitly consider interactions and the interactions can be the major source of variation in many situations.

    0
    #85124

    Tierradentro
    Participant

    Do not waste your time.  Let the parts talk to you (Shainin Method).
    Fix your Parts. 
    Shainin techniques is powerful and faster than other DOE’s.
    Good Luck

    0
    #85131

    Ø6 Sigma BB Coordinator
    Participant

    I have studied Shinin method. I also develope its mothodology myself.
    As my study, I have found that component search does capture interaction (believe me). The way it capture the interaction may be difficult to understand. I explain as the following.
    As the experiment you explained, you part was tested at least 4 runs.
    Let’s define
    A = your part                          B = the rest (all other parts)
    WOW   —–   A = -1      B = -1
    BOB      ——  A = +1     B = +1 
    Then they swapped which equal to 2 more runs as the following :
    A =  – 1   B = +1           and         A= +1       B = -1
    These run is just 2X2 factorial design.  However, B consits of many factors combined into 1 factor.
    When your part shows significant effect.
    It meanst that your part (A) have significant effect or have interaction with other parts. AB interaction means your part have interaction with one of the rest parts but we do not know which part until we continue searching the rest. 
    My conclusion is that component search capture interaction.
    Actually, it can capture 2-way interaction, 3-way interaction or even 4-way interaction.     
    Hope this helps
    Six Sigma Black Belt Coordinator      

    0
    #85133

    Ø6 Sigma BB Coordinator
    Participant

    Sorry.
    I explain wrongly a little bit.
    A = The parts which you and your competitor produce.
    A = -1   is your part.
    A= +1 is your competitor’s part.
    B = -1 are parts from suppliers C,D,E,F …
    B= +1 are parts from suppliers Z,Y,X,V ….
     
    Six Sigma Black Belt Coordinator

    0
    #85135

    vidyut bapat
    Member

    dont defend your parts.
    Talk to your  parts instead (try paired comparison )  and find out what is wrong and fix them.
    your customer will be happy then.
    otherwise he may switchover to your competetion.
    With best wishes,
    Vidyut Bapat

    0
    #85145

    Marc
    Participant

    All –
    I would recommend to those who would like a better understanding of the power of the Shainin techniques (which is specifically to identify the Red X when it is an interaction) that they read Keki Bhote’s World Class Quality, 2nd ed. The Shainin philosophy states that most of what causes variation are interactions and something you would least expect and therefore would not be able to “brainstorm” with any predictable success. As someone who has spent years doing the traditional engineering methods and then switched to the Shainin approach, I can assure you that the Shainin method is light years better. It took me a year to de-tox from the old way and to fully appreciate the Shainin method. In all due respect, I can understand where there would be people who would challenge it. If you really want to solve some chronic quality problems, then learn to talk to the parts instead of the “experts”.
    All best,
    Marc
     

    0
    #85155

    vidyut bapat
    Member

    Marc
    I fully agree and support your views.
    After learning about shainin  methods, our group solved a 15 year old chronic problem in our electronic components manufacturing unit within 3 months. the defect rate reduced from 8000ppm to literally 0 ppm.
    everyone who is interested in rapid improvements should learn and practice shainin.
    Vidyut Bapat

    0
    #85161

    Marc
    Participant

    Vidyut,
    Thanks for your response. I am very excited to hear about the outstanding results you got by taking the Shainin approach. I’ve had the same kind of experience on many high profile projects in which there were high defect and warranty rates, where design engineers tinkered with the problem for years (literally), and then they give up and call me and we solve the problem within a day. It wasn’t because I was clairvoyant: all we did was talk to the parts; it’s that simple. It’s amazing how hard it is to get people to do that. Best of luck in your work.
    Sincerely,
    Marc McKeel

    0
    #85178

    BenR
    Participant

    As a result of the original message in this thread back on April 5, I bought the Bhote book. Had only heard of Shanin in conjunction with the Red X.
    Used the components search technique and in two days helped solve an issue we had been fighting for a while.
    I was probably lucky this first time around, but I have some new tools I intend to use when the situation calls for them.
    BenR

    0
    #85345

    Marc
    Participant

    Ben,
    Great job with the Components Search! It’s exciting when you know you’ve found the number 1 problem (the Red X) and not some tiny, 10% issue that someone’s theory might have offered up. Keep going!
    Best,
    Marc

    0
    #85459

    reinaldo
    Participant

    BOB and WOW are both part of what Dr. Deming call The Heavy Losses.
    They are both lack of constancy of purpose, short-term thinking, failure to optimize trough the time.
    Do the rights things rights from the very begining (long-term planning for a quality product)and control the variation during the production/manufacturing process.
    Uses the Taguchi’s Loss Function to have a robust design!

    0
    #85471

    Mikel
    Member

    Reinaldo,
    Wow, what topic are you addressing?
    BOB means best of best
    WOW means worst of worst

    0
    #86179

    Lee
    Member

    Marc-  How’s it going?  I’ve spent the last three years trying to apply Shainin in service/transactional arenas.  It’s been very helpful to apply the BOB/WOW concept to distributions but it’s been difficult to find a corollary to a components search.  Once a transaction happens, it’s hard to take it apart and put it back together :)  We need to get in touch…I’ve got a ton of stories for you.  Hope everything is going well!
    Thanks,
    Sal (From B&D)

    0
    #86539

    DS
    Participant

    Jay,
    If you haven’t already gotten Keki Bhote’s book “World Class Quality,” that has to be your next step.  Have you actually seen the component search computations and results?  Did your OEM follow the procedure Bhote outlines?  Was the Stage 1 (assembly process) test done correctly?  If not, inferences based on component swapping may be faulty. 
    Assuming Stage 1 was passed, your statements imply that complete reversal of BoB to WoW and vice versa may not have occurred.  If so, there is some other factor involved besides your part.  As Bhote shows, data from this process can be fitted in to a factorial analysis.  The process will reveal interactions if done correctly and the interaction actually exists.
    This process is very reliable if applied correctly.
     
     
     
     

    0
    #87101

    “Ken”
    Participant

    Sal:
     
    I’d also be interested in the application of Shainin methods for business transactions.  Right now I am working on improving throughput for the quoting & order entry process.  Any information, please send my way.
    Thanks.
    knoll(at)cincom.com

    0
    #87110

    Ron
    Member

    Why are they using Shainan?  Use statistically valid techniques contained ion the six sigma body of knowledge..
     
    Shainan is so past century I can’t believe people still refer to it.
     
     

    0
    #87120

    Huh?
    Participant

    Ron,
    Just about all of the statistical methodologies in Six Sigma are as old if not older than Shainin techniques. Based on your logic we better stop using Designed Experiments, Control Charts, etc. What else do you suggest we use if age, by itself, makes tools obsolete? There isn’t anything I been taught in Six Sigma training that I haven’t been aware of and/or used for the past 17 years as a quality engineer.
    I don’t know if you condone and use pre-control (taught in Six Sigma), but guess who was one of the individuals to develop it? Could it be Dorian Shainin?
    All the tools (Shainin, Six Sigma, whatever) are pretty much worthless if your organization has an uniformed and uncommitted management. They love to say how the tools are so great until the data and statistical inferences tell them what they don’t want to hear. (Of course, none of the Executive Staff has attended any training. No need to, since one of them a short article about how Six Sigma will solve everything).
    Waiting to hear your reply.
    Huh?
     

    0
    #87294

    Mike Shea
    Participant

    Ron
    As a person getting involved with Shanin or “Red X” techniques, I would not get defensive about your product or process.  I have used this technique to discover if I have an assembly, supplier or product issue.  This is a very valuable tool, learn to use it.  It really does let the part(s) or process “Talk to you”
    Just my two cents worth.

    0
    #87327

    Jorge Mtz
    Participant

    The whole purpose of a tool is to help you to solve a problem. Your problem !! Doesn’t matter if is an old tool or a brand-new just out of the oven and highly sophisticated weapon. If you understand it and know how to use it, go for it. and if it really help you, then that tool serve its purpose.
    I personally have a great respect for shainin tools, I already use them & got good results. So my recomendation is to learn the methodology and add it to your tool box.

    0
    #88170

    Rick Pastor
    Member

     
    BOB WOW Interactions at What Sacrifice?  The following will show that a BOB WOW experiment “can” and “cannot” provide information on interactions but you have to make statistical sacrifices.  This sacrifices has not been mentioned in previous replies to the original message.  One pair of BOB WOW samples does not allow for an analysis of interaction while more than one allows the experimenter to evaluate interactions.  Dan states that 5 to 6 is a good number.  An example illustrates why one pair does not include interactions and more than 1 pair includes interactions.  At the same time, the example raises another question.  How to perform the experiment with more than one pair?
    BOB WOW Without Interaction:  Let set the experiment up.  We will perform a full factorial experiment:
     Let  A and B be a two factor two level full factorial experiment where
     A = your part    B = the subassembly into which your part fits
    WOW:  A = -1      B = -1                 (A,B)=(-1,-1)
    BOB:    A = +1     B = +1              (A,B)=(+1,+1)
     Assume that the customer selects one good and one bad “fully assembled part” that is there are no repetitions.  The customer decides to use a continuous scale to describe the idea of good (10) and bad (0).  The customer performs the BOB WOW experiment with the following results. 
     Pattern     A                   B                   Outcome
    -+                -1                   1                     3
    +-                1                   -1                     8
    —                 -1                   -1                   0
    ++               1                   1                     10
     
    Running a full factorial using the JMP software package gives
     
    Analysis of Variance
    Source    DF    Sum of Squares    Mean Square     F Ratio
    Model       3      62.750000             20.9167               …….
    Error         0      0.000000               …….                  Prob > F
    C. Total    3       62.750000                                       …… .
     
    By including the interaction A*B in the model
     Outcome = c0 + c1*A + c2*B + c3*A*B + error)
     there are no degrees of freedom for error, . 
     BOB WOW With Interactions: On the other hand if the experiment performed by the customer used two good and two bad fully assembled parts (one repetition) the results might look like the data in the table below.
     Pattern      A                    B                    Outcome
    -+                 -1                    1                      3
    +-                 1                    -1                      8
    —                  -1                    -1                    0
    ++                1                    1                      10
    -+                 -1                    1                     4
    +-                 1                    -1                     7
    —                  -1                    -1                    2
    ++                1                    1                       9
    Running JMP full factorial again gives 
    Analysis of Variance
    Source       DF     Sum of Squares   Mean Square     F Ratio
    Model          3       88.375000            29.4583            33.6667
    Error            4      3.500000               0.8750              Prob > F
    C. Total       7      91.875000                                       0.0027
     
    This time we have enough data that we can begin to evaluate the interactions between factors A and factor B.
     
    What is the Statistical Sacrifice?: In order to calculate averages and standard deviations we select n samples randomly from a population without replacement.  For example, if we want to now the average height of men in Illinois, we do not measure the same man n times and take an average.  In contrast, if the components in a BOB WOW experiment are simply exchanged, the same experimental unit is used in different runs and this does not make for a great DOE — independence. 
     
    Perhaps a better way is to treat the outcomes (-1,+1) and (+1,-1) as a B vs. C experiment.  In this case, we can say that the BOB WOW does not detect interaction.  Also, the experiment requires multiple runs and the customer might want to do the work.
     

    0
    #98035

    Victor Gonzalez
    Member

    Obviously this is a component search experiment .
    You need to know what is the green Y , this means the output or factor they are considering to know wether a part is good or bad.
    They also need to demonstrate the measurement system validation .
    Once this is done, you need to know the decision limits, if your part changes the BOB to WOW completely ( a complete reversal ) you dont have a solid defense.  You need to solve the problem in your part.
    However, if the reversal is not complete, this means there is an interaction and there is another component interacting with yours to cause the problem.
    If the response is not outside the decision limits, your component is not a contributor to the problem .

    0
    #100042

    Marc McKeel
    Participant

    Hi Ron,
    Have you ever actually used the Shainin techniques? In all due respect, most of the Red X critics I’ve met have either never used the methods or have dabbled in them without really using them properly. It seems that if a methodology doesn’t square with a theory or mathematical logic, it tends to be dismissed out of hand. I used to be that way until I actually started using the Red X approach, and now I would never consider an X-to-Y approach over the much more efficient Y-to-X method.
    All best,
    Marc McKeel

    0
    #100066

    Diaconu
    Participant

    Marc,
    As a practical matter my last two 6S projects had to be approached from an Y to x perspective because our tribal knowldedge simply wasn’t sufficient to allow us to differentiate the importance of one x from any other. By starting at the output and examining all of the causes that could contribute to it is not only faster but it allows one to theorize about processes where their true nature is not fully understood. It’s not strictly speaking a 6S approach but it worked.
    It sounds like the Shainin methodolgy is closer to this approach than 6S is. If we had pursued a result from an x to y direction in either of these projects we would probably still be working on them now.
    Once again it’s the right tool for the right job and I will certainly investigate the Shainin methodology further.
    Thanks
    Mia
     
     

    0
    #132417

    R.M.Parkhi
    Participant

    Dear Sir,
    I am following Shainin techniques for last twenty years. They are wonderful, useful for every problem, unlike Taguchi,s cumbersome & suitable for a few cases only.
    Thanks for your views.
    Regrds,
    R.M.Parkhi

    0
    #186691

    zeus alvarico
    Member

    Hi everyone,
    I’ve been using Shainin DoE approcahes / tools to solve not only manufactuuring process / test problems but also with  the product design (electronics) problems  for almost 10 years .I find it very simple effective and useful for those who are dealing with marginal cases where there are too many variables affecting the problem.
    I learned  those techniques from my previous company (ASTEC Power Inc(now  EMERSON) – seminar by Keki R. Bhote).
    I post this message to support Shainin LLC  and I hope my new company will realize the need to apply Shainin DoE on most of our cases.   
     Zeus A. 

    0
    #186692

    Thothathiri
    Member

    I am reading the book written by Keki Bhote, its very good in explaining with case studies from Motorola practiced in the year 1980’s. I fully support for implementation of Shainin methodologies in organisation to get 100:1 improvement in short time.

    0
    #186694

    Mikel
    Member

    The only problem is Shainin methods were abandoned at Motorola
    by 1985 in favor of better methods.Keki is a good guy but was also seen as ineffective and put out of
    any role in the improvement of process or product by 1985.Shainin methods were useful when methods were done mainly by
    hand but the availability of DOS based stat tools for the PC
    basically did away with any advantage there by 1986.As a data point, Shainin methods were preferred by GM and
    Chrysler but abandoned by Ford in the 1990’s. GM and Chrysler
    have now also abandoned this approach.Both of you guys, look into other methods before buying the
    Shainin hype.

    0
    #186699

    Robert Butler
    Participant

      Since you mentioned Shainin DOE the post below may be of some interest.
    https://www.isixsigma.com/forum/showmessage.asp?messageID=63748
     

    0
    #187339

    X A Rand
    Member

    Jay, Can you tell us how this was resolved? Were you able to find a mathematical trick to get you off the hook?”the question is, how can i disprove these results? obviously, the shainin study captures zero interactions, that’s my first course of defense.” You must understand that EVERY interaction that matters is contained in this small sample of parts.
    X A Rand

    0
    #187342

    Mikel
    Member

    First you are responding to a six year old post and second there is no
    defending Jay’s part. If the problem follows the part, the part needs to
    be understood. Zero energy should be given to disputing the results.
    100% should be given to understanding why the problem is following
    the part.Shainin’s methods aren’t efficient and they are outdated, but a
    component swap that says there is a difference is correct. Understand
    the method before trying to critique the result.

    0
    #187347

    X A Rand
    Member

    Stan,Two sets of parts and swaps show the problem follows Jay’s part. Could someone draw the conclusion that the method is inefficient from this example? Is more information needed to draw that conclusion? X

    0
    #187351

    Mikel
    Member

    Did you read what I wrote?Shainin methods in general are inefficient, but no, if the problem
    follows the parts, the parts are a problem. It is a first step – now how to characterize the parts. I personally
    would not use a Shainin method for the next step.Component / Variable Search plus Multi-Vari are two of the
    “Shainin” methods I make sure everyone knows. Neither was
    created by Shainin but they were marketed very effectively by
    Shainin as part of his cult.

    0
    #187354

    X A Rand
    Member

    Stan,I though I was reading what you’ve written. Now that the problem follows the part what efficient method would you use to characterize the parts as your next step? Could you recommend a source for Component / Variable Search?Thank you for the help.
    XAR

    0
    #187392

    Anonymous
    Guest

    I think Keki Bhote’s book ‘World Class Quality’ covers some of the techniques.I wouldn’t recommend variables search, although I’ve used component search many times to solve difficult problems (high dielectric miniature microwave antenna and ink-jet modulation)The critical thing to remember about component search is to have repeatable breakdown and rebuild performance. Sometimes this involves intricate solder removal, re-soldering, etc.There is a discussion about Shainin Techniques here:http://elsmar.com/Forums/archive/index.php/t-9607.htmlGood luck!

    0
    #187395

    Severino
    Participant

    “Statistical Engineering” by Steiner and Mckay contains an algorithm very similar to the Shainin approach, but modernizes it to utilize statistical  techniques available in Minitab.  The book is loaded with examples and the supplemental CD contains a lot of indepth discussion including comparison and criticism of many of the Shainin techniques (i.e. Variables Search, Paired Comparisons, etc.). 
    The key takeaway from Shainin as I see it is to let the parts guide you to the answer.  Once you get into that mind set, you quickly abandon tools such as fishbone diagrams as your first approach to problem solving (although there are situations where they too are useful).

    0
Viewing 40 posts - 1 through 40 (of 40 total)

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