iSixSigma

Why only six sigma not seven or eight sigma.

Six Sigma – iSixSigma Forums Old Forums General Why only six sigma not seven or eight sigma.

Viewing 68 posts - 1 through 68 (of 68 total)
  • Author
    Posts
  • #31829

    HVGUPTA
    Participant

    I got little confused when workmen asked me as to why only six sigma, why not seven or eight sigma. Can any body help me to find out the answere for this question? 

    0
    #84290

    DD
    Participant

    A good question indeed. There is no harm in going from 6 to 7 og 8. In fact airline industry operate at 10 or more for flight take off & landing, but for all practical purposses in an industry where humal life is not at risk six sigma is a good enough improvement benchmark.
    More than that may not be worth the Time, money$$ and effort.
    Thats my take, Hope it answers your query.
    Regards
    DD

    0
    #84294

    Mikel
    Member

    DD,
    You don’t know what you are talking about. Airline safety is only about 6.4 sigma. There is no organization that has reached 6 sigma, so the question of 7 or 8 is only theory and kind of dumb.

    0
    #84301

    Mike Carnell
    Participant

    HV,
    I am not sure why you are confused?
    What is your current Sigma level? If you are running at a typical industry average you are in the 3-4 range. Your actual short term goal is probably in the 5 range. Most industries top out around 5-5.2 before thy need to start moving into DFSS. If you are getting pulled into some philosohical navel gazing over that question then you aren’t going to spend a lot of time moving your sigma level (at least up).
    John Lupienski of Motorola Elma, New York is as good as anyone walking the face of the earth at Six Sigma or just plain process improvement. His factory delivers just over 100 products (automotive). He presented a graph at a Toronto IQPC conference that showed his progess over the last 7 years moving the factory (measured at the customer) from 5.1 sigma to 5.8. That makes John about 12 – 22 years away from 7-8 sigma. Unless you are very young you probably woun’t be seeing it any time soon. The discussion is NVA.
    You have a few issues. First, sigma isn’t a very sensitive number. When your process is screwed up it takes big numbers to make it move but that is when improvements are easy. When it gets to the 5 and up range it takes less to make it move but they are tougher improvements. A discussion around 7-8 is a waste of time. Two, why would they ask you about something they don’t particularly understand? Maybe because you get confused and go away. Maybe that was their intent in the first place. When you are confused what happens to SS from their perspective? It went away. Three, there is a good chance you opened this esoteric discussion yourself. The advantage to using sigma is it lets you compare dissimilar thing (basically lets you kill off the old stalling technique “We’re comparing apples to oranges”). It is a management metric. At the shop floor level (product level) the metric that makes sense is ppm or dpu. Use the correct terminology in the correct environment. If you have a buring desire to pontificate on sigma do it somwhere where you don’t do any damage.
    Good luck.

    0
    #84458

    Amitabha Chattopadhyay
    Participant

    Implementation of Six Sigma ensures “Conformance to Specification” at a pre-defined conifidence level. Neither we can afford to set a very loose specification (as otherwise bound by market / customer demand) to achieve very easily Six Sigma level of performance nor we can undermine the high confidence level as set in Six Sigma Model.
    This means that Six Sigma itself is pretty diffficult to achieve and reknowned companies are all struglling to achieve this. So we can not even dream of  7 Sigma and above.
    It is not our whims and fancy that decides at what sigma level we need to operate rather it is the organizational capability vis a vis the bench mark (which customer adores) that decides the level.

    0
    #84459

    belt
    Participant

    6 Sigma takes care of 99.73% of the process deviation. This leaves only 0.27% as suspect product. Any effort to reduce this 0.27% will result only in extra efforts, $$$$$$ & time…..without the equivalent amount of benefits.
    Does that answer your query.

    0
    #84463

    Dr. Steve Pickering
    Participant

    To all,
    I have read all the replies, and I am amazed at some of the responses, I was not going validate this forum with a reply until I got to Mike Carnell’s reply, Mike, 3 fair points well made !
    Stan, I have an Aeronautical Scicence Ph.D and 15 years Industry experience, I am also an Academy Trained MBB, And I know it is a proven fact that there are Civil Aerospace processes operating at 11-12 Sigma, The 6.4 sigma that you refer to actually applies to ALL Aircraft including Military Vehicles and applies only to Aircraft landings, Please do not comment on subjects you have little or no knowledge of.
    belt, Not quite sure what level of Belt you think you may have achieved, but 6 Sigma does not account for 99.73% of defects, The correct figure is 99.99966%, that’s for centered Sigma, If you take into account shift, Then the number drops to 99.865% but still equates to 3.4 ppm – Another person who pretends to know what they are talking about.
    Now, for the poor sole who posted the original message, My advice is to concentrate on your subject and be in a position to answer the simple questions that crop up from time to time, Or leave the teaching to those of us who know what we are doing.
    To All, Please do not re-post a reply, I will not be reading them.
    Kind Regards,
    Dr. Steve Pickering

    0
    #84464

    jagdish
    Participant

    This is a question that has no logical answer. However  here is a background.
    Normal industry standards of a process is +- 3Sigma. When Motorola realized that they have to do something to drastically improve the process and reduce the variation it was taken as +-6Sigma (reduce the variation be a huge factor ). 6sigma is something that they came up with and made it as a Industry practice. The theme in Sixsigma is REDUCE variations.
    Practically it is very tough to bring your processes to 6sigma and sustain it. If you are able to get 7 or 8 sigma it is really excellent.

    0
    #84467

    Mikel
    Member

    Dear Dr. Steve,
    A real PhD in Aeronautical Science and Academy trained? Wow, you must know everything!
    I know whaer the original data comes from, not some hype you learned from spin Dr. Mikel. You are the one who should not comment if you are just passing on tribal BS from the good doctor.
    Dr. Steve, since you are sooo smart and soo well educated, why don’t you explain what 11 sigma means in terms of number of take offs and landing without incident.
    In case you are unclear of this, I am in awe.

    0
    #84468

    Mikel
    Member

    Wow, Dr. Steve, you are right about this one. This guy doesn’t have a clue.

    0
    #84471

    Manoj Bhardwaj
    Participant

    Dear Mr.H.V.Gupta,
    Have you ever thought of asking NASA the Sigma level of the prestigeous Space Shuttle Programme? It is not a question of number of the Sigma Level attached to the process but assessment and review of the performance level of the process ,as felt by the customers.I dont think you would like to ask parents of Kalpana Chawla as to how much is the Sigma level of the space shuttle programme.
    Sigma level can not be decided till you know the Voice of the Customers.The level can be improved if required .
    Manoj Bhardwaj 
     

    0
    #84475

    Anurag Asthana
    Participant

    Six Sigma is / was just a benchmark, when you reach that, you ought to move to higher levels, the goal is zero failure or infinite sigma.

    0
    #84485

    R. Ramirez
    Participant

    Addressed to any Ph.D:Two quotation from Dr. Wheeler’s Understanding Statistical Process Control book:1)”…Dr. Taguchi’s concept….On Target with minimum variance”….”2)”….Zero defects, Six Sigma Quality, Cost of Quality and all other specification based NOSTRUM miss this point”My quotation (a simple mortal human)…even if you have 3.4 ppm!

    0
    #84493

    Sinnicks
    Participant

    Gents,
    This stream of answers and hypothetical stuff doesnt really answer the poor mans question.
    I have been asked this before too, my answer was simple, it is possible but we are faaaaar away today. So lets walk before we can run and get our current process performance to a level which we can mention in public.
    Depends what you want to do to the green belt…want to make his/her head explode and humiliate him/her (not my prefered method..then start spouting statistics/theory at him/her) Want to get him/her to focus on the problem at hand and stop blocking then tell them to think about pudding later…we are not even at the starter yet.
    :-)
     
     

    0
    #84494

    Sanjeev Goyal
    Member

    Dear Sir/madam
    The same question u wud have asked if it was 7/8 sigma.
    regards
    Sanjeev

    0
    #84496

    Patrick L. COlvin
    Participant

    When Measuring any process a number of variants can cause process shift.
    The normal distribution of the process falls within a given graph which when plotted loks like this

    Thes lines will never actually touch the x plane
    if your process is in control the center line of this graph will be labeled as 0 the others are then labeled positve and negative to either side .
    the over all width of this graph is the amount of process control you have.
    so you end up measuring out three in either direction thus
     SIX SIGMA

    0
    #84497

    Patrick L. COlvin
    Participant

    When Measuring any process a number of variants can cause process shift.
    The normal distribution of the process falls within a given graph which when plotted looks like this

    These lines will never actually touch the x plane
    if your process is in control the center line of this graph will be labeled as 0 the others are then labeled positve and negative to either side .
    the over all width of this graph is the amount of process control you have.
    so you end up measuring out three in either direction thus
     SIX SIGMA

    0
    #84505

    Abodorra
    Participant

    I think a Six sigma gets us to 99.999% or somthing close to that. A lot of green belt projects start out with a sigma of 2 or 3 and getting to 4 or 5 is a major challenge. A six would be a stretch for many organizations because it is near perfection. In my opinion, a 7 or 8 would certainly push the limits and force six sigma practitioners to drive for extraordinary results which I suppose is possible in some settings, but certainly not all.

    0
    #84507

    Anonymous
    Participant

    Hatsoff to Dr.Steve

    0
    #84508

    Arora
    Participant

    Hi Guys, Do u want to know about an organisation, which is implmenting 8 or 9 Sigma? If ur answer is yes, then click here…. http://www.guardian.co.uk/Archive/Article/0,4273,4447391,00.html
    Do you want to know more about it, then click here…..
    http://www.google.com/search?num=50&hl=en&lr=lang_en&ie=UTF-8&oe=UTF-8&newwindow=1&q=Mumbai+tiffin+delivery+network&spell=1
    Who want Quality gurus?
     

    0
    #84510

    rama
    Participant

    Be Proud of as a Indian… We teach the things to the world,

    0
    #84517

    Mike Carnell
    Participant

    belt,
    Just a point. If infact six sigma was 99.73% then it would cause a serious problem for those people who have written books, given speechs, etc, that claimed it was 3.4 ppm. You need to have a discussion with someone who understands the concept.
    That is a pretty convention (as well as unsubstantiated) statement – with regards to the $$$$$$$..
    Good luck.

    0
    #84529

    marcel
    Participant

    Belt,
    You appear to be as confused about 6-sigma as others that I have encountered since I started my journey on the path to 6-sigma.
    6-sigma is not +/- 3-sigma, where 99.73% falls; for the correct answer I suggest you get some training, and, as the good Doctor stated, leave the dissemination of 6-sigma to those who can…
    Marcel

    0
    #84533

    Pandwe J Lubamba
    Participant

    I agree, Stan.
    I don’t think there are any organisations operating at 6 sigma.
    If you try the sigma calculator you will find 95% yield at 3.14 sigma, 99% yield at 3.83 sigma and 99.9% (exceptional) at 4.59 sigma.  Six sigma is at 99.9996%!
    I hope that helps.
    Pandwe.

    0
    #84557

    Mikel
    Member

    Since Dr. Steve is an ostrich with his head in the sand (will not be reading this post), I offer the following for everyonelse to consider.
    # of commercial aircraft in operation in the world –
    13,500 (approximately)
    Average # of takeoffs and landing per day –
    6
    Average # of days in service per year –
    300
    Near as I can tell, 11 sigma would mean no issues on take off and landing in over a decade – Sounds like a pretty bogus number to me. Anyone out there think there have been no issues on take off and landing in a decade. If you do go with Dr. Stevie. He also thinks Dr. Mikel is something special.

    0
    #84577

    Dr. Steve Pickering
    Participant

    Stan,
    It would seem I have hit a raw nerve with you, Why ?
    Your reading is obviously not that hot, Otherwise you would have correctly interpreted the actual wording of my comments,
    “And I know it is a proven fact that there are Civil Aerospace processes operating at 11-12 Sigma”,
    I did not specify that this was for just take-offs & landings, I merely pointed out that some processes in this industry are proven to be operating at this figure!
    As for Dr. Mikel Harry, I believe he is a good statistician and a clever guy, I am not a betting man, But I would hazard a guess he has amassed a formiddable fortune, through preaching, practising and passing on his Six Sigma knowledge, Maybe it is not I who has is head in the sand afterall, and even so, Maybe better in the sand than up your own ass, (as you Americans often say)!
    Wake up lad, Stop having a go at all who may oppose your views and get on with the job in hand, IMPROVEMENT !
    Dr. Steve Pickering

    0
    #84586

    Mikel
    Member

    Doc,
    I just asked for your data. Tell us what it means to run at 11-12 sigma and give us an example, not an anecdote, of a process that is.
    The point is that is it is not useful to pass along this BS when most are at 4 sigma. To talk about even 6 sigma being achieved as an enterprise is not useful as none have achieved it.
    As for Dr. Mike, I agree with all that you have said except the practicing 6 sigma, theere is no evidence he has ever done this himself, just a lot of taking credit for the work of others. He is insanely rich, and if that is your scorecard, he is a man to be followed.
    Give us data Doc!

    0
    #84596

    Mike Carnell
    Participant

    Dr. Steve Pickering,
    I’m not particularly interested in the >6 sigma discussion. I stated my opinions in an earlier post.
    Just a point. Before you pass judgement on Mikel and choose to follow him to far you should probably get to know him a little better. There is one thing that cannot be argued – he did get this industry rolling outside Motorola. He was/is the great purveyor of Six Sigma.
    If you were around at the ASQ Quality Congress debacle in Indianapolis “The great Six Sigma debate” where Mikel proposed solving world hunger with Six Sigma? What is the guideline for six sigma project selection “you can’t solve world hunger.”
    Let’s run a survey: How many of you guys that hired the Academy have actually worked a project – where you had to interact with people at the process level have actually had Mike with you?
    The cowboy persona? Mikel is from where? Rich is from where?
    In the words of Andre Aggasi – “image is everything.”
    Maybe not the great practitioner you imagine – strictly my opinion. Away from Six Sigma and the image – a nice person, good father, and a former Marine.
    Just my opinion – I could be wrong.

    0
    #84800

    Mikel
    Member

    Dr. Steve,
    I am still waiting for you to explain to everybody what it means to run at 11 sigma. I am guessing that you do not actually know how to figure it out.
    Alsogive us a specific example of something in real life that is.

    0
    #84862

    Mikel
    Member

    Still waiting.

    0
    #84870

    Jackie
    Participant

    A quote from Dr. Steve’s last post on this subject:
    “To All, Please do not re-post a reply, I will not be reading them.”
    Looks like you’re going to be waiting awhile Stan.

    0
    #84874

    Mikel
    Member

    Jackie,
    Look closer, he was kidding, he replied after that.
    Truth is, he doesn’t know what he is talking about with his claim of 11 sigma processes and he is hiding. I don’t think he knows how to figure out what 11 sigma means – most don’t.
    Thanks for dashing my expectation of a lively discussion based on data instead of rumour mongering.

    0
    #84876

    Jackie
    Participant

    Stan,
    You got me.  I missed the later post….where he was (hard as this is to believe) more insulting than you.
    Have fun.
    Jackie

    0
    #84893

    Mikel
    Member

    More insulting or not, he is always less informed and claims data he does not have. You should not be insulting when working from such a weak platform.

    0
    #84898

    mcintosh
    Participant

    With a 1.5 sigma shift, the maximum sigma score in the “sigma calculator” (top right of this screen) is 7.98, and this happens for a DPMO of 0.00005 or 1 defect per 20,000,000,000 opportunities (that would be 20 billions in all the world except outside USA, where it would be 20,000 millions).
    I wonder if a single aviation process has already created that ammout of opportunities to be able to say that, with 1 or less defects occurred, the process is better than 7.98 sigmas. For 11 sigmas it would have to be 1 defect in severeal thousands times those opportunities (inagine a number like 200,000,000,000,000; or 200 triillions or billions, deppending if you are inside or outside USA)
    I remembes having seen an European avirplane design standard (JAR) that said something like that the chances for catastrophic filure shall be less than 1 in 10 millions. That is 6.7 sigmas. Far from 11.
    I also remember that in this forum someone explained that the “design input” for the transatlantic airways was that the chances of a person died in an in-flight colision should be less than 1/1000 the chances of dying because a natural cause. Roughly (I know it is not correct, but just to get an idea), a person lives 75 years or about 650000 hours. That is 1 death (defect) every 650000 “lived hours” (opprtunities). If we consider that the transatlantic part of the flight is 5 hours long, then the chances to die beacuse of a natural cause in that part of the flight are 1 in 130000 (5 opportunities). If we take 1/1000 of that it is 1 in 130 millions. That is 7.1 sigmas. Still far from 11.
    If the aircrafts and the airways are designed for let’s say 7 sigmas, I don’t see why would they behave better than that. And even if there was an aviation process than never created a defect, still it would be hard to have enough information (enough “successful” oportunities) to say that it is at an 11 sigma level.

    0
    #84902

    Mikel
    Member

    Tom,
    Amen.
    You just joined Jackie as one of my heroes.

    0
    #84913

    Ronald
    Participant

    Tom and Stan-
    In absolute agreement!  Even nature and its processes are not above a 7-8 Sigma.  Take gene replication, cell mitosis, bio-chemical reactions.  Things that happen hundred of trillions of times.  These things have defective occurences in the billions leading to defects, cancer, diseases.  Even God is not at a 11 Sigma level.  I guess the aviation industry have found supernatural processes.
    Besides,  if CTQ processes were at >8 Sigma, the industry and the insurance companies would be putting up billboards the size of Cleveland to advertise their accomplishments. 
    I guess if I do something 1 time and it is sucessful.  I should never do it again so I can have a infinite sigma level.

    0
    #84922

    Long
    Member

    According to my preliminary calculations, 11 sigma is the equivalent of 1 defect for every 1,426,971,118,249,190,000,000,000,000 opportunities.
    If the opportunities are occurring at the rate of one hundred million per second, it would take about 7.5 billion years to record enough opportunities to rate a process at 11 sigma.
    Since the generally accepted age for the earth and the rest of the solar system is about 4.55 billion years, I think it would be safe to say that there are no processes that can be rated 11 sigma.

    0
    #85833

    Junior
    Participant

    Question 1: Patrick L. Colvin, I saw that reply once, ignored it, saw it again & my curiosity got better of me ! I thought 3 sigmas on either side of the mean (within specs) is a 3 Sigma process. For a Process Sigma of 6, you need SIX sigmas within specs on either side of the mean. Am I missing something here?
    Question 2/3: Can someone please explain what happens to Process Sigma computations when the distribution is not normal & when you have a one-sided spec limit?
    Just the answers will do please, I already know I’m a half-lettered ignoramus ;-)

    0
    #85834

    Strumms
    Member

    I was going through the replies and having a good laugh. Pretty interesting and active discussion. Junior, it’s 3 SDs on each side of the mean in a 3 Sigma process… when you have a 6 Sigma process, it’s 6 SDs on each side of the mean.
    I think it was mentioned earlier about how the ends of the binomial curve never reaches the x axis. So therefore, in theory there’s never a chance for 100% accuracy. To me, in my opinion only, you can have more than 8 Sigma BUT in theory only. I think most industries at the moment have difficulty in operating high sigma level processes. If you actually calculate it, a 3 Sigma process will have a SD of 0.08, 4 Sigma will have 0.06 and 6 Sigma will have a 0.04 SD. So if in effect we go up to 11 Sigma, we’d have a 0 SD… to me, that’s practically impossible due to the inherent variation in every process/part (random variation). If you disagree on this, a good example to see would be the Variable Gage R&R analysis charts. Check out Chart 1.
    If data is not normal, transform it… or use the Weibull distribution to make your predictions. If you have skewed or a data set that has a lower/upper bound of zero/non-zero, you can try using the log or natural log functions to transform it (or even square/cube root the set, whichever function is applicable). But my preferred method, use the Box-Cox transformation in Minitab, then once that’s done, calculate the Sigma level.
    Just a note… the weibull distribution is for when you have those instances where you can’t convert the non-normal data to normal… the weibull distribution basically is an estimate of the percentage of out of spec products.
    Hope my answers help you in both your question…
    David T.

    0
    #85835

    C.Stokes
    Participant

    A1. What you said is correct: a 3 sigma process has 3 S.D to the right and 3 S.D. to the left of the mean that are within your set tolerances.
    Process capabilites indicators are the techniques to use:
    If your process is centerd and at 3sigma your CP would be =1.  The CP for a six sigma process the CP would be =2. (multiple your CP by 3 to work out your sigma level)
    A2. If your process is not centered (Cpku not equal to Cpkl) then the Cpk for the lesser of the two values (x3) would give you your actual  sigma level.  The Cp value (x3) would give you your potential sigma level if you managed to center your process.
    You can use the CPk calculation method for one-side specs as well (use either Cpku for uper specs, or Cpkl for lower specs)
    If your data is not normal then the calculations do not apply.
    Either try to fit your data to a known distribution (and calculate the probabilities thae hard way), normalise the data first (and report that it has been normalised) …. alternatively use non-parametric SPC to really control your process and forget about trying to calculate a sigma value!
    Hope this helps
     

    0
    #85848

    Martínez
    Participant

    Hello Mike, I need some clarification and confirmation. Some of these people are loose canons so I wanted to go right to you. First of all there are two ways to measure it: Attribute (how many people died or were hurt in airplane accidents as a % of travelers) or Variable (I am given an upper and lower spec limit and I count the standard deviations between those and my target) In attribute I can see all of the arguments. People can chooses different measurements and start arguing over it. Thats OK with me. With variable though it seems that there is no question. Moreover how can people say that 10-12 Sigma is ‘an unattainable ideal’ or whatever was said. If my supplier wants a piece of paper larger than a postage stamp and smaller than a doorway and I consistently supply 8 1/2 x 11 then I can easily hit 11, 12 or 200 Sigma.

    0
    #85851

    Mikel
    Member

    On question #1, you are right. Mr. Colvin obviously was trained with Annie and does not have a clue.
    On question #2, you have two choices. Transform to normal if it is legit or find the correct underlying distribution and find your margin with respect to expectations. Sigma level, after all, is just a statement of margin.

    0
    #85856

    Loose Canon Stan
    Participant

    Mikey is on hiatus from here it seems, so you get the loose canon.
    Yes, in theory 11 sigma is possible, but we are talking reality. The scenario Dr. Stevie claimed to have data on does not exist. Read one of the posts where someone took me up on my challange to define what 11 sigma means. If you know some reality, not some fable like Dr. Stevie – bring us data and I will stop being a “loose canon”

    0
    #85858

    John H.
    Participant

    David T.
    In Theory and in practice, Natural Mixing Processes(ex: Thermodynamic Processes) operate at nearly Infinite sigma levels. In Professor Kenneth Denbigh’s notable text on the Principles of Chemical Equilibrium , he uses the example that the probability of a density relative change of .001% in 1 cm^3 of air is smaller than     10^-(10^8) and would not be observed for Trillions of years. In reading the posts on the subject, I could help but wonder if  Dr. Steve Pickering  was waiting for a “sucker bet” on an example of an 11 sigma process
    -John H.          
     

    0
    #85859

    Junior
    Participant

    Strumms, thanx for the insights. However, am curious to know how you got the SD values for various Process Sigmas (“3 Sigma process will have a SD of 0.08…”)?? I would have thought the SD value depends on the actual dataset you’re dealing with….how did it become a standard derivative of Process Sigma?

    0
    #85863

    Strumms
    Member

    You’re right Junior. What I was trying to put across was that as your sigma level increases, your SD decreases closer to 0. I was referencing a Standard Normal Distribution, thus the figures. My apologies for being confusing.

    0
    #85866

    C.Stokes
    Participant

    A few thoughts on sigma calcs for non-normal data:
     If you have derived a process mean and S.D. from individual counts and the data is not normally distributed, then as everybody has been saying it would be inappropriate to use the normal sigma calcs (you’d have to transform first, or fit to an appropriate distribution).
    However, if you took sub-samples from your process and worked out the ‘grand’ process mean (mean of samples) and estimated S.D., then the Central Limit Theorem would imply that the distribution of your samples would be normal.  Consequently you could then do standard sigma calculations on this data (but report that it was on sub-samples not individuals).
    Am I correct in this assumption or am I missing something?

    0
    #85868

    Junior
    Participant

    Strumms, am still not clear about what those figures mean. I would have thought the SD of a Standard Normal Distribution is 1 regardless of the Process Sigma level. Could you pls briefly explain the 0.8 fig you talked about for a 3 Sigma Process?

    0
    #85880

    Strumms
    Member

    Hi Junior. I shouldn’t have mentioned about the corresponding figures. The point I wanted to bring across was the decrease in SD values as your Sigma level increases. As I’ve replied before, you’re absolutely correct in that for a Standard Normal Distribution, the SD is 1 and the mu is 0.

    0
    #86586

    Dave Gentile
    Participant

    > 6 Sigma?  I have no data, but think it applies to:
    1 Returnable or un-buyable products in Supermarkets
    2 Unviewable TV-hours vs. TV-hours broadcast
    3 Driving miles lost due to “bad” gasoline

    0
    #86593

    Howard Cooper
    Participant

    What a “string” of comments you created. I hope this response will give you some practical, usable value. Other’s comments certainly gave you enough background on 6 or 7 or 8 Sigma, but here is an easy, cost effective way to move from 5 to 6 or from 6 to 7 or 8 Sigma. Let us keep in mind “The Goal.” The goal is to make money! Using DMAIC is an excellent way to train people to focus on problems and generate solutions, so the problem will not crop up again. The goal is to point a motivated team toward gaining increased efficiency, productivity and increased profits. We may call it Six Sigma but it often never really gets “there” unless we address the stresses that cause malfunctions, failures and downtime, to the computers and machine tools – the other important but ignored part of that “team”!

    0
    #86625

    James G. Chirumbolo McKee
    Participant

    Assuming you are treating flights individually and a crashed flight is a defect then the Sigma value is really closer to 3.5 to 4.0 (good years being more towards four). Some may claim higher numbers if they change the assumptions, but quite simply if you look at the total number of takeoffs per year as compared to the number of planes that crash you will get something in this range.

    0
    #87960

    Curtis Aldrich
    Participant

    Jim,
    Doesn’t GE aircraft engines strive for this standard?  Johnson&Johnson’s Ethicon Endosurgery strives for a similar quality, though their metrics for acceptance could be much looser than those for a commercial turbofan.  Accepted manufacturing tolerances, however, can be loosened to allow any manufacturer to lay meaningless claim to the almost holy sixsigma standard.  This may be a difficult game for the airlines to play in the scenario you pose.  There really isn’t much grey area in the metric of crash vs. no crash.  ; )
    But seriously…Email me if you see this.  I have some questions regarding the ex-pad on Jefferson.
    Thanks,
    Curtis

    0
    #87963

    James G. Chirumbolo McKee
    Participant

    Curtis,
    GE does strive and will continue to strive for this level of performance. Where we are now is merely a step in the journey towards this achievement. It will take the balance of cost, quality and time to get there.
    Cheers,
    -Jim
    [email protected]

    0
    #87965

    Curtis Aldrich
    Participant

    Jim,
    Did you graduate from Mariemont High School in 1991?
    Curtis

    0
    #87968

    James G. Chirumbolo McKee
    Participant

    Does a bear #%@% in the woods? How many people do you know who have a last name like mine?
    Yes it is me.
     
    Cheers,
    -J
    [email protected]

    0
    #87974

    Ron
    Member

    Purely by chance!  At 3.4 PPM
     It is considered excellent in most manufacturing organizations, however in the airline industry, medical fields etc. you would want better !
     

    0
    #87987

    fernando
    Participant

    The importance of 6sigma is not in the single process yield, but in the rolled yield. If you have a process/product with 20 characteristics each of them operating at 3 sigma (long term), the rolled yield is only 25%.
    If you have a very complex system with 1000 characteristics, you need a 6sigma capability for each of them in order to maintain a rolled yield of 99.3%
    Another example: 100 chracteristics, each of them having capability 4.5 sigma (99.865). The rolled yeld drops to 87.4%

    0
    #87988

    MG
    Participant

    I am under the impression that 6s is almost unachievable firstly due to financial restrictions imposed by the knights of the round table (upper management) and secondly the fact that each subsequent process improvement yields a lower return in relation to investment, estimated @ 50%. 
    Equally, if you have achieved 6s 7 or 8 would just be icing on the cake.

    0
    #87991

    jaideep
    Participant

    There are organisations working at aboove 6 sigma level.
    example is motorola where the DPMO is lesser than 3.4
    An agreeable answer to why not 7 or 8 sigma would be that the returns w.r.t the investment in terms of time & money put in is v v less.
     

    0
    #87995

    Mikel
    Member

    Whoever told you Motorola was above 6 sigma was lying to you.
    The rest of your answer is just theoretical jiberish.

    0
    #112844

    Shreesh S Kate
    Member

    Much has been written about the Dabbawalas (Lunch Box delivery boys) of Mumbai.  Ever since the popularity of their craft grew I have on many occassions, while travelling on the  Mumbai Local Train, observed the dabbawalas at work and have seen it from Six Sigma point of view and tried to figure out what could be the secret behind their success.
    They are organized as a Co-operative Society (no-profit organization).They are about 5000 strong handling about 200,000 dabbas every day one way – that makes 400,000 transactions a day – for a charge of about US $ 4.00 (four) per month per dabba.
    Some lessons may be learnt from the following observations.

    The consignor and the consignee is the same person – no room for inter-party disputes. They do not issue a document / consignment note. They do not obtain an acknowledgement (‘legal’ ??) and there is no record of delivery.
    Although the No of transactions per day is large each person handles a limited No transactions at a time, manually, with no processing involved except sorting (manageable scale – around 50 dabbas per person on an average at a time).
    They do not deliver to the exact address of the person – they deliver it to the floor or wing or building No, etc., The consignee/consignor identifies his dabba (lunch box container) from the lot unloaded at his floor/wing/bldg based on his personal identification marks on the container and he puts it back in the lot after lunch – it saves time & manpower. It shortens the address tremendously and avoids writing and reading of full address and name – and avoids those many ‘error opportunities’.
    The ‘address’ is coded in specific standard format using color paints (requires visual recognition – not reading). The color code containes 3/4 characters for origin and 3/4 characters for the destination. The pattern of colours/signs on a sorted lot of 50 Nos to be handled by one person has a very specific color pattern – any odd pattern becomes easily recognizable. This by-passes a major source of error (‘defect opportunity’ in Six Sigma parlance) – reading a non-formatted address and sorting on that basis (like in case of post card). The ‘visual signals’ are used by Japanese extensively in car manufacturing.
    The color code address is almost permanent on the lid of the metal container – not put each time for each delivery – those many error opportunities are saved. 
    The external container that houses the lunch box inside is of standard shape & size. The containers are palced in a single tire in a wooden crate (the lids of entire lot are visble). The color code pattern on the lids of the entire lot of 50 can be visually scrutinized very quickly in one span of vision.
    The color codes are such that each person at origin / destination checks whether he has recd the right lot/container or not. Thus it undergoes very quick checks at each point where the dabbas change hands.
    It also by-passes a major activity of document processing which in turn obviates a major step of flow of dabbas & documents to Offices, data capture, distribution from offices, office staff, document processing, phones, computers, record of delivery, etc., (done for courier service) – it avoids those many ‘error opportunities’.
    The dabbawals never lose personal custody/contact with the dabba crates throughout the delivery process although they use 2/3 different means of transport – they almost travel with their consignment.
    The transactions are extremely repetitive in nature – origin and destination remain the same over a period for a dabba – so the ‘address’ color codes for a container are constant over a long period. Because of this the color code pattern of a lot of 50 Nos in a crate also remains constant for a long period.
    The same person handles the same lot over a long period every day – the color pattern on lids gets mastered / remembered easily by the dabbawal for a lot of 50 – any error is easily obvious as it immediately shows as an odd item in the population. Every dabbawala knows exactly what to expect in his lot.
    The operation leverages on the efficient Mumbai Local Trains.
    The operation is non-commercial in nature – it is philanthropic in content.
    The consignment and its contents are also unique – it is low cost/value, perishable, cannot be repacked/rehandled, contents are non-standard, no alternate commercial usage, no resale value, the container (dabba) is used and has little alternate usage – it renders the consignment practically non-reusable hence non-pilferable – it is secure with no extra cost or processes.
    The effectiveness of operation leverages on above properties and uniqueness of the operation and the nature of the consignment. It also depends on low stake of the customers, the personal integrity of the Dabbawalas and the trust of the customers.
    The dabbawals of Mumbai operate in a supremely simple and low cost manner in an environment characterized by very high repetitiveness, certainty and predictability – practically nothing changes for months in the entire course of operation, except an odd addition/deletion of containers and maybe a few persons.
    Repetitive/predictable transactions with high visual content get mastered by human minds easily although statistically they add up to a certain number and a certain sigma score. Wonder if there is any process of ‘discount’ on the Six Sigma score for these kind of extremely repetitive/predictable transactions – SS professionals to pl scrutinize this.
    The success of dabbawala operation in Six Sigma terms lies in the extreme repetitiveness/ certainty/ predictability of the nature of the operation itself coupled with an equally simple design (the whole thing is Designed For Six Sigma-DFFS). The only ‘process’ involved is sorting, loading, unloading. It does not need any technology, infrastructure or qualified persons to operate – and that is the strength of it – and as such the number of variables (EFFECTIVE ‘defect opportunities’) is reduced to minimum.
    Can this operation be done on a mass scale or over long distances ?? Can it be replicated in baggage handling in Airlines for instance ? It is hard to imagine other such applications relevant for business/industry where similar modus-operandi can be replicated.
    My colleague in M&M had a brainwave to start a daily dabba service from Mumbai to US for indian students there ?? !!
    Someone should do a study of our legendary age-old Indian Postal System – it is principally the same as dabbawala operation but on a massive scale – traditionally no technology is employed – and they generally do a very neat, reliable job. The No of transactions is very high – the error opportunities are far higher – the six sigma score must be very high.
    This is no attempt to discredit the good-old-reliable dabbawalas of Mumbai but one needs to basically ascertain the facts notwithstanding the media hype.
    Will SS professional comment on this unique Operation in the World.
    Shreesh  S  Kate
     

    0
    #137364

    Chandran
    Participant

    As we know from Normal Distribution curve
    The area of 1s covers approximately 68% of all values,
    at 2s there are 95% and
    6s (six sigma) contains 99,99996% of all values.
    The theoretical value of 100% can not be reached in practice, because the curve meets the x axis in the infinity
     
    Now For Dabbawallas PPM Calculation:
     
    No of Dabbas Served / day=200000
    No of Dabbas Served per month= 200,000X25= 5,000,000
    Error Rate is 1 in 2 Months
    I.e. 1 error in 20,000,000 transactions (Collection & Delivery)
     
    PPM=0.058 or 6.8 Sigma
     
    Now six sigma gives us 3.4PPM. Is there anything beyond to measure that??
    The current six-sigma level is decided by Motorola to have a 10-fold increase in profits. Why, because most of Motorola’s products are complex and have many steps in the manufacturing process. With many parts as well. It was based on a typical 1200 part step process. And also 1.5 sigma shift.
    Juran indicated back in 1980 when referring to a 3 sigma wide bell shape curve on a normal distribution, that if your specification was at 4 sigma, then there was a better chance to catch bad parts before shipping them when you were doing SPC, (Statistical Process Control).
     
    Now can we rate Dabbawallas seven or eight sigma?
    Dabbawallas collect Dabba from Home & Deliver it to offices, This cycle is reversed in the  afternoon
     
      The CTQ (critical-to-quality parameters) address before lunchtime:
    (1) Lunch boxes must be delivered on time, and
    (2) Recipients must receive their own lunch boxes and not some one else’s.
     
    Humble Dabbawallas are still not aware of this calculation and they don’t accept making 34 mistakes in two months, which gives them 6 Sigma
     
    “ We make a mistake perhaps once in two months . Our livelihood depends on delivering efficiently”  told Raghunath Medge, President, NMTBSA
     
    Then Why Motorola did not go for Seven or Eight Sigma?
    As it just wasn’t practical from an economic point of view for Motorola .At some point spending money to train and monitor people and processes just wasn’t going to return on the investment. Of course all of this goes same for GE, 3M, Dupont, Honeywell, etc.
    Six sigma is the best productivity toolkit that exists today; The figure of six was arrived statistically by looking at the current average maturity of most business enterprises. We need to wait for new Toolkit till world becomes a more orderly and predictable (even with increasing entropy or chaos) place to live!
     
    Till then we can rate our Dabbawallas with Superlative term ‘Beyond Six Sigma’
    Six Sigma has one purpose: To serve the business. The business does not serve Six Sigma.

    0
    #137368

    The Soft Drink Vendor
    Member

    Yeah, yeah, yeah….. Hard to believe that the humble dabbas make only one error in 20 million opportunities. Maybe this is an example of a selection process where errors are not reported, recorded or consolidated.
    I love the quote which includes the word “perhaps”. Who wouldn’t be proud of a process where you can simply make up the data…..
    I know in my Dabbawalla business, which is advertised frequently on conservative talk radio (you know the one, make six figures working from home in your PJ’s, even with bad credit) we purposely don’t keep detailed records and if a client complains, he knows what we’ll do to his lunch next time. Toasted roadkill anyone?
    We soft drink vendors think we’ve done well when only half of our customers are unhappy. Dr. Pepper any one? None for you, Dr. Eugene.
    Hard to believe more of those dabbawallas don’t get bumped off by Mumbai cab drivers who refuse to use their lights to save their battery in the night hours.
    Maybe GE or Honeywell is thinking about getting into the dabbawalla business too. Better watch your back.

    0
    #165773

    Roger not the same
    Member

    I realize that this is an old mail trail, but I’m praying that the use of the term “loose canon” is not just a lame spelling error but a clever pun on the disparity of math calculations (and egregious spelling errors) on this hopefully erudite/literate mail trail.

    0
    #165775

    Mike Carnell
    Participant

    Roger (not the same),
    I am not sure who you were expecting a reply from but the use of the term “canon” was not my post so I can’t add much clarification in terms of intent.
    I did reread the 4 1/2 year old string and thought that MBB brought us a significant point with the example:
    “If my supplier wants a piece of paper larger than a postage stamp and smaller than a doorway and I consistently supply 8 1/2 x 11 then I can easily hit 11, 12 or 200 Sigma.”
    That is an accurate example of what can be done when someone gives you a specification wide enough to pitch a dog through. Let’s look at what can be done when you get a situation like that. It is easy to satisfy the customer just as MBB said. If I do supply a 8 1/2 X 11 sheet of paper (or even A4) what have I done for the company I work for? I have actually suboptimized the profitability of the company. I can supply a piece of paper that is slightly bigger than a postage stamp which will cut my cost, probably increased profitability and still satisfied my customer. The process that would cut that small piece of paper has a standard deviation. I can use that standard deviation to back the target for the cutting process off the lower specification a certain distance depending on what percent defective I am willing to ship (if you believe in the 1.5 sigma shift use that if you don’t use something else – it makes no difference to me). The result is you now have two happy customers 1. the external customer is still getting what they want 2. the internal customer (your company) is making more money.
    Anyone who has ever plated precious metals as a part of their process knows this to be true. A real life comment from a VP I worked for in terms of plating gold thicker “I can drive a Lincoln Continental off a f__ing bridge every day for what you are costing me in scrap.” He wanted me plating on the lower spec with a standad deviation of 0.
    Pounding out a high sigma value is not always the end game in deploying six sigma. regardless of what we may or may not want to believe most of us are here or are working for someone who is here to make a profit. If you are driving the six sigma deployment bus and the only gage on your dashboard is sigma you will eventually crash the bus.
    Just my opinion.

    0
    #165795

    Brandon
    Participant

    Great explanation & example Mike.
    Guess that’s why it’s called continuous improvement – there’s no end to how good we can get.

    0
Viewing 68 posts - 1 through 68 (of 68 total)

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