iSixSigma

Why Shouldn’t We Include Spec Limits in Control Charts?

Six Sigma – iSixSigma Forums General Forums Tools & Templates Why Shouldn’t We Include Spec Limits in Control Charts?

Viewing 22 posts - 1 through 22 (of 22 total)
  • Author
    Posts
  • #240834

    tgause
    Participant

    I’m confident I fully comprehend the difference between spec limits and control limits. The former is defined by the voice of the customer, whereas the latter is the voice of the process. Former is determined by the customer; latter by the calculating 3 sigma from the mean.

    So, why don’t we include spec limit lines on the control chart, specifically on IMR charts? I understand why we shouldn’t do it for X-bar charts, but why is it suggested that this is not a good idea for IMR charts? What harm could come from including these two additional lines on a control chart (other than the possible confusion a reader could have and mix up the control and specification limit lines)?

    I’ve read that spec lines can be included on histograms, box plots, and probability plots (see last sentence on this article). But, if they can be included on histograms, isn’t the control chart basically a histogram on its side (see the graphs on this article)? I would think including them could help us accomplish two things in one graph: 1) to determine see if a process is control and 2) to determine if process is meeting customer specs.

    0
    #240841

    Robert Butler
    Participant

    I don’t see any reason why you can’t if you want to.  In most instances though the question will be: which customer spec limits are you going to include on your chart?  I worked in industry for many years before shifting over to medicine and I can’t recall a single product where we were making something for just one customer.

    There is also the issue of having customer specs that are well inside your control limits but, because of their location, the fact that they are is of no major consequence.  For example, let’s say you manufacture polychrome fuzzwarts and the control limits for this sought after product are +- 3 polycrinkles.  Customer #1 has a spec limit of -3 or less to -1 polycrinkles, customer #2 has a spec limit of 2 polycrinkles to 3 or more polycrinkles and everyone else is happy with anything between from -3 to + 3 polycrinkles.  True, you will have to do lot selection for two of the groups but, if the cost of such an action is less than the cost of shifting the process when orders come in from the two populations then there is a good chance you will leave the process alone.  Under such circumstances the inclusion of customer spec limits will just clutter the graph.

    1
    #240844

    Strayer
    Participant

    Per @rbutler we don’t want to the clutter the graph and risk obfuscation.  I’ve seen spec limits added for a presentation when someone wanted to show that the process was statistically under control, but out of spec.  I think that was a good use.  But what if it was out of control but within spec?  Should we disregard what the control chart is telling us?

    1
    #240845

    Mike Carnell
    Participant

    @chebetz I am with @rbutler and @straydog on some level. There is no statistical jail. You want them on there put them on. Don’t put them on and walk away. Pay attention to what you have done and make sure you aren’t causing a mess. This whole thing is about continuous improvement. You want to try something, do it and if it works keep doing it and don’t worry about what other people think. If it doesn’t work have the integrity to say you screwed it up and remove it.

    Here is a couple examples. I had a BB I was mentoring in Skelmersdale England. Did a box plot and there were 3 boxes close together and 1 that was not even close. His plan was to adjust the one to the 3. The 3 were actually out of spec and the single plot was the only one that was in. That is a mistake you can make if you ignore specifications.

    I had a wave solder process on a bomb fuse line. We ran to WS 6536 which require binocular inspection of every solder joint. Idiotic requirement but I had 5 visual inspection lines of 3 inspectors each. Each group of 3 operated as an inspection team. Each team plotted the number of defects per board on a control chart. The issue came when some little nerd saw that the boards were distributed randomly to each inspection line as they came of the wave solder machine. This hysterical nerd dragged a lot of books into my office to convince me I was breaking statistical laws and she would have my head on a spike outside her tent before sundown. Those boards had to be in the order they were processed through wave solder. Right now there are a bunch of nerds shaking their heads in agreement. Without even a question of why off she went. Well I still have my head and she probably hates me to this day. As I said there were 5 lines inspecting from a single process. That should be pretty much a homogeneous product. My control charts were there so I could walk past and make sure no group of inspectors had fallen down some cosmic bunny hole and started doing something off the wall like calling a bunch of defects that were not there. I was controlling my inspection process not the wave solder process.

    Each board had a serial number. That serial number (bar code) was entered into a data base with the number and location of defects. The wave solder machine had a bar code reader as well. It, the wave solder machine, pulled data from the data base in the order processed (from reading the bar codes) and constructed a control chart for the engineer that ran wave solder. He controlled the process because I had control of inspection. Pretty cool how that works.

    We were quite happy with our controls on that process. Without the courtesy of a question some little nerd with a copy of Grant and Leavenworth (Statistical Quality Control Handbook – good reference material – you need a copy) had made the discovery of a lifetime and was going to be a famous SPC cop. Our program was the only Motorola program to win 2 CEO Quality Awards and a Navy Salty Dog Award. We broke the rules but we understood what we were doing when we did. I have never been great at following rules and really don’t care much what people think (obviously if someone I respect has an issue then it means something). If you want to be mediocre and do all the same stuff exactly as the book says and your peers do and not think – well mediocre isn’t a bad thing. Just remember when you do that you are easy to replace even with a cocker spaniel. If you want to step up, what is that Eagles song? All Ready Gone. “….then you’ll have to eat your lunch all by yourself.” There are worse things in life.

    Check out Mario Perez Wilson’s book Six Sigma (write that title down it is hard to remember). Mario and I worked together on this project, FMU-139. I left and he did FZU-48 by himself (actually with Jim Blanden another very intelligent guy people perceived as a Neanderthal-zero tact and he knew more about wine than the yuppies in management so it is easier to disparage than learn from him). The book is not about those programs specifically but the experience from those programs flows through to Mario’s writing naturally. You might find it enlightening.

    This has nothing to do with statistics but you need people like Jim Blanden. Mario as well but he is a very classy guy. Jim is more basic. I was driving to work and my car broke down across the street from Jim’s apartment. I walked over to catch a ride and both his cars were there so I knew he was home. Knocked and no answer. Knocked again with no answer. As I walked away I heard the door open and there was Jim naked and dripping wet (second floor overlooking a golf course). I said “Jim you are naked.” Jim said “of course I was in the shower,” I said “you could have put cloths on to answer the door.” Jim said “why would I do that? I am going to get back into the shower.” That is the personality you need around you to some extent to make change. People on the floor loved him. Management hated him but when he said turn right it was always a good idea to make the turn and then ask why. I have told that story in various situations. I always thought it would be a great interview story. I have found that about 99% of the people laugh. That other 1% that doesn’t laugh. Those are the interesting people.

    Just my opinion.

    0
    #240850

    tgause
    Participant

    @rbutler, @straydog, @mike-carnell – Thank you all for your responses.

    Gotta ask: You guys seem to respond to most questions I’ve seen posted on this forum. Is this your day job or something? Do you guys sit around all day answering people’s questions on this forum? Every time I’ve posted a question, I know I can count on getting a response from one, if not all of you (and I look forward to it!).

    Thank you for all you do and for sharing your knowledge and expertise with the rest of us! You’re helping to make the world a better place when you share your passion with others. And it’s apparent you’re passionate about process improvement. I look forward to talking with you again soon.

    Your humble student,

    “Grasshopper”

    1
    #240852

    Robert Butler
    Participant

    Well @Straydog and @Mike-Carnell that guy @chebetz has found us out!  I guess it’s time we ask Mike Cyger to tell Katie to change our screen names again. I think I’ll go with Dr. Theodore Sabbatini this time around.  :-)

    I can’t answer for Straydog or Mike but given our interactions over the years I’m pretty sure their answer will be the same as mine – specifically – no , the forum is not a day job and I don’t sit around waiting for the computer to signal me that someone has posted something new.

    One of my engineers pointed me to this site back in 2002.  I looked over some of the answers to basic statistics questions and couldn’t believe how wrong many of them were.  So, with vicious self-interest as a driving force, I started answering various and sundry stat questions.  I enjoy responding because this gives me an excellent chance to not only revisit some of my textbooks for details but also practice the art of explaining difficult concepts to non-statisticians.  I will usually drop by in the morning before starting work and again in the evening after work to see if there is anything interesting.  If there is and I decide to respond I’ll usually do that during lunch or, if it is short enough, during a coffee break.  If the response is going to be too long or if I want the think about it I will wait until after work to write a response.

    1
    #240854

    Mike Carnell
    Participant

    @chebetz, @rbutler and @straydog Just to do one upsmanship I found this site in 2001. I was being sued for violating a noncompete so I couldn’t work for myself. The guy suing me used time as a strategy and dragged these things out trying to run you out of money. It also gave the market time to forget you and move on. I dedicated 4 hours a day to answering questions. Abut 60 days later I was doing a speech in Las Vegas and a guy in the front row said “I know you. You’re that guy on isixsigma.” It was working. Here I am 18 years later.

    Up until about 2010 this site saw about 40 new posts per day. It was and probably still is the epicenter for Six Sigma. That and we are very reluctant to ignore @michaelcyger and @katiebarry. They are very sensitive. It was what Bill Hathaway at Moresteam called “Food Fights” and there was a group of us called the Usual Suspects (Stan, HeBe Geebe, Darth and Stevo) that Mike Cyger referred to us as his Jerry Springer factor. It was pretty exciting, my lawsuit was over (about 18 months and I was working again) but if I had to make a choice between watching TV or answering questions I would answer questions. The attached picture is from a iSixSigma Live conference in Miami. 4 of the five Usual Suspects were behind a lit screen – Stand didn’t show up well because he was on the phone. They did this at the end extemporaneously.

    Between being self employed and up until about 2012 I slept about 3-4 hours a day there was still a lot of time to answer questions. I am a huge fan of a lot of the people who post here regularly. I have learned more about statistics from Robert Butler than anyone else and read most everything he posts. I have made friends with people all over the world. I am even nice to Darth sometimes.

    I have a day job as a LSS consultant and a few other things. I tried to retire last year and figured that was going to kill me faster than anything else so back to work.

    That’s my story and I am sticking to it.

    Attachments:
    1. You must be signed in to download files.
    1
    #240971

    Geoffrey Withnell
    Participant

    tgause:

    In regard to spec limits on a control chart, one thing they most certainly do NOT do, is help determine if a process is in control.  A process is in control when the process inputs are all normal cause inputs, and there are no special causes creating unusual variation.  This has nothing to do with the specification.  A process can be in very good control, and be producing completely outside the spec limits.  Process capability determines whether the process results are inside the spec limits.  Once a process is in control, and the natural variation shown by that control is inside the spec limits, the focus should be on maintaining that state of control.  If this is done, the spec limits will be met.  The problem with putting the spec limits on the control chart, is that there is a temptation to ignore out of control conditions that do not appear to approach the spec limits.  This is VERY risky.  If the process is no longer in control, there are outside cause(s) at work, and statistically speaking, one has NO ability to predict the process outcome.

    0
    #241019

    Paul Keller
    Participant

    If process personnel adjust the process to meet specifications, and the process is inherently incapable of meeting the requirements (i.e. Cpk<1), they are tampering with the process, which Deming showed in Out of the Crisis actually increases process variation.

    It’s best to first analyze your processes for statistical control. If not in control, remove special causes. If in control and not capable, reduce common cause variation. If in control and capable, then use the control limits (and run test rules) as your guide for adjusting the process. That is, only respond to special cause variation, which includes run test and out of control violations.

    0
    #241031

    Chris Seider
    Participant

    Just ask yourself why are you wanting to do this?

    I would caution you against putting spec limits on SPC charts BECAUSE if you’re using SPC as an active tool to control the process (or why else put them in the “field”), then you have to be wary that the nuance of an XBAR vs I chart will be lost by most “operators”.

    Once they learn one way, they will think it works that way for all charts.  I’d hate to think someone down the road said, hey yea we let that batch of toilet paper rolls out because the Xbar was greater than the lower spec limit.  Jack in printing told me about his charts and said if his point was within the spec limits all is good.  …Never mind that Jack was using an I-MR chart…..the confusion isn’t worth it–in my opinion.

    If you’re wanting to add spec limits to your SPC charts you’re using for data gathering or analysis–it’s less impactful.

     

    My two cents….that and 5 bucks MIGHT get you a venti Latte at Starbucks.  :)

    0
    #241032

    Mike Carnell
    Participant

    @chebetz I need you to clear something up. In your post you spoke about adding spec limits to the control chart i.e. “What harm could come from including these two additional lines on a control chart (other than the possible confusion a reader could have and mix up the control and specification limit lines)?” My interpretation of that comment is that you are not running to spec limits. You continue to run to control limits and just have spec limits as “additional lines.”

    Now we are getting posts about running to spec limits? Which is it?

    0
    #241034

    tgause
    Participant

    @mike-carnell:

    To be clear, I was NOT suggesting that we replace the UCL and LCL with USL and LSL. I fully understand why the control limits are there, how to calculate them, and what they help tell us, the readers. They are a must on a control chart.

    I was wondering what harm, if any–other than possible confusion on the reader’s part–would come from including specification limits in addition to the control limits on a control chart? In other words, there would be 5 lines on the control charts:

    • the mean – required
    • the two control limits (UCL and LCL) – required
    • and the two spec limits (USL and LSL) – optional

    By having both the control and spec limits on the chart, wouldn’t we be able to determine not only if the process is in control, but also if it’s meeting customer specifications?

    0
    #241037

    Mike Carnell
    Participant

    @Chebetz That was my interpretation of your post. What you are seeing is the typical reaction to doing something that doesn’t match what people put in books. I have always had issues with rules to some extent. This is paraphrased but “I didn’t care yesterday. I don’t care today. I won’t care tomorrow.”

    I have always been more concerned with results and understanding risk. Adding a couple lines on a graph doesn’t represent any risk as long as you explain i.e. train people what you are doing and you don’t just walk away and provide some support.

    It has been a very long time since I have launched a new production line. When I start up I run control limits at 95%. I am driving more out of control points because I want who ever is supporting the launch looking at everything. You cannot believe how much flack I took every time I did it. You know how much I cared? Zero, nada, zilch. It was my line and my responsibility and could care less if they had to spend more time on the line than they liked.

    This is your project. Understand the risk and adjust for it or if it is to large don’t do it. I don’t see any actual risk to this other than you are going to have some idiot with a stats book regurgitating stuff you learned a very long time ago.

    Just my opinion.

    0
    #241837

    doe
    Participant

    Customer calls (like an automaker) and says: “You can’t use spec limits on your control chart”.  The same customer calls and says: “Your parts won’t fit, get in here and sort them”.

     

    Your replay should be:

    “I won’t sort anything because my process is well within control”…then hang up on them.

     

    Said elsewhere in this message string, there is no statistical jail that will confine you should you use spec limits in charting. In the example, above, customers want “every part” to meet…spec limits…not control limits. Therefore, if they want / need statistical part control use spec limits on your charting.

     

    Use pre-control charts. They satisfy both schools of thought.

    0
    #241945

    Chris Seider
    Participant

    @[email protected]  Did you really state you’d tell the customer you wouldn’t help them out and give them a stupid phrase of “but my process is in control”.    Have you considered that something downstream of your process is impacting the product OR that variation appeared outside of your typical subgrouping that you’ve designed your control charts around.

    I guess you’ve never acted as the company rep when having to investigate an issue with your product/service.

    1
    #242046

    Mike Carnell
    Participant

    @[email protected] The first time you hang up on an Automotive OEM you need to pack your stuff when you leave that night because they will have you fired by morning. That is the worst advice (I am seriously wanting to put a string of expletives in here just to make sure you understand how bad that advice was) I have ever seen provided on this forum and I have been posting here since 2001.

    I am the one that said there is no statistical jail but at no point did I say you controlled to the spec limits which is what you are suggesting with PreControl. First most automotive companies are looking for a Cpk of 1.33 minimum. If you use PreControl you will end up controlling to a Cpk of 1.0 (you are letting the process run rail to rail so 1.0 is the best you will get). Precontrol is a Shanin technique. That will have you in trouble with a lot of people in the automotive business that are adverse to those techniques although most don’t understand why.

    Should anyone need to sort product at an automotive OEM call Jeff Glowacki at Quality Liaison Service of NA out of Nashville. He is great at getting good people to sort parts and he is very good at dealing with irate customers.

    0
    #242354

    Darth
    Participant

    @mike-carnell   It is nice that you remember old Darth.  It’s just that I can’t recall you ever being nice to me.  You, Robert and Chris sure are active again posting sometimes useful information.  Has anyone spoken to Fake Mike to see if he is interested in revitalizing the Forum or getting back into the Conference business?  Even the Brady Bunch is doing a retro reunion.

    0
    #242356

    Mike Carnell
    Participant

    @Darth remember you? How cod we forget you. Without you here we have to answer a bunch of homework questions.

    I could be more active except my computer will not boot. That would be my LENOVO computer will not boot up. It is dead as in it responds to nothing. That was my LENOVO computer. Maybe someone from LENOVO will read this since they do not appear to be straining themselves making computers.

    Fake Mike is out running across the state of Washington. Considering all the nonsense that goes on here i would be running for the border as well. Idaho has to be looking pretty good about now.

    The days of the DF being the epicenter of the SS world may have passed us by. Nobody worries about fixing anything but everybody wants a certificate so HR will let them apply for a job. Actually we get some decent questions here. We have hit the point where a mutual aquaintance of ours sat down one weekend day in the morning and by that evening had 5 BB certs and 1 MBB cert. Not even worth throwing a chair about this.

    The good news is Katie is still with us.

    0
    #242357

    Katie Barry
    Keymaster

    @Darth Here’s where you can learn more about Fake Mike’s run! https://mike.run/ Here’s the direct link to his Instagram for the run: https://www.instagram.com/mikedotrun/

    As @Mike-Carnell mentioned, I’m still here! :)

    0
    #242369

    Darth
    Participant

    @KatieBarry, @mike-carnell

    Jeez…..hotels, restaurants, Uber….what kind of woosie run is FM doing???  How about sleeping out under the beautiful Fall sky with the sparkling stars….hunting and gathering his food and fishing from the running rivers and drinking water from the mountain fed lakes.  That’s what Real Mike would do if he didn’t have to worry about his crappy Lenovo computer.

    Wow 5 BBs and 1 MBB by the time Saturday Night Live comes on…..that’s a heck of a productive workday.  Betcha he didn’t have a crappy Lenovo computer to contend with.

    Katie, you were always the heart and soul of iSickSigma and everyone knew that.  Just wish we could somehow recapture the magic and incredible contributions of the early days.  There are enough of us that still care about the Way of the Ancients and the Path of the Unusual Suspects.  LOL.

    0
    #242380

    Katie Barry
    Keymaster

    Katie, you were always the heart and soul of iSickSigma and everyone knew that.  Just wish we could somehow recapture the magic and incredible contributions of the early days.  There are enough of us that still care about the Way of the Ancients and the Path of the Unusual Suspects.  LOL.

    Aw, @Darth — Back for a day and you’re already my favorite! :)

    0
    #242381

    Mike Carnell
    Participant

    <p style=”text-align: left;”>@Darth none of that hard Corps stuff these days. I babysit grandchildren. That makes Six Sigma deoyments look simple. The up side is dealing with all the Champions over the years has prepared me very well for trying to reason with my 2 year old grandson.</p>
    The Lenovo was laid to rest today after lasting about 18 months, and after we had opened it up and removed the flash drive. 25 years and my first Lenovo and it dies. It hasn’t even been on one international flight. Not what they used to be.

    So FM is running across Washington in an Uber? I’ll bet that is some tough training.

    0
Viewing 22 posts - 1 through 22 (of 22 total)

You must be logged in to reply to this topic.