In the following example, we go over a welding defect issue with our example company, and figure out how to solve the problem using Six Sigma. If you have any questions about this Six Sigma Complete Project Example, please leave a comment below!
At its core, six sigma is really fairly simple some people tend to over complicate it and that’s a shame. It doesn’t need to be that way. What I’m going to do in this video is walk you through a complete Six Sigma project example from beginning to end to show you how the tools work together and how a simple project might unfold.
Now what I’ve chosen for this one is a manufacturing example, but if you’re doing a business transaction or business process example it’s really no different.
The metrics change because in production quite often it’s scrap rate, but in business processes it’ll tend to be things like error rates and cycle times and loss rates and things like that. So let’s look at the information that generated the project and that would be this. I’m just going to pull that right up hide behind it and what we have is we have a day shift and a swing shift going and on each of the shifts we have Elbonians and we have Mudians. I think the Mudians are from the country where we important Muda. I’m not sure but as we look at it on this nested Pareto chart which is a patented feature of quikSigma, we find that the Elbonians on day shift or having a rather bad reject rate as compared to the Mudians and the Elbonions in other settings so that’s the basis of the problem. That’s the chart that launches the project right there and I can look here and I’ve got in a week I have a hundred and eighty-eight defects on day to day shift with the Mudians as opposed to 64 here and 72 here and only 45 here so big difference.
So how might I start a a Six Sigma project? Well the defect we’re seeing is in the weld quality and so up here we define who’s going to do the project is the old who, what, when, where, why formulation. Who’s going to do this well? We’ve got a project leader. We’ve got some people that are going to serve on the team. We have a project champion who is someone whom the organization that manages a group of project team leaders, and we’ve got a mentor supplied whose job is to help with the technical to help and we’ve recruited somebody from finance to give an unbiased opinion of the quality of the project. So we have to make a business case. Why are we doing this?
Well, defective welds on our Shire 55 product cost us about 87 dollars each to rework and the rate of defects is higher on the day shift with the Elbonian workers and that’s what our Pareto chart told us. If we can reduce defects there to normal and that’s compared to other instances in the company, we will reduce our production costs by over half-a-million dollars a year. Now if you are on the elevator with president of the company or the CFO and he asks you why you’re doing the project this is the statement you give him and if that kind of a statement doesn’t get his attention well I’m not sure what would so we have to then state our smart objective. We have to define done by june 30th of this year reduce Elbonian day shift well defects from a 188 per week to 65 per week without compromising weld strength or worker safety. So these last two the first part is the main objective and then the weld strength and worker safety or what we call side conditions. We have to make a statement of scope, that is does this include everything everywhere?
No, in fact we very specifically limited ourselves to the Shire 55 model and the Elbonian workgroup on the day shift and we have to choose an indicator for our process and i’m going to choose weekly defective welds now if you want to refine that a little bit you could make that a percent because production goes up and down the defects would usually go up and down as well as you increase the volume.
So baseline right now we’re at a 188 per week, our goal is to get it down to 65, which is about were the other bars are and entitlement just simply says that if everything went perfectly and we kept working on this a long time we think we might get it down to 40. So then I have to make a statement of why we’re going to do it. Well the financial information is that I’ve got a cost-reduction I’m going to reduce my cost of goods sold. That’s going to fall through to the bottom line and here’s the arithmetic down here that says that that’s what’s going to happen a 188 a week I’m going to drive it to 65-87 dollars a piece 52 weeks a year and it adds up to a little over five hundred and fifty thousand dollars.
When I’m satisfied that I have my charger complete I can mark that is done and when everybody signed off that this is the project we’re going to do I can mark that is complete and the defined phase is then complete. The next thing we might want to turn our attention to is the measurement systems analysis and we have a couple of tools that we can use.
Well I’m going to cheat and take a little shortcut down here under MSA complete I’ve got a little box that i can say well why isn’t there a measurement systems analysis and I’m just simply going to say see the gauge consistency study done as an I-MR chart and all I’ve done there is this little gauge consistency study and what’s happening here is that I have taken the tools that are used by the Mudian day shift and swing shift in the Elbonian swing and a
shift and I’ve simply made a dozen or so measurements of a standard item that’s about the size of our work piece that we work on and just take a look to see if everybody’s getting about the right answer and if they’re consistent about it and we see oh, my goodness, certainly appears that we’ve got a problem here on the Elbonian day shift and you might start to get excited and they go I got this one solved.
Well probably not and what we found here is a problem that does need attention and that we’re going to fix but the specs on the item are fairly broad and this is not likely our core cause. So at this point we know what we’re going to do what we’re trying to do and we checked our measurement system and it’s apparently good enough to give us answers that are meaningful.
So now we’re scratching our head a little bit wondering what to do next. Well it’s logical at that point to turn our attention to the process map and what I’m going to do is just blow this up to full page so that you can see it and what we have over here is the CTQ is critical to quality. What are we trying to do? Well we said in our project charter that we wanted to reduce the defective rate and we wanted to maintain worker safety and maintain weld strength. Okay, so some of you out there undoubtedly are welders, I am not. So I made this up and you can probably improve on it but it’s a teaching example.
Over on this side we list our key process output variables. These are the meters that we would like to watch to see how each of our process steps has gone. Over here are the key process and input variables. These are the knobs we would like to be able to adjust or that we can adjust to make the process come out like we want it. So and that’s the whole point is to get these CTQs and the knobs we can adjust. So I go through and for example down here in the performing the weld things that I know a little bit about is the wire feed feed rate and the type of the wire and if you’ve got some shielding gas that you’re flowing over it why how much of that and one thing that might be important sometimes is the time since the surface prep was done got a surface prep step back here and if some processes if you allow too long to elapse then you will have difficulty getting a good weld.
So having collected that the next issue is which of all of these input variables is most important? Now we could go through, in fact I’m going to shrink that down a little bit, we could go through if we wanted to and we could analyze each of those input variables but that would be very costly and there’s no reason to do it. There’s a really well established rule, not law but pretty good rule, that most of the problems come from a few variables. Alright? So what we need is some simple tool for assessing which one’s of these we might spend most of our time on and that would be the cause and effect chart and here are my CTQ’s these are automatically carried forward for me.
Here are all of the KPIVs that I collected. Those are automatically carried forward for me. I assign weights to each of these CTQs, ok, and then I fill out this matrix. I can put in 0,1,3 or 9 depending on how much this input variable influences this CTQ, ok, and then I just hit calculate and it does a little bit of arithmetic and ranges these in descending order of importance. So it’s just a rough quick-and-dirty selection of what’s most important. Now instead of having to do a few dozen variables, I do a few variables.
Makes life a lot simpler. So I want to study these in detail and so I might logically carry those through to my FMEA. Well, as long as I’m here I might as well
check off the process map and I might as well check off the C&E matrix and the Pareto chart is done. We’ve got some more stuff to do so I’m going to look later with this so I’m going to leave that unchecked. Now, if you notice here, the heavy variables are cleaning procedure, time since surface prep, the material physical properties, and a couple of other things. So let’s go to the FMEA and see what’s been done for us there. Well what’s been done is the variables that we have designated as important, have been carried through and I’ve got one of these sheets for each of my important KPIVs and I’m looking at that and I start to fill this out and you know if it’s been too long since surface prep then I can have surface oxidation, I have poor welds, and I’m going to give that a severity of seven and now I’m kind of stumped because now I have to tell how often this happens? What’s the occurrence? I don’t know.
Well long before the people in our company ever heard the word gamba which is really fine concept we used to say, go and see. Go look at the process so I’m going to go back to my trusty old by I-MR chart which tends to tell me a lot. I’m going to go over here and i’m going to look at the Elbonian day shift defectives and these are the number of defectives by day. Here’s Monday, Tuesday, Wednesday, Thursday, Friday, Monday, Tuesday, Wednesday, Thursday, Friday, Monday, Tuesday, Wednesday, oh my goodness, anybody see a pattern here? In fact this was exactly what was happening at our client.
As we look at this made-up example what happened is the Elbonian day shift fellows were a little faster and what they would do is they would do the surface prep for Monday’s batch on Friday afternoon and leave them over the weekend. The film of oxide would form not even something you could see but it definitely did affect the weld.
So this problem is very quickly coming unraveled here. I’m going to check, I’m going to look, that’s not done. Let’s get back over here and do that. This is happening a lot. This is happening routinely. So I’m going to give it a fairly high occurrence number and you know it’s just everywhere I check, I’m going to give that a seven and i get an RPN number that says I should maybe take some action, well this is fairly obvious, isn’t it? Change the procedure so the welding takes place quickly after surface prep. Ok, and then I can set my little carry forward arrow and at this point I’m starting to suspect I’ve solved the problem that looks really really likely.
So I’m going to go down to my action plan. Let’s, I should check off my FMEA even though I haven’t done everything yet but for a classroom example we’ve done enough. And here’s my action plan that’s automatically carried forward when somebody clicks the little carry forward arrow. Some of these get populated for me. So then the action is going to be change the procedure so it takes place quickly and we’re going to assign that to someone in production engineering and that’s due in a couple of days and then we want to describe the results of that.
Well let’s go back here to our trusty I-MR charts and this time I put a specification limit down here. We wanted to get down to 65 per week and that’s 13 per day and I can look at that pretty easily in an I-MR chart and that says that I’m running an average of two and a half a day. We’ve really succeeded very nicely. The process is stable and predictable.
According to this evidence I can do this over and over again and get the same result. And I can just look at the specification limit up here and my actual production is a long way from the spec limit and in fact those of you that want to look at something numerical I’ve got a PPK down here of 3.42 which says that I’ve succeeded wonderfully wonderfully well and I have very good evidence that the problem has been solved.
Now the next thing that we need to do is draw this to a close. We have to have a control plan so that whatever gain we make doesn’t just simply drift away. Well there’s nothing complicated about a control plan and sometimes we get carried away and we have layers and layers of of that kind of thing. And you’ll notice that i skipped the whole bunch of stuff. Hey, when you’ve solved the problem declare victory.
These tools are not hoops for you to jump through. These are tools to help you get to the conclusion and when you get to the conclusion, declare victory and go home. So I’ve only got one item on my on my action plan and I checked the little carry forward arrow and that will bring up the control plan and that will carry forward the first couple of columns in your control plant. That is, what’s the variable that you’ve adjusted? And what improvement of you made? And what error indicator? Well, in this case what we’ve got is a little poka yoke kinda thing, where when the operator finishes the surface prep on a batch of parts he sets a little timer and those parts have to be welded before the timer goes off and that ensures that we don’t have this problem. So whether you’re working on a business process problem or whether you’re working on a production or an engineering problem, the principles are the same. The variables that you’re trying to get to work ride change and each tool then feeds the next and we build decision upon decision until we come to the right to the right
conclusion. Hopefully you can see that this is really fairly simple. Yeah there are some just wonderful advanced statistical tools that we get to use, but not nearly as often as we’d like to most of the time the basic tools will solve the problem and that’s all that’s required.