FMEA makes Six Sigma Obsolete!!

welcome to complexity made simple my name is paul allen and before we get into today's video well i'm just going to remind you the three books that are on sale drink tea and read the paper if you're a green belt in a black belt and you want simple instruction on how to apply your skill design of experiments for 21st century engineers and finally a statistical process control for small batch production they are all available from lulu.com and the links are in the video below welcome to complexity made simple my name is paul allen on the subject of today's video newsletter where we're gonna go and take another look at fmea my last video just about 10 days ago i was talking about fmei and i showed you how to fill an fmea out correctly detailed high fidelity solution well today what we're going to do i was i was at a six sigma certification event uh just a few days ago and this project was presented to me and when you see it it's a very simple project it's a very simple problem uh the company is quite small there's there's about there's only about 50 people in the company they make electromechanical assemblies typically and what you see is a very practical simple problem that they had very typical of a 50-person small and medium enterprise in the uk um so it's not the biggest problem in the world but what they had to do because they allowed disaster to creep into their process because they allowed chaos to creep into their process they then had to spend five months correcting the disaster that they'd allowed to be created and this is a great example of it's a simple process but if you treat it like that if you treat it with carelessness if you treat any process carelessly you go at it half-assed and you don't give it the respect that it deserves it will cost you fortunes now you'll see that this when we look at this presentation it's a very simple thing to fix so what we're going to do is we're going to take a look at the presentation and then what i'm going to do is i'm going to say well what if we'd have done an fmea let's show you what the fmea would have looked like for this process and how it would have made you sit up and go no no we need to take this process seriously it's about to cause some really serious problems in our organization so this is an example of why six sigma actually should never exist these three month four month projects should never exist in organizations they only exist because you design disaster in and when the pain gets big enough and painful enough you go back and you treat the process with respect but if you used fmea and you used fmea at the beginning six sigma would never exist so what this is a video about is how to put people like me out of business i don't want to keep talking about six sigma i'd love to talk about why fmea is a brilliant tool let's take a look at the presentation okay so here's the presentation i've changed the name of the company just to um to give them a little bit of privacy on this um you'll see it's just been it's just been presented just a few days ago and the process the project took them a few months four or five months i think in the end to uh to sort out so let's take a look at this thing so what's it saying process the process cannot be controlled um in relation to machining of some plastic extrusion and you'll see this is a very simple process they are just taking a simple plastic extrusion they're doing two simple machining activities at the end of the plastic extrusion they're not going well and they're creating problems and it says here look it's creating variability with the product quality and something to point out the actual plastic extrusion costs about 20 pence the final product item costs about a hundred pound it is turning up the problem turns up in the finished item the problem does not turn up in the 20 pence channel that they should have machined correctly it only shows up at the end and when it shows up at the end they have to throw the the whole item away a hundred pound goes straight in the bin they then also get inaccurate they get inaccurate stock records so they run out of this this simple plastic extrusion because they're using much more of it than they need to so lots of costa pol quality going on here that just isn't isn't needed now here's 50 days of the reject data the the main graph the big the big peaks appear this is the production quantity so this is this is how many they made the graph we're interested in though is this thing down here because this is the this is the reject rate you can see the the data on the left hand side and if i just blow this up it probably makes it just a little bit easier to see 18.5 defect rate so that's what we've got currently an 18 defect rate but very variable so could get could get a lot worse than 18 18 on average i'm just looking at some of the numbers there's one here look 42 there's one here 55 percent 28 percent yeah 28 so some of these 53 you know when you're getting defect rates like that you just cannot control anything around this process so with a lot of these solutions a lot of these chaotic problems what do i encourage my client to do well i encourage you to go process flow cause and effect ncx slp and we want to get the chaos under control that is what most six sigma projects are about getting chaos under control now this is the process you can see look that it's a simple piece of plastic channel you could probably see better in this in this photo here um if i if i blow the picture up like that you can see a little bit better but they're machining a sort of a an area here this flat spot here needs to be machined off and the other thing they do they machine a slot which you can see in this photo here you see that semi-circular slot in the end and they're doing it by hand so they have a dremel in their hands and they are doing it by hand with skill process currently takes 83 seconds and it's got an 18 18 defect rate then they go and take a look at the the input variables to the process so this is a very standard approach what are all the inputs to the process because inputs control outputs so now we have a more detailed problem statement that they're talking about so you know no control over machining the process 18 defect rate 63 second process but 20 seconds worth of clean up time etc we've also got health and safety issues with this manual process so not great and then of course the the uh the objectives well simple things like reduce the cycle time by 50 percent take the reject rate down to zero uh somewhere it's probably going to say does he say health and safety did a little it doesn't doesn't mention health and safety but really really should do but anyway there are the obvious the obvious objectives for this project any other variables that we saw earlier and because we don't have control here in large swathes of these inputs when you let inputs like that be all out of control what do you create chaos you create an unreliable unpredictable process known as chaos and that's what they've allowed to happen because they treated this simple process carelessly they didn't come up with a high fidelity solution detailed planned under control a controlled process not a chaotic process we need high fidelity solutions for for world-class businesses so now they start getting into controlling some of the variables and this first one is uh calibrating the source so this is a calibration process this is purely about cutting the uh the piece of channel to the right length so they put a piece in here which is exactly 100 mil and then they calibrate it to 100 mil this little yellow box of tricks here which is on the saw has a tendency to drift out every day so every day there is a little problem with it and if they don't calibrate it every time they use the saw they are likely to see error in the cut length of these extrusions then they don't fit together properly and they get all sorts of problems so we've got a calibration routine then they need to get control of all these manual dremel chaos using skill to try and get around the problem and basically all they do is they buy something a little bit more industrial and what do they do they jig it and with a jig you end up with two uh places on the jig stage one and stage two and stage one machines the uh the surface the bottom of the channel where there's a little little cutaway section and area two when you push the channel in there you will machine the slot in the bottom so very simple solution here it actually is in action uh if i blow it up and get a little bit get a little bit closer look one of the reasons why they put these examples on the top is because people started putting the channel the wrong way around you can see the channel the open end of the channel must face towards the right to do the one process it must face towards the bottom to do the second process and people were misorientating the um the piece that wasn't clear know that they have an example it's much it's much clearer um anything else to say yeah you can see the two i've pushed it in the top there so you can see they've pushed it in here first time and then they're pushing it in the bottom for the second time and he does the two it does the two cycles so very simple and by the way just to point out they wanted to get remember 83 seconds was the original cycle take 83 seconds they wanted to cut that in half they've smashed it uh enough 75 percent of it now that's a really good example people talk about quality costing you money quality doesn't cost you money quality makes you money because when your process gets reliable and consistent and delivers every time all the time you cannot have processes like that that are slow and sloppy now a process that's on it doing the job every time all the time will be fast efficient and they will make you money quality makes money and here's another great example and then of course this these are all these are all maintenance these are all maintenance routines that i've put around the saw and the um and the dremel what you tend to find is if you don't do maintenance correctly that is where variation starts chaos starts from maintenance if you allow sloppy maintenance if you cut across maintenance routines chaos will just enter your process in huge quantities and money will pour down the drain maintenance is super important so they've got they've got things about maintaining the the saw blade keeping the cutting fluid etc up to level making sure that the air pressure's right this is all on the saw by the way so they've got maintenance routines in place they've put a jig in place to take all the skill out of the machining processes and now if we take a look what's happened well the reject rate effectively has gone to zero these terrible results up here which are showing in the first few and the results that you're seeing here this was because they got the orientation so just something had gone wrong the orientation wasn't obvious and once they made the orientation obvious what's happened the defect rate has gone to zero fantastic here's what they achieved they're saying the defect rate is 3.8 but quite honestly looking at it now it's it's stuck at zero pretty much they've achieved what they what they wanted to achieve um cycle time reduced by 79 percent save 3 000 pound um by the way the defects look 10 000 pound per annum this has been going on for three years this had cost around 30 000 pounds can you believe that because we didn't have a high fidelity solution we just decided ah all we have to do whenever you hear an engineer say all i have to do is if you hear that phrase all they have to do that is a red flag that phrase is going to cost you thousands of pounds and in this case that phrase costs this company 30 000 pound because some engineer went at the process half asked and just said all you have to do is dremel the end it's easy and no it wasn't we were relying on skill which cost 30 000 pounds the jig cost them a few hundred pounds they could have saved this money if they'd have gone for an fmea first as last if they'd have treated this process with respect so here is where they've got to you can see now that all of these things have got c's next to them they have a control in place they have a low defect rate high speed process high quality low cost get it high quality low cost brilliant yeah so that's exactly what we want that's what six sigma is about high quality low cost okay so there's the presentation it's it was a simple thing wasn't it but it was a simple thing that's cost them thirty thousand pounds now as i said at the beginning of this video none of this disaster needed to happen all they needed to do was to treat the process with the respect and get the engineer and this is what i'm talking about the engineer does the fmea nobody else you don't have a big team of people spending hours and days wading through paperwork and creating bureaucratic fmeas now now this is a is a document that helps the engineer design a high quality high fidelity well-controlled process right from day one so let's take a look at the fmea document that the engineer should have completed before this process went live let's take a look okay now i've pre-filled out i've pre-filled out the fma just for speed of this uh presentation um but if we go through the um if we go through the boxes and i'll just just maybe indicate them as we as we talk about them we'll just just sort of color them in gray so the first thing to say look is we've got the process step over in the column a of the fmea form so you can see the first box look cut to length so in this side of the of the document we've got the process step in every case cut to length using the dremel to hand machine the recess using the dremel to hand machine the six millimeter semicircular slot now this is the process that was in that presentation this is the original process the way that it was it was intended to go ahead so this is the engineer thinking all we've got to do is give him a dremel and let him get on with it so we're going to let him design the process just the way he was going to okay look so let's go cut to length what's critical to quality now a reminder about the critical to quality critical to quality is the input to the process it's the input to the process that's really important so what's what's critical to quality about cutting the the channel to length well that the blade is sharp and that the calibration is of the scale is done then we go so having looked at those two let me just color those in and then it says if we don't have sharpness what will we get well if blade sharpness if the blade sharpness goes wrong failure mode now this is again how can the input fail so the input was blade sharpness how can the input flat fail well the blade goes beyond each usable condition that's the failure mode for the input side this is really important inputs and outputs if you get these mixed up the fmea doesn't work properly it breaks down and it becomes such an unwieldy uninvaluable unvaluable useless i suppose is a better phrase such a useless tool if you get the inputs and the outputs mixed up so this is really important inputs control outputs failure mode how can the input fail the blade goes beyond its condition what will happen the failure effect well now we get heavy burs and a hundred percent reject rate i've given that a severity rating of seven now something that's really important at this point the severity rating have a clear scale now i don't have a scale here that i'm referring to for why i've picked seven but you should write down your and it should be your language your definition for your company so your engineers can easily choose that but to me something that results in 100 rework is going to be high on the on the severity scale rework is definitely not allowed ever and therefore if you're getting it 100 of it it's going to be in the high scale as far as severity can is concerned then it says well what's the cause well there's no indication of blade light so nobody's ever said what the blade life is we have no idea how old the blade is etc etc so no indication of blade blade life then it says occurrence now my occurrence is on one to four scale let me show you my occurrence rating that i tend to use here is my occurrence rating so you can see hello how how likely are we to make the mistake are we able to make the mistake once a day once a week once a month once a year now you can add to this if you wanted to so if you want to add more categories of course you could go once every two years you could go once every five years you could add maybe you know once an hour up here if you really wanted to if you want to try and make 10 categories you definitely could but the reason why i've used this language the reason for it is to make it easy for you to judge don't forget you're often trying to judge a process that doesn't exist yet how can you judge how likely you are to make the mistake you have no evidence of reject rates or anything at this point so you're asking the question well how likely are we how often would the blade be blunt and you could be going here i think i've decided a three so i've decided given how often we use this tool the blade is going to go blunt once a week and if there's no rules we're going to start making this mistake once a week and it's easy to make that judgment whereas if you start to say is it a five percent defect right at a ten percent defect rate or a forty percent defect rate you have no evidence but if you say yeah we're more likely to make the mistake once a week we'll more than likely make the mistake once a month it's really easy to make a a good guess because that's what you're doing here you're making a good guess so get your your scales so that you can tune into them and there's no arguments very little argument clear and unambiguous these the language that you need to use here needs to be clear and unambiguous so let's go back to the fmea form okay so no indication of blade life i've scored it to three and then he says what's the current control well there is no control what we do is we run to disaster what i mean by that is we run until we get the problem and the problem becomes so acute that someone eventually says i can't use this anymore please change the blade you run to disaster and that to me is a 10. no control whatsoever and you can see if i work out the risk priority number 7 times 3 times 10 i get 210 now for me the the worst i can be with my scoring system is 400. you have 400 so for me above 200 is going to be something that is clearly actionable so i'm above 200 the box goes red let me show you by the way the scored in routine for the chance of seeing the error so the controls at the end there that i gave i gave it a 10 no chance of seeing the editor talks i haven't got any controls at all you can see the language that i've used is based on the control you are using so i've not tried to say sometimes they say there is um that there is a um you know a 10 chance the problem will go undetected there's a 20 chance the problem will go undetected there's a 30 percent there's a 40 i have no idea how you would score those numbers but if you say i have a mistake proofing device fitted you can easily say oh i can therefore score it a one i have a mistake proven device fitted but it needs an alarm type so i'm halfway into making the mistake then i get an alarm that would be level two i have daily tpm routines weekly tpm routines spc i might actually swap those two over looking at it at the moment i think those two should be the other way around but you can see that these are clear and unambiguous scores if you have spc in place to control the problem you straight away just give it a five there is no argument so in my case i give it a 10 because there was no controls so the whole thing now we've gone all the way across and that's telling me to do something so i've got to put a control in place to control the sharpness of the blade so action would have been taken to redesign that process the scale calibration what have we got well the scale reads the incorrect length that's how the input can go wrong all the channels will be cut to the wrong length that's an eight this is getting more painful now we know that the electronic scale is a daily error built into it so every day the scale tends to be out slightly so for some reason the piece of equipment isn't particularly stable therefore we're liable to make the mistake daily so now i'll give it a four and we don't have any controls in place and again i'm scoring it 320. now on a scale of 1 to 400 320 is a big big problem so we would be doing something about that what else have we got let's take a look using the dremel for the hand machining the recess what are the variables tool condition would be one of them skill of the operator is the other one they're the two inputs if we don't get the tool condition right we'll get rough finish now i've called this a minor issue so i've scored it very low because the surface finish isn't that important on machining the recess so this one look because it's got a low severity etc has come in at 60 even though i've got i've got now control it's a run to disaster type problem the actual problem that it causes is pretty minor so it's it's green i'm in theory i'm not going to do anything so all of that go across there that's all being dealt with then we've got the unskilled operator machining the recess too long what would happen if they get this wrong so we've identified that well what it will do the resin that we put in the product will run out and it ruins the finished product it multiplies the error by about ten thousand percent i'm going to give that an eight what's the reason well we're using manual scale and control what's the current controls nothing at all we're using skill therefore 320 it's a disaster waiting to happen now i'm not going to finish the rest of this um fmea off but if you look at this if you look at this it would be telling the engineer if you allow this this poorly controlled low fidelity solution low fidelity you're just going all we have to do and we and we think that's going to produce a good result by saying the words all we have to do if you allow this low fidelity solution to go forward you run the risk of high reject rates lack of control on the computer in terms of stock accuracy all kinds of disasters are going to befall you by the way those disasters have been going for the last three years in this company on this very process had they filled out the fmea and all i'm asking them to do is to fill out the fmea personally use the tool to help the engineer design high fidelity process controls then the process would have gone right first time this project never would have existed never would have existed and it didn't need to exist did it if you look at it it's such a simple solution and yet they allowed it to happen that's why fmea is so important think about a process that's more complicated than this that you're not doing fmeas on the fmea is a money-making monster because it allows you to go live with a new process right first type and that's why you should be doing it and not doing six sigma projects now there's my example six sigma versus fmea i know which one i'd sooner do even though i love six sigma fmeas are the way forwards the way forward and if you use fmea you will definitely make more money now if you wish to get more detail on this you can buy my book drink tea and read the paper if you go to lulu.com you can get it either as a paperback book or as an electronic download in there it shows you fmea and it gives you advice on how to create a scale which is clear and unambiguous how to fill out each of the boxes making sure that the inputs and the outputs are in the right box and just shows you how an engineer should use an fmea to make more money [Music] you

Channel: Paul Allen

Views: 555

Keywords: Six Sigma, FMEA, Process thinking, SPC, Mistake proofing, MSA, DOE, TPM, TPS, t test, p value, hypothesis test, Standard Deviation, https://youtu.be/QH984PnwRDE, https://youtu.be/AGJ1QYI2B4c, cpk, process capability, https://youtu.be/f_fjqCpd67Q, https://youtu.be/gsD8V2_eZ0A

Id: o_tTEYOTej0

Length: 32min 4sec (1924 seconds)

Published: Mon Sep 27 2021