Failure Modes & Effects Analysis (FMEA)

Good afternoon, we begin again and this time I am going to complete my two lectures. One lecture is just over, the second one is going to be on FMEA. And in this, we take a look at how we can predict the rate of failure of certain products. These could be finished products or these could be for example, a process. So, this is quite often utilized right at the design stage, right at the design stage itself, you would like to do FMEA to try to make sure you can prevent, you can take all these preventive steps to look at the manner in which a product or service may fail. The outline of this hour's talk is going to be what is FMEA, little bit about the history of it, what are the benefits, what will be the applications of FMEA, this is something you would like to be able to see. The procedure for doing it, I will take give you a glimpse of the work sheet, which is there and couple of examples and a summary, this is what my plan is for this session with you. What exactly is FMEA? It is a technique to try to improve system performance or to try to design systems that would resist failure, both of those are achieved by doing FMEA. Why do you do FMEA? Well, first of all, let us try to understand the definition and the purpose and the benefits and so on. I should just caution you that FMEA is a team approach, it generally cannot be done by just one person trying to work out the details of it. Something you got to remember is, the whole purpose for FMEA is try to reduce, is basically try to reduce the impact of failures and in doing that, what it really does is, is prevents a lot of crisis. If you would not do FMEA, then the design or the process would be released as is without any kind of you know kind of prediction of the chance that it might fail, it may fail to function as required or it may fail to deliver the function that is the performance that is that is again expected of it. That is something that would be there, if you do not do FMEA. In FMEA, FMEA is kind of a preventive predictive technique, that is what it does and it leads to actions, it leads to actions that will lead to improvement in performance. What exactly is the result? Well, there are two dimensions, whenever something breaks down or some failure takes place, there are generally speaking two or three things that happen. So, I am going to show you that slowly. The first thing of course, is there is a severity of the impact that can be high or low depending on the nature of the feedback, nature of the failure that took place. You might get some feedback saying that the impact was high or the impact was low. The other aspect is the probability of it, the probability of it occurring that again can be high or low. So, if you got a situation, when you got a high probability of failure and the impact is large and that the consequence, the adverse consequences lead to things that you do not really want, then you got a high risk situation and that is that red area there. If you look at the, if you look at the picture there, there is that red area and then actually says, if my product ends up there, if my process ends up in this area, I have got to make every effort to either reduce basically the likelihood. So, I should try to push it down this way or I should try to take some preventive steps. So, I can, I can reduce the impact. The whole idea is to shift a process or a system which is here, more to this side, more to this side. As you go from here to here, basically you are trying to reduce the impact, also trying to reduce the likelihood and this is achieved by doing FMEA. This exactly, this objective is achieved by doing FMEA. Let us take a look at the output. What exactly happens at the output? Now, if we look at the system for a few minutes, for a couple of minutes, you will find there are certain components in the system that has gone through, this particular system has already gone through some kind of FMEA analysis. So, the component names are given there, the components that make up the system, then failure modes and what different ways could that component fail, for example, loose wiring or a failure of the plug or something like that itself. What are the causes of that failure? It could be vibration or something else, that also is pinned down there, then the effect of that failure. Causes of failure would need to, then again you got something called the effect of the failure, then the correction of the problem. What all things I can do to try to reduce either the impact or the likelihood of that thing and then I put some general comments there and this is what I do with every component that makes up the system itself. So, this is going to be, this work shift is going to be the output, at the end of your FMEA analysis, you will end up with a work shift like this. And of course, in addition, you will also end up with some priorities, because you could do many different things and the FMEA approach, the FMEA approach also tells you, it gives you a sorting of priorities, it tells you what are the things that you must take some steps to try to prevent now, try to take some corrective action now. What are the other things that will probably, that could probably wait and you probably take some time to figure those things out and probably take some corrective actions. But the ones that have high likelihood of failure and also have a high impact, those actually are the ones that should be addressed as quickly as you can. When you do that, you are really following a structured approach, you will identify the potential failure modes, you will basically identify what different ways could this system fail, you assess the failures to determine the prioritizing actions and this is something that we got to be able to do, because you must what attack the more important ones first. Then of course, once you have gone through the analysis, you got to document the process, you got to make sure, you release the action items to those people who are supposed to be able to act on this. FMEA is intended to basically rate this severity of failure mode, failure modes are those different ways by which a product could fail or a service could fail. Identify actions to reduce the occurrence, basically to impact the likelihood itself and of course, test the adequacy of controls, you might have some control actions put in place, you would like to probably check them out, in that the control actions themselves do not fail, this is something you would like to be able to do. Some history of FMEA, it was created by the aerospace industry in the 1960s to try to make sure that there were not as many failures and flight is something that can cause to, that can lead to a loss of life, that is something very very costly and you would like to avoid as much as you can and of course, the loss of the equipment, that also is something you would like to minimize. Ford began using this in the automotive area to try to reduce the breakdown, breakdowns of cars and is where it was adopted by the big three, when they move slowly toward QS 9000, they started using FMEA. The big threes are the GM, Chrysler and Ford, these are the big three automakers, they started using FMEA and the automotive action group, industry action group, this is a group that basically formalizes lot of action, which are going to be there to benefit the companies, to benefit these three auto companies there. So, what does, what are the benefits of FMEA, safety is something that you get out of this, you improve quality and also improve reliability, all those three things you basically impact. If you continue with the benefits, the potential benefits can be, can be potential, now those are, those are the previous ones for the real benefits. Now, there are some potential benefits, the company image may improve, user satisfaction is also something that might also improve, lower development cost, because if you already done FMEA, it probably puts you at a higher ground sort of and also the presence of a historical record, this is also something that is like a benefit. You will find out, how did I reach certain decision, decisions during design or in the development stage, what all, what all things did I go through to try to do this, that comes very handy when you are trying to do the same job again, you got a history there. And the applications where you can apply FMEA is concept development, when you just basically trying to conceive a new product, that is like one place you could probably use FMEA, design FMEA, this is called DFMA, this is DFMA, this is one place where FMEA can be very, very effective, because you are really trying to prevent, before the new product comes along, before the new design is converted to a manufactured product, you are trying to predict as much as you can about the likelihood of its failure, this is something you would like to be able to do as much as possible. And of course, process FMEA, so many processes that break down in process and you would like to be able to see, can I, can I reduce the likelihood of its failure or its impact, this is something you would like to be able to do. Now, more and more services are becoming a very significant aspect of our, our economy worldwide, globally and services can also fail, because of variety of different reasons. So, we are also getting into now doing FMEA for services and of course, equipment FMEA is done for there also. If you are trying to do, if you are looking for a procedure, you got to start with the component level and you would like to see, what are the functions of each of these parts and sort of try to come up with some sort of a basis to, to say that the product or the components, they fail in this manner, this is something you should be able to do. If you continue with the procedures, you determine the effects of those failures there, this is something you would like to be able to figure out. And then of course, the severity, what is the impact and what is the likelihood, both of these things you got to figure out as much as possible, whatever it is you are trying to improve, it could be for example, the cover of this, it could be the cover of this CD, this could fail in certain ways, it may actually damage the disc, the disc is inside and if this design is not right, it may actually end up damaging the disc or it may just come off the locks, it may come off the hinges for example, or it may crack or it may scratch the disc and so on, so forth. So, many different ways, this cover itself could really lead to a failure, a failure of the total system. So, you would like to basically understand the different ways it could fail, that will be the mode, then like to make sure what is the likelihood for this happening and then what is going to be the impact and then will I be able to detect, with that you will probably work at various components, for example, you might look at the hinge, you might look at the cover, you might look at the plane surface there, the finish of it and so on and so forth, that little paper that is there, for example, some paper, some paper is there, can that come off, can that actually do interact with the disc in any manner at all. When the disc is put in place, it should not the surface, the surface underneath should not probably touch the bottom of the thing, so in fact, there should be some design modification right there, you may not be able to see it, but there should be some projection there, that holds the disc, it is a very delicate thing, it should be held there in the right place, in the right place and it should be portable also, I should be able to carry this in my suitcase, for example, or in my briefcase or sometimes I put it, stick it right with my laptop and I carry it across town and so on and so forth, so that is all something, this this guy should be able to prevent, my loss is going to be loss of data, loss of the disc itself, that is going to be my impact and what is the likelihood, this is something we got to figure out and if there are actions required, then of course, we will modify the hinge or modify the stand or modify molding or do something there to try to do this, this is something we should be able to do, when you do it all, eventually you will got you got these numbers there, you got the likelihood and you got the impact, these should be quantified, then you got something called detection probability, you should be able to detect that there is a problem there, that also is part and parcel of doing SMEA, you multiply these three quantities, which is the probability, the impact, quantified and the detection capability, if you multiply these three, you will end up with something that is called the risk priority number, if you end up with something that is like, if it is if detectability is low, obviously it is more serious, so you would like to give it a rating that is high, because it is got poor detectability, these are these are things when they are multiplied together, you end up with something called the RPN, the risk priority number and risk priority number basically helps you to try to prioritize the different actions, which are which this this particular design deserves for example, you would have RPN for the hinge, you will have RPN for the surface, you will have RPN for various things and then, that would give that would give you a hint as to what should be the priority, in which I should be able to do it, you could even do SMEA for your glass for example, various places you could damage this glass, never mind breaking it, but when you just lay it on the on on a surface for example, could it could the glass get scratched, could that impair my vision for example, could it collect dust and so on, so those are some of the things you would like to be able to predict as much as you can right in the design stage itself, so there is almost nothing that should escape your SMEA, you end up with something that I call the that we call the worksheet, here is an example, here is an automotive company and that has got an auto assembly plant, and it is got a paint shop that has got a tank and this tank hold different types of paints, which are required to paint the vehicle, these on the surface of the vehicle for example, and there is a there is some sort of automatic system that fills up the tank, the paint tank, then it clicks and then the automatic spraying starts and so on and so forth, that thing goes on. Now, one could actually ask in what different ways could the function be impaired, see if you look at the worksheet now, the function is fill the paint tub, nothing more and nothing less, just fill the tub, failure mode is that the high level sensor never trips, even if the trip got the tank got full, the high level sensor did not trip, what is the effect of it, the result of this failure liquid spills on customer floor for example, the user. In this case, this paint painting machine has been supplied by somebody else, the car manufacturer does not produce this paint painting system, so it has been produced by somebody else, it has been mounted there, it has been installed there, so the automotive auto manufacturer is really a customer, if paint spills over, it will fall on the customer's floor, so that is like a consequence, that is like an effect. What is the severity, and this severity is rated on a scale that is like between 0 or 1 and 10, 10 is of course, have very severe, in this case, the engineer he sort of felt looking at the damage on the clean up required and so on and so forth, he gave it a rating of 8, he said if there are frequent failures like this, that would really end up probably shutting down the production line. What are the causes, what could have caused this, now I am getting into that fishbone diagram, you might remember we had a we had a little diagram called the fishbone diagram, where we had the effect at the end, and we had various causes leading up to that, that fishbone diagram is a brainstorming tool by which you basically, people who are working in that area, they would know very well, what all different things could contribute to this failure there. So, that is something that I will be doing there, and I would be tracing down to the causes. And the occurrence rating, and this is something that of course, is the likelihood, how likely it is to be, and they have given it a low rating, that means, not very likely to fail there. The current controls, current control is based on the you know the field to the low level sensor, there is a there is a low level sensor, the moment the low level sensor you know indicates that the tank is empty, or this paint tub is empty, it starts pouring liquid, it starts pouring, it should it should stop the moment, level gets to the right level there. What is the chance of detecting that the level has been, the the the the tub has become full, that detection probability they have given it a number of 4, the the the capabilities have rated it 5, and of course, how critical is it, is it critical, is it non-critical, now critical is something that can lead to a loss of life, or loss of property, hopefully those things are not happening, just some mess on the floor, so that can be probably cleaned up, not a not a big deal there, so it is it is being regarded as something that is non-critical. If it is critical, action has to be there now, then of course, you multiply 8, which is severity times the likelihood, which is like 2, which is 16 multiplied by detection capability, which is 5, you end up with a number that is 80, 80 is the r p n for this particular item there, now this r p n is the is the number that will indicate, how important it is for me to fix this problem there, this potential problem there, is going to be, now this 80 is to be compared to other things that might have probably picked up an r p n of 20, 30, 40 or may be 200, we do not really know, what all different things can go wrong in this plant, so that is something we got to figure out, but before that of course, we got to do the r p n for every kind of function that can stop, because of various types of failure there. Then, what are the recommended actions, recommended actions are perform the cost analysis of adding another sensor half way perhaps, between the low and high level sensors, that is like a recommendation, because that could reduce the likelihood, and who is going to be responsible for it, there are an engineer, Mr. Ravi Kapoor, and he was told that as of 10th of October 2009, kindly get this fixed, and then this got filed, this worksheet got filed, and the report we have back is the action was taken, and the the problem has been fixed, and they have taken this corrective action, which is like they have installed a level sensor, that is half way between the full and and the empty position. So, this is this is our worksheet, this this worksheet basically helps you identify, what could go wrong, find out it is impact, find out it is likelihood, find out it is detection capability, and then work out something called the r p n, which is going to give you a relative importance, relative significance of fixing this problem as opposed to others, and then of course, the preventive action, or the the corrective action, and then assigning this the responsibility for taking the corrective action to someone, and reporting back on saying. So, just see how we have prevented a potential problem, and this generally these turn out to be a high return project, their ROIs are pretty good, and this is just a basically a flow chart of what we went through, assign a level to each process or system component, which we did that was the paint job, paint tub, list the function of each component, we also did that, it is supposed to fill up the thing, list potential failure mode, we did that, describe the effect of the failure there, we did that, we described the failure severity, that also we did, spill on the floor and so forth, determine the probability of failure we did that, and the detection rate, detection detection rate was also something that we did, then we ended up with the with the r p n, and then of course, we took took action to reduce the highest risk, that is something that we did. What exactly is FMEA, if you if you get into the nuts and bolts of this a little bit, there are two key things that you must go after, one is the failure mode, any errors or defects in a process design or item, especially those that affect the customer, can be potential or actual failure mode, these are the different ways the functionality can can can be harmed in some way, or it can it can break down, and the effect analysis is the consequence, effect analysis is basically the consequence of something has broken down, what has been the impact of it, these two things you got to keep in mind. How is it evolved, if you go back to 1940s for example, that is when the army started first using this, if the US army started using FMEA or some variation of FMEA, then of course, from that point on, if you take a take a look at today, there was a serious application of FMEA, which was introduced by the Ford company in the 1970s, they had a large number of accident with their Pinto cars, and these led to loss of life and loss of property, and there they did FMEA, and they located certain certain aspects of the design, that had not been properly addressed, when the design was commercialized. So, this is like something that, these these these results or the action items emerged out of FMEA, doing FMEA, FMEA is also used to try to improve production, you have ongoing production going, you have an existing design, that also can be done, who are the different people who use this, all kinds of people, aviation people, nuclear power people, aerospace people, chemical industries, automotive industries, they have been doing it for a long time, and the goal really is to try to prevent prevent accidents from occurring, that is really the goal. This is an FMEA cycle, and if you look at step one, it is detect a failure mode, different try to find out, what different ways can a product or or a component or a service break down. So, detect the failure mode, and look at the severity, look at the look at the look at the severity of it, which is like, what could be the impact from that, you would get to know the severity of it, what is the likelihood of it, and what is the detection number. We saw all of these things, when we worked the worked at the worksheet, then you work out something called the RPN, once you have the RPN, it can help you prioritize the the actions, the corrective actions, as opposed to working working working on something else, and getting your getting your system improved. There are certain terms, which are used in FMEA, there is something called FMEA, it is the manner in which the failure is observed, and the effect of the failure is basically the consequence of the failure itself, that is there, then there are other things, for example, they could be local effect, end effect and so on so forth, but there is the important one, which is the severity, it is the consequence of that failure, basically it is the worst potential consequence of a failure, and that is determined by the degree of injury, property damage or system damage, that could ultimately occur, that is like something that tells you how serious could could this matter be. And let us try to see the manner, in which a system fails, what what exactly is something called the failure mode, let us take a look at an example, in how many ways can a particular, let us say a vehicle can stop functioning, engine is not working, and partial function is, so then there could be, that is a mode of failure, engine is not working, or not all function of the engine, the engine starts, but it quits, so that means, not all functions are properly working, so these could be the modes, these could be the different manners, different ways by which a by which a vehicle could could would have would have difficulty starting, getting going. If you look at other examples, if you look at other example, for example, there could be technology mode, the failure could could take place, because of technology, for example, if there is a space craft flying, and the on board computer fails for some reason, that is a critical failure, there could be conceptual failure, and this can happen sometimes, when person a person does not fully understand, how to carry out his task, and this happens a lot, this could happen for example, I will give you couple of examples, one is you know the people, who are who are at the flight deck, they are the flight controllers for example, at the airport, you know there are certain training, that they have to go through, the other kind of people, who also have to go go through this kind of training, is people who do weather forecasting, and they forecast these different, they they look at pictures, and from that they with their training, they can figure out, there is going to be a massive storm, or snow storm, or tornado, or something like that, that is something they are trained to do, they could be conceptual breakdown there, because because because the person training is not complete, or some aspect of it was not covered very well, so there is a conceptual breakdown there, they of course, they could be organizational failures also, for example, you may not have the, you may not have well defined job roles, and the result is that, also people probably do not understand mission priorities, and the the result is this, there is failure in salary administration, because you penalize someone, because he has not done some job, the problem is probably his job, the responsibilities of the roles they were not defined very well, whose failure is this, it is not the person's failure, it is actually the organization failure, organization's failure, that led to this kind of consequence, this is something we got to be able to do, got to be able to fix, if there is a weakness in design, if there is a weakness of the design itself, if the design of the cover is not good enough, I will probably have a problem there, I I am showing you an example here, which is like a design defect, look at the two parts of it, these are these are these are parts, that are supposed to supposed to fit together, now this is the polar connection, polar means, there is a positive end, that must go into the positive hole there, there is a there is a negative end, that must go negative negative plug, that must go into a negative hole there, if they are of equal size, you know and I know that, there is every chance that someone is going to connect it backward, connect the plus to the minus, and the minus to the plus, if the design is like this, so that can lead to a massive failure, that can lead to a disastrous failure, tripping and everything else, and fire perhaps and so on so forth, change that to this system, look at this now, the plus has got a bigger plug, and it is got a bigger hole there, and the minus has got a narrower plug, and a narrower hole, there cannot be a misfit there, there just cannot be a misfit by doing, the plus is not going to go in there, and the minus is not going to be easy for for it to be pushed in, if this cannot go in there, so this is something has to prevent a misconnection, a wrong connection, this is like we we we look at a modification, we had this design to start with, and we change that to this design, this design which is the modified design, and this actually is a way by which I am I am converting an old design to a new design, and it turn out to be a good design in the end, failure mode it could result in what we call an effect, and it is due to what we call a cause, these are the couple of things we got to remember, we got to remember that with every failure mode, we have an effect and also we have a cause, it is the the effect is something that actually results from it, but the causes are the ones that cause the failure to occur, so we got to identify these things, we got to understand these things, take a look at another example, a text a piece of text is unreadable, and that leads to the customer not being satisfied, and this could be because of a variety of different things, perhaps the writing surface was known was not was not smooth, perhaps other problems were there, and so on so forth, that is why the text was not readable, so there is the effect which is here dissatisfied customer, but it was caused by this surface roughness, so there is the effect, and there is the cause of it, it is due to this, and many times of course, you would be able to fish this out, you would be able to basically figure out what cause what, if you construct the fishbone diagram, and you do something in a very disciplined way that is called brainstorming, the only thing you do there in brainstorming is, you do not control the the number or the different diversity of the ideas that come up, that that are given by people, this is something you do not try to control or do quality check on, let it pour out, and then hopefully some gems will also emerge as a result of brainstorming, so the fishbone diagram is a great tool to try to do brainstorming, and basically figure out many different ways by which this failure would have taken place, could have taken place, so FMEA can lead to brainstorming, this is something you would like to be able to figure out, and some procedure is given here, you can follow this or you can follow any other one, my favorite is to start with the fishbone diagram, and to try to populate the diagram by asking people many different questions, and you could use the what we call the post it note, so you could do anything else, there are number of alternatives available, and basically what you should be able to do is, let the ideas pour, and there are some steps that are provided here, step 1, step 2, step 3, step 4 and so on forth, eventually the idea is to be able to sort out the roots, to be able to get to the roots, once you got the roots there, you got a starting point, you can actually for now looking for prevention, in trying to reduce the occurrence itself or perhaps the impact of it, if you trace it down to the root, you got a chance, you got a good chance there, and so brainstorming would be like one great way to get there. Now, that will give you the cause, you also have to find out what is the probability that the failure will occur, this is something that can occur at various different places, there could be a very high chance of failure, in which case the chance is something that I have to worry about, so I look at the ranking, and if it is very high that will lead to a very high R P N, and I want to give it some attention, I want to, so basically it is going to be helping you, in trying to prioritize the different action items, if something is likely to happen, if some failure is likely to happen, please go after it, do not leave it alone. And then of course, you have got to look at severity also, and severity or the impact of it that can be done this way, you look at the severity, if it is a safety failure, it belongs to class S, if it is unacceptable you know industry calls it class A, if it is a big risk, relatively big risk and correction is recommended, it will be class B, there is minimum risk that will be class C, and if you are willing to live with it, it will be class D, so these are the different ways by which I categorize my severity, and of course, numbers could be applied here, I could probably say 10 for S, 8 for this, and 6 for this, and 4 for this, and 2 for this, with that I have numbers, and those numbers come very handy, once I go out and I try to do basically what we call R P N calculation. So, what are the various possible actions I will be taking, if you look at the possible actions, I will try to eliminate the fact failure, or I will try to reduce the effect of it, or I will try to reduce the occurrence, so either I eliminate it all together by taking some design action, or by rectification action, or I try to reduce the effect of it, I take some preventive steps here, or I reduce the occurrence, any of these things if these are done, they are going to impact your R P N, and that is going to be a big big plus. If you look at again the different alternatives that we got, corrective actions we have got, when eliminate the failure mode itself, that means, I just remove the root, wherever the problem started with, I minimize the severity, this could be another way to try to reduce the impact, reduce the chance of it, and improve detection, any of these things, if you act on any of these fronts, you are going to be gaining, and lo and behold we are back with the worksheet here, and I see there this item 80, this is the same paint of filling problem, and it is got a R P N of 50, hopefully the other failure modes have led to an R P N that is less than this, therefore this is the one that should get some action, some action is planned here, and this resulted from looking at severity, looking at the likelihood, and looking at detectability, and this is a non critical item, so I do not really have to work on it immediately, but it is something I should take a look at, and that is assigned to an engineer, and the report is that yes, he acted on 10th of October 2009, and he took the corrective action there, I have provided you here with a worksheet, and of course, you will find worksheets either on internet, or you will find it in reliability books, and what I would like you to try is again going back to the bicycle problem there, we did an FTA on your bicycle, now this time try to find out can you do an FMEA on a bicycle, on the total system as a bicycle, and just for one moment I would like to ask you a question, and the question is this, are the two wheels of the bicycle as far as reliability goes, are they in parallel, or are they in series, please try to figure out the answer. Industry uses FMEA very extensively, and basically you know this is something that is done very routinely, and it is been prompted by the automotive industry action group, AIIG, they have basically prompted people to do FMEA, and you would be doing design review, that is like something that is done, that is like another mode of doing FMEA, the design FMEA, this leads right into our design failure mode and effects analysis. Now, we are looking at designs themselves, and we will be following the same routine that we did for basic FMEA, identify all the ways in which the failure can occur, and this of course, we are trying to do this, when we got the design on the drawing board, estimate the effect and the seriousness of the failure, and recommend correct action, and the corrective action here is going to be a modification on the design. So, again design failure mode effects analysis, this is going to be first identify the failure modes, look at the different effects of failures on customers for example, look at the severity, look at potential causes of failures, and then look for modification of the design to try to eliminate this problem there. So, details of doing FMEA, those are laid out in these different slides there, again I have got the failure mode there, I have got the causes there, I have got the effect there, I have got the correction there, and I have got to make sure that this is done in a team manner, because it is not so easy. Let us take at this example, let us take a look at this example there, there was a particular storm, there was a huge storm that even brought a tornado with it, unfortunately the people who are watching this storm, they failed to recognize something called the mini super cell that was being formed in the storm system, they failed to recognize this, because there was a shortage in their training, and this is something that became a major issue, which the consequence was this, yes there was some loss of life, because it was not a very teeny mini storm, it turned out to be a big tornado, and it led to loss of life and property, although it was not fortunately not very big, but it was something that people should have caught early on, and this was missed because of lack of training, lack of comprehensive training. So, that was pointed at something that was there, if you look at the design of the total system, what is the total system here, it is the man plus the system, all the communication devices and display devices and everything else, and the sensors and so on and so forth, and the data analysis program and so on, all those things together they make it the system, in the final at the end of the day, it is the person who makes that decision, and if he is not competent enough to process all these things, that could lead to a failure, and this was traced, this particular failure was traced to a problem that occurred with incomplete training of the person who is in charge of forecasting storm. Here is another example of a design defect, an engineer, this guy he was riding a train and he failed to notice a red signal due to fog, and he this led to a major accident, loss of life and so on and so forth. Now, this led to injuries and loss of life and of course, damage to equipment, so we saw the cause, the cause was that the person could not see the red signal, and this was done by manually checking, he was looking out and he would check that way, whether the signal is up or down, or is it red, or is it green and so forth, he just could not see anything, because the fog there, and he might be coming from a different line, he might not be familiar with that area there, so he might not be looking out to see the signal, you know from far enough distance to be able to stop the train or slow it down, that he could not do. The corrective thing would be here to do not depend on manual detection, try to see can you provide some technology, whereby this signal up or down, signal green or red could be detected by going there, this is something that we should be able to do, this is something that we should be able to do, and what we should really do is, as much as possible we should try to see, if we could prevent problems like this. The bottom line is going to be again calculation of the RPM, so this is something we would like to be able to do, we would like to work out the components of the RPM, and again the risk priority number, it is made up of the severity, there will be a rating, there will be a number of the probability of the event taking place, and the detection capability, these three things they must come together, and they should give you, if they give you a high RPM, that is a serious matter, you got to take some action there, you got to take some corrective action there, and these this comes out, these numbers come out as the result of this FMEA analysis. So, I have got here the same formula there, S is the severity, O is the occurrence of failure, which is the probability, and D is the detection capability, these things together when they are multiplied, when they are multiplied, you end up with RPM, RPM is the risk priority number, and that basically, if you got RPMs on different actions there, you should really go with the RPM, that is called the highest RPM, first return that one, then go to the next one, then go to the next one and so on, and with each of these RPMs leading up to some corrective action, that is really what you are after, so you do a FMEA in such a way, that you are able to find these high impact, high likelihood, high likelihood and low detection capability items first, once those are taken care of, then you move down to the next ones and so on and so forth, this is something we should be able to, we should be able to do without any trouble at all, if I if I am doing FMEA. So, the summary of conducting FMEA, a cross functional team must do FMEA, identify functions of each component, find try to find at least one mode of failure, one mode of potential failure, define the effects, look at the severity and look at the detection capability, this is something that you got to be able to do, once you have done that, you can calculate your RPM number, RPM would come very handy in term in terms of prioritizing the different number there, once you have done that, you you then got to be creative, you got to then come up with solution and solutions would come from engineers, to try to improve performance, how do you actually do this, either you would like to eliminate the failure mode itself, somehow something is there that breaks down, because of that, minimize the severity of it, that is also something, that is like like another course of action that is there, reduce the occurrence, reduce the likelihood of the probability of of it and improve detection capability, like in our case for the for the railway signal, it was not be able to see the red light, the red signal, that was like the problem there, so you try to improve that by putting in some technology, that will probably, there might be some sort of a not necessarily audio, but some other kind of signal coming through, and that should be detected automatically, in addition to whatever the person is trying to do, that should be there, but some failures are more preventable than others, and this we have seen so many times, let us take a look at that warning system again, again notice here, we had that supercell, supercell was formed in the storm system, and this led to a conceptual failure, because the person did not have enough training to be able to do it, how could I change that, I could really change that by changing the technology, perhaps I could use a different kind of technology, and not be dependent on a person, not be dependent on a person to be able to recognize that, because he is doing a lot of signal processing in his mind, through his eyes he is trying to detect very some very fine things, should it be left like this, I mean look at the consequence, look at the loss of property, loss of life, look at the the tremendous cost that would lead to, I could perhaps upgrade technology, and this is something we should all try to do, this is like one place, where it is not very difficult to figure out what I should be doing right, you just should not hesitate to take that corrective action, that is the message, basically doing just FMEA is not enough, you should be able to take the prevent the failure and loss, this is something for which you got to act, and that is something that comes along very handy, once you try to do this. In certain situations, you will try to quantify hazards, and I will give you an example, if you look at the hazard scoring matrix, this is like another matrix, that is also very similar to what we did, we we calculated RPN, and what this guy is doing, what this matrix approach is doing, it is again, it is ranking the severity in terms of the consequences, catastrophic impact, major impact, moderate impact, minor impact, so he is got some ratings there, and then you got you got probabilities, and if it is frequent, it is catastrophic, you give it this, and if it is frequent and major, you give it this number, and so on so forth, and if it is if the probability is pretty remote, pretty infrequent, it is a minor sort of severity, you give it number 1, so here I have got items here, that actually have a hazard scoring, that is 16 or 12 or 9 or 8 and so on, if if an item ends up, if a if a if a FMEA leads to this sort of scores, on the hazard scoring matrix, this is like another approach, in place of doing the RPN, you could do it this way, and you would again end up with very similar analysis, here is an example, which is of a which is using a decision tree, so what it says is, let us say first you have done this, and then you got a number there, if the hazard score is 8 or higher, you move this way, and you try to sort of see, does a control measure exist, for the identified hazard, if it is you got your action done, you do not really have to go further, just activate that action there, if it is not, then you see if you could improve the detectability of that particular thing there, and then you proceed to what what we call a hazard guided FMEA, this is something that you should be doing, if you have good detection capability, just go ahead and act, there is no real issue there, if suppose it is not a high rating, it is a rating like 3, 4, 5, 6 and so on so forth, then you come down this branch, and then you ask yourself the question, is this a single point of weakness, if it is yes, then you probably go through the other boxes and so on, if it is not, you stop right there, if it is not a single point of weakness, you stop right there, stop basically means, you got to move on now, to other actions that are possible there. So, what I would suggest is, try to look up some references on hazard guided failure mode and effect analysis, here is an example, you know this is this example is taken from the health care industry, and it turns out that there are certain patient alarms, which are you know they are they are connected to the patient, if there is a patient in ICU for example, you do not really have a doctor or a or a nurse sitting there round the clock, and watching every movement of the patient or his BP's or everything else changing there, that is not something that you do, but you got alarms, you got the body alarmed in various ways, and those alarms trigger action. So, basically what we do is, in certain places, we would like to probably provide care obviously, and monitor alarm, this these monitor alarms would lead to lead to the kind of actions that we would like to see, we got to make sure the alarms function, the alarms do not themselves break down, and for this I probably have to go into this a little bit, to make sure that the alarms are functional, I can rely on those alarms, I have got 3 there action item number 3 there, 3 a will have one consequence, and 3 b will have another consequence. In 3 a periodically, I check the monitor monitor status, I mean to just to make sure that you know the the alarm is active, and the alarm is still looking at the person there and so on so forth, and of course, the other part is the respond to alarm, that is also something I should be able to do, once the alarm is there, so periodically checks the monitor status. So, has there been an alarm, and now this step could fail, if the nurse fail to check the status, then of course, is something there or misread or misinterpreted the the the alarm that came along or partially checked, any of these things could lead to a failure there, similarly I would like actually to get to this point, when it is like I respond to alarms, and what are the different ways, even if there was an alarm the person probably did not respond, this happens once in a while, or he responded slowly or too late, and by that time the patient was out of control, I mean this is like something we should be able to do, now to try to get into this you do an FMEA, to try to prevent those sort of occurrences you get into this, So I have here 3 B 1, which is like if you look at the first there 3 B 1, did not respond, why could he not respond, ignored the alarm, did not hear the alarm, or the alarm volume was too low, or it was in a remote location, or probably the care giver was busy, and therefore, you know just missed the alarm, and you do exactly what we did earlier, and you come up with outcomes, you come up with with outcomes and action plans and so on so forth, so we are also doing here, what we had done earlier, we are doing something very close to FMEA, that is what we are doing there, here if you see I have got the failure mode, which is the cause, look at the severity, look at the frequency, and look at the actions, and look at the outcome, these are exactly the same that we followed in the earlier case at all. Now, there is some slight differences between root cause analysis and so on, there are some similarities, and there are some differences there, people tend to prefer one to the other, it all again depends on the industry, engineering people they tend to go by FMEA, and health care people they may go by the hazard FMEA, that is like something that is there, we got to recall that FMEA is a tool, it allows you to prevent basically system product or process problems before they occur, FMEA reduces cost by identifying system product and process improvement opportunities in the early development cycle of the product itself, and it creates a more robust process, that is something that it does, it prioritizes actions to try to decrease risk or failure, and it also helps us evaluate systems and design from a new vantage point, this is the point of view of trying to be take a vantage position of prevention, this is something that we try to do. FMEA is of course, a systematic process, and we have gone through the steps earlier, and so you basically try to evaluate, you identify, you evaluate, then you try to quantify, then you try to prioritize, then you plan on some actions there, and you document the activities, so these can be followed up later on. Purpose and benefits there are obviously, are pretty major benefits of doing FMEA, this is something that is because of its logical nature, and it is done quite frequently, it is done quite frequently to be able to take care of this. Benefits of FMEA again, I have repeated this, I have mentioned this before, I repeat it again, it leads to higher reliability, better quality, increased safety, and has customer satisfaction, it also leads to cost savings, it also leads to what we call continuous improvement, and there are some examples, there are some examples there, when it says, if it cost 100, when it is discovered in the field, that is a pretty high cost, it may cost dollar 10, if it is discovered during final test, these are like the product development life cycle stages, it may cost only 1 dollar, if the problem is discovered to an incoming inspection, there is like before the, before you put the, took that part, and you put that, assembled it in a complete system there, so you looked at your transistors, or the chips and so on, before they entered the system, and it is even better, if you can do this, all this right at the design stage itself, so start your DF, start your design, start your FMEA right at the design stage itself, this is called DFMEA, what are the tools available to be able to do this, there is benchmarking available, there may be other people who are doing this job better, customer warranty reports that can lead to failure, indices and so on, so forth, design checklist or guidelines, field complaints, internal failure analysis, internal test standards, lessons learned, return materials reports, expert knowledge, and certainly warranty service, these are the sources from which you can find out, how things, how different things can go wrong, in what modes do they fail, what is their impact, what is their likelihood, all these things can be found, if you got a good database, what are the possible outcomes, what if you would do, would do FMEA, you can discover the failure modes, and it can lead to a meeting certain legal requirement, that also can be there, duty cycle requirements can also be impacted, product functions obviously, can be impacted by doing this, some of the key product characteristics, those can be enhanced by doing this, and certain product verification and validation, validation is the actual use, verification means, it confers to specs, and the of course, to be able to do FMEA, you got to do your PT meeting, which is like guys would be sitting around, and they would be basically getting briefed on it, and here is a little example, a perform DFMEA, so the team has been now looked, asked to we are trying to design a new pressure request, and what all things could go wrong with this, and for that you will start your basically approach, determine the scope, what is it that we are trying to do, gather background, reference material and so on so forth, I am going to give you one or two illustrations on this, create block diagram, so it is easy to understand, what the process is, what is actually going on, identify team members, prepare the agenda, schedule and milestones, and of course, identify item functions, failure modes and their effect, this is something we should be able to do. For the pressure cooker, safety valve releases pressure before it reaches dangerous level, this could be like one, this is this is its function, these are the safety function, now it could fail, because if this for some reason, the safety valve does not release, that could lead to a major explosion, thermostat open circuit through a coil, before temperature rises above 250 Celsius, and this also is a dangerous situation, so the thermostat, so the safety valve could fail, or the thermostat could fail, or the pressure gauge is divided into green and red areas, and danger is indicated, when the pointer is in the red section there, now there could be a problem there also, so you can actually see three modes of failure right away, define the scope that means, let us take a look at these systems there, and you could look at the electrical system, or you could look at the safety valve, thermostat, pressure gauge, whatever you want to focus, your your focus is going to be safety, and for that you again start with the block diagram, which is like the pressure cooker, safety valve, pressure gauge, these are sub components of it, it is a like your FTA, this is very much like your FTA, the electrical system could fail, because of heating coil failure, or cord failure, or plug failure there, and the safety valve could fail, because of valve spring not functioning properly, or the valve casing not functioning properly, there are certain assumptions you make, you make sure that the components that are used, those are functioning properly, and that could lead, that could lead to basically a problem in failure. So, again let us get right back to our work sheet, and let us review what we have done so far, some function is not being delivered, this is something we would like to be able to check, we would like to be able to prevent, potential causes for this, in what different ways could this function stop, so for example, the engine does not work, in what different reasons, what different reasons are there why the engine does not work, what is the effect of it, well you you can see right away what would be the effect of the engine not starting there, what is the severity, what is the impact of it, and then of course, I have got some class, and that would depend on the severity class there, as I showed you earlier, potential causes or the mechanisms of failure, this would be like your mode, and you do your cause and effect diagram there, occurrence this is again the probability of it, and current design control do they do they do some prevention, or do they do detection, and then detection rate, this is something that I have also got to figure out. So, I have these three quantities severity, occurrence, likelihood, and detection capability, these lead to my R P N, and the highest R P N would immediately get some attention there, so there will be some recommendation given to to take care of this response, and target completion there, these are there, who is going to be taking that action there, and some other follow ups there are there, so this is the worksheet that you should end up doing this. And I have these slides here, and these slides should be there, you can just flip through them up to a certain point, they will tell you what are the functions expected, what are the how failure modes are going to take place, what how are you going to discover the effects of that failure there, how are you going to find out severity, how are you going to find the class, how are you going to find the different ways causes of failure, how are you going to find the frequency of occurrence, how are you going to be find the current design controls if they are there, how are you going to look at the detection capability, how are you going to calculate the R P N, this is something you would like to be able to do. And of course, you want to be able to make sure that the high R P N items are there, all looked at before you go very further, and recommended actions of course, this is something that you got to be able to end with when you end up with your FMEA, you got some FMEA actions there. Thank you very much, I really appreciate your spending time in browsing through these videos here, and we are approaching fast the end of this 40 40 session sequence of our lecture on six sigma. So, the course will be ending with two more lectures, and there are two very interesting topics that are coming up for you in the next two lectures. So, see you soon. Thank you very much.