 This is the first class on machinery for diagnostics and signal processing. In this introduction class I will be basically telling you about the course content and what we are going to learn in this course and how the course will be delivered and what will be the basic content of the 40 lectures which will follow. Well, this course will be actually in 5 modules and before that let me tell you if you want to know anything more about this course you can always go to this website which will be periodically updated with you know handouts on your question papers, tutorials, quizzes, etcetera. So, I will be letting you know when to go and see that website. But as I was coming back to this 5 modules, the very first module is on the maintenance principles and question is you know in a machine the idea is you know what is maintenance and what kind of maintenance are available to us and then the question comes you know can I just use and throw a machine or do I need to invest to keep it running. But these are serious issues I mean you think of a steel plant I mean obviously I cannot afford to lose a blast furnace, I cannot afford to lose a rolling mill, but certainly I can afford to lose in a few motors and pumps here and there. So, you all can realize you know it depends on the cost of the capital equipment involved, it depends on the criticality of the equipment involved. So, our maintenance efforts will be proportional to the criticality of that equipment to the strategic importance of that machinery etcetera. So, there are techniques which will let us know like which kind of maintenance of course, we do not know what are the available maintenance to us, but we will once we know in the second lecture what are the maintenance principles available to us and knowing that in the third lecture we will try to know by a method which is predominantly used by a lot of industrial engineering engineers is this failure modes effects and criticality analysis or which is known as Femica. How by Femica we can decide as to what kind of maintenance is to be done where in a large plant is it always breakdown maintenance is it always predictive maintenance is it always periodic maintenance. So, these are issues which we will understand. And of course, you know always people ask this question well one is my machine going to fail, I have maintained my machine it is running nice, but then I would like to know how long is it going to live. So, once we have diagnosed the faults in the machine people always are interested in to know the remaining useful life of the machine. They would like to know the remaining useful life of the machine and then how do we estimate that and this is known as the prognostics of the machine. So, we will discuss regarding certain case studies on the machine prognostics and then the once we know about the diagnostics. So, this will be this four lectures will be in the module one that is the introduction followed by the maintenance principles then failure modes effects and criticality analysis then followed by fault diagnostics and prognostics. Now, let me tell you briefly something about this maintenance. The focus of this course will be something towards what is known as CBM which is condition based maintenance. That means I have to know the present condition of the machine. One question is my machine is running how do I know its present condition? Well, when the machine is running it gives us certain signals. We hear a sound coming out of it. We feel its vibration. We feel its temperature going up. So, these are the telly-tale signs or the signals which convey what kind of condition the machine is presently in. So, we have to acquire this condition out of the machine by certain signals. Now the signals could be visual, could be in a tactile, could be electrical, could be vibration and so on. But again worldwide about 70 percent of the cases in the industry are through vibration based CBM. So, this vibration is a very very important signal parameter. Which is to be nicely understood and the vibration from this machine has to be analyzed so that we know the machine's present condition. We are able to diagnose the machine's present condition. It is very similar to like I was telling you in the last meeting that we will become human doctors. Like when a patient goes to a doctor or doctor measures the ECG, doctor takes the temperature and measures the blood pressure and these are the parameters of the indicator of the person's health. Similarly, in a machinery a vibration is a very very strong indicator of the machine's present condition. So, we have to understand the basics of vibration and in fact the next module of this course is towards that that we need to have a thorough understanding of the machinery vibration. Because this vibration is going to convey to us the machine's condition and there are many applications of machinery vibration. We will use those applications which are useful or beneficial to us in finding out the condition of a machine. For example, the response of a single-leg freedom system, a 2-degree freedom system and then effect of the you know the basically every machine if you look at it, it is nothing but a shaft which is mounted or held on couple of bearings and this shaft could be carrying a gear, could be carrying a pulley and so on. So, everything on a machine and then we could be capturing the signal from this bearing location by a transducer and this could be our rotating shaft and this could be a pulley or a gear or an impeller. There could be one or many and they could be meshing with few more and so on and everything could be complicated and then the most important thing here is we have bearing. So, the dynamics and this could be in a casing and then we have our very basic machine. Now, this machine could be coupled with another machine and so on with a coupling. The dynamics of this machine is going to give rise to certain response at this bearing locations. This response could be a linear vibration in terms of a displacement, it could be a rotation theta. So, these responses which are may be vibratory in nature are actually are the signals which has to be analyzed by us, analyzed or interpreted. So, that we can find out the machines present in the condition. So, you all can appreciate that the understanding of vibration is at most important to understanding or conducting CBM in a any machinery. So, we will be focusing few lectures on the basics of machinery vibration, then particular engineering applications of vibration in terms of the unbalanced response, in terms of the vibration isolation, in terms of how do you measure the base excitation, because of a motion at the base of the machine, how do you measure them. So, these are some of the engineering applications of vibration which we will be studying, which are used in machinery maintenance and then we will talk about rotor dynamics in which a shaft is rotating, what about its natural frequency, what is critical speeds, what are the whirling of shaft and so on. Because as I was telling every machine has these few elements common and if I was to less than any machinery, any plant let us talk about say you know are familiar steel plant, cement plant, paper mill may be a power station. If you just think about you know very wildly about the contents of the machines in these plants, I am sure all of you will agree with me that the common elements in these plants are rotating shafts, there will be bearings, there will be gears, I mean this comes without thinking in fact, there will be fans and impellers, there will be electrical motors, of course motors themselves have rotors and bearings, there could be you know pulleys, chains, belts etcetera. So, you talk of any plant, these are the very common basic machine elements which are present in those machineries. It is very similar to you talk of 10 different human beings, but they all have a heart, they have all have a lung, they all have a liver. So, these are in fact the organs of the human being, so they are the organs of the machines, we are going to eventually become machine doctors. So, now we have to see how the vibration from these machines manifest in themselves certain characteristics features of these components. So, when I when I measure the vibration from these machines, I would like to know what kind of inherent characteristics these vibration signals have which pertain to the machine common machine element, is it bearing, is it a pulley, is it a gear, is it a shaft which is misaligned, shaft which is cracked, shaft which has loose components, they all have characteristic vibration signal, all these common machine elements. So, this helps us do a fault diagnosis in that machine. For example, just for the sake of an example I will give you, say for example which we have not studied here, but then I will briefly introduce you to what is known as a vibration frequency spectrum, certain vibration. Suppose a good bearing has certain response and this is for a good bearing, but if the bearing was bad maybe you know I would get this response and certain high frequency. So, this is the response of a bad bearing, the vibration, vibration response. So, just by having a understanding of these two spectra from a good bearing and a bad bearing by looking at the spectra itself or finding out the features of this spectra, maybe the RMS amplitude at high frequency is high for a bad bearing which is not there for a good bearing, I can say if the bearing is unknown to me whether the bearing is good or bad. So, this is the power of vibration signature analysis. I will be going to the specifics as to what these values are depending on the design, depending on the parameters of the machine components. That we will discuss later on, but I will just tell you briefly the power of vibration based signature analysis or this is in fact a signature of a bearing, a good bearing and a bad bearing. So, by doing a vibration based signature analysis I can find out whether the machine is bad or good. Now, I was telling you about that 70 percent of the cases in the world are vibration based, well then obviously the next question you all would be wondering what you do with the next 30 percent. The remaining 20 percent is actually what is known as by a technique known as wear and debris analysis. In this technique what we do is from a gear box for example, there is lubricating oil in the gear box. With time what happens, these mating components wear out. Once these components wear out, these particles get deposited in this lubricating oil. So, in the process what happens the lubricating oil loses some of its physical properties because the lubricating oil has got contaminated, its specific gravity may change, its viscosity may change, its certain acid number, base number may change and then it may also undergo certain impurities. In fact, this is what is known as the oil analysis and next next is this impurities themselves could be analyzed. The impurities I mean the wear particles. The size, the shape of the wear particles give us a clue as to what kind of wearing mechanism occurs because wear is going to happen because when two components have a relative motion between themselves they will be wear and then of course we try to reduce it by putting a layer of oil in between them. So, this particles which get removed from this wearing particle wearing components because of friction will get deposited in the oil. So, this wear particles could be analyzed then I can get to clue as to what kind of forces created this wearing. So, wear debris analysis is an indirect method of finding out if there are abnormal forcing conditions in the machine. Normal wear happens, but suddenly you see the composition or the percentage of wear particles has increased drastically. You can understand that you know something is really wrong with it and then that is why we have to take care of how to reduce the successive forces. Another is because this lubricating oil has been subjected to such impurities it is going to lose its properties. So we can do the property measurement of this lubricating oil and get to get some clue as to what is wrong with this machine and this is usually done in the nearly about 20 percent of the cases do wear and debris analysis. But the problem with wear and debris analysis is that we cannot do this right at the gear box, right at the place where we have picked up the oil because the wear particles themselves could be of the size of 2 micron to about you know 30 micron. And imagine our say for example a coal mill or a cement plant and if you have been to such plants they are very dusty the amount of particles in the air which are suspended are more than 2 micron and 10 micron. So, we are going to contaminate once we do such an analysis right there in the plant we are going to contaminate the oil ourselves and that is going to give us a wrong reading. So, there is a there are procedures as to how we need to sample the soil what is the procedure so that we do not contaminate it during the sampling and once we have sampled it we take it back to a specialized lab wherein we try to do the analysis find out the properties of the soil properties of the lubricating sorry properties of the contaminants and so on. And the remaining 10 percent of the cases of CBM are what is known as the NDT non-destructive tests like in ultrasonics could be x-rays or radiography could be thermography thermal imaging. Another very important and upcoming area is what is known as this motor current signature analysis MCSA. This means for example, if you if you think of any electrical motor or any mechanical unit and there is a coupling and it is driving a blower they are all on a foundation I have this motor maybe this is a pump and I have electrical source. A very interesting thing happens is you know all of you would have realized for example, if I talk of a motor the supply current to this motor is usually a 50 hertz sine wave. Now, what happens once in such a system if this pump there is a fault what it is going to give is it is going to give a load torque on this system because of the defect will be loading. So, there will be a torsional load on this electrical motor and this torsional load has a frequency which is characteristic of the defect frequency of this mechanical unit which is the pump. Now, because of this torsional load what happens this motor is subject to a torsional loading and the rotor of the motor undergoes a torsional oscillation and because of this torsional oscillation the flux or the current drawn by this gets an additional torsional oscillation. So, the current no longer becomes a sinusoidal, but it gets what is known as modulated. So, if you look at the envelope we will discuss this in detail, but just to tell you about this. So, this will the motor current changes from being a perfect sinusoidal to an amplitude modulated wave it becomes something like this. So, this has many frequencies and this if I analyze this motor current I will see the frequency of the pump defect frequency will appear in the current spectrum of the electric. This is a very strong and a very powerful statement with lot of engineering implications in the sense imagine I am having a pump. So, for example, a submersible pump which is underground I have no access to the pump, but sitting somewhere in the control room I would like to know the condition of the pump. Obviously, I cannot go down under the ground and put an instrumentation to measure the vibration of this pump. So, how do I measure or monitor the health of such a pump? So, in the control room if I was able to measure the current drawn by the electrical motor and if I do such an analysis I will obviously get a chance get a clue as to maybe the pumps defect frequency will show up in the spectrum and this is very powerful technique. In fact, in IIT we have been working in this area about the last decade and we have been very successful with gear boxes. In fact, now we have a patent that with any rotating machines you can find out the from the current spectrum the fault in that rotating machines. You can imagine submersible pump, you can imagine a system inside a nuclear reactor where in a person is cannot go in, but then how do I find out the defect in such a system where it is not accessible and you know compared to vibration measurements the measurements of current is really very easy and then we can very easily remotely do it and as it. In fact, this is this I still want to put it in an upcoming area of CVM and just to let you know, you know I was talking about the signals. Now, of course, signals from the machinery to the place where we are doing an analysis the signals have to be drawn by a cable, but nowadays the technology is such that we need not even have cables, it could be all through wireless. I could be sitting here maybe with an iPad or an iPod and then or the mobile and then try to remotely find out what is happening to my gas turbine in Alaska. It is not a science section, it is possible. We can know what is wrong with the gas turbine which is running in Alaska perhaps through wireless communication through the internet. You have some transmitters carry the signal over the wireless, download it here at Kallagpur and then try to analyze it on your laptop. This is possible and we will do it. So as I was telling you, this is what is the how this 100 percent of remaining 10 percent of the condition mounting is done, but by now so vibration is a very, very important subject which is to be well understood and which is to which will be used predominantly in this course towards a condition based monitoring maintenance. And as I was telling you, in CBM I have a machine, I have to put a transducer to capture some signal and I have an analysis unit. This could be a transducer and then this analysis unit is in different types. So if you look at the lecture number 8 that is on time to one signal analysis, the basic signal which comes out of the transducer as you see in an oscilloscope, I am sure all of you would have used oscilloscope in your graduate. So by doing a time domain analysis of such signals we can find out the features of the signal in terms of its certain time domain features like its RMS values, mean value, quotas use, crest factor etc. I am sure in your first year electrical circuits you would have studied sine wave, cosine wave and you must have done the integration to find out the RMS value, mean value, max value, crest factor etc. And these are actually the indicators of the signal and such analysis helps us. For example the machine is bad, its RMS value of the signal is going to go up, its bearing is bad, its quotas is going to go up, go up. So by simple time domain analysis we can find out certain features of the signal which will be telling us the condition of this machine. The next lecture is on the frequency domain signal analysis. For example, the same signal in the frequency domain say may be some vibration signal, this could be component A, component B, component C, component D. Every component in a machinery, for example I have a machine whatever is the internal may be I will not be specific but something some components are there say this is component A, component B, component C, component D, anything a generic machine. So every component and they are rotating has a characteristic frequency, it is like this signature of the machine. So this signature is going to change if the machine's condition changes. So this is what is going to happen. I mean a signature could change in the form may be one component could be increasing with time. So these are if I put certain base level I will get alarm that well something is wrong with this machine, component B in this machine is having a problem because its level has gone up beyond the threshold level of which I have kept. Who decides the threshold level is something we are going to find out and how. But if this level has gone beyond the threshold level I will tell because it is component B something is wrong with component B and that is very important in frequency domain signal analysis. There are many more methods of frequency question is you know how do a convert a signal from the time domain to frequency domain. I am writing this as time and this as f, this is in time and this is in frequency because the power of frequency domain analysis is such that any signal which you see in the time domain like you see in an oscilloscope can be broken up into the frequency domain and in this frequency domain you can see these components showing up and how do you do that is something which you are going to find out excuse me and to do this conversion we need to convert the signal from the time domain into the frequency domain we need to take the help of a computer. Certain algorithms have to be implemented in a computer so that this conversion from the time to the frequency domain is possible. The question is how do I take my signals which are analog in nature which I see in an oscilloscope which I have been just measured with a transistor monitored on the machine how do I take it into the computer and that is what we are going to study in the lecture number 10 that is on computer aided data acquisition. What are the issues of sampling frequencies? Why is a 12 bit machine inferior to a 24 bit machine and so on? What is the need of amplitude resolution? How small a signal can be detect? How fast a signal can be sample or store have because see for example in a real wall machine if I have a real wall signal in the time domain something like this if I take the signal into the computer I should be also getting it like the same values because computer only stores numbers in forms of data points. Only question is if the data points are close enough I will get a two representation of the original signal which is in black rather if I pick up signals one here and one here and then one here and one here and one here and ask my computer what is the signal it is going to understand it is going to draw like this. So you can realize you know this is what is known as the delta t or which is known as the sampling interval. The power of this sampling interval is very much in the sense unless or the inverse of it is nothing but the sampling frequency to retain the features of the signal I have to always sample at a much much higher rate as there are certain theorems and algorithms for that which I will play later on. So computer data equation is a process by which I should not lose my real wall signals because of some interpretation my computer has done then otherwise all my interpretation will be wrong. So computer is a fool I mean whatever you give it that is what it is going to interpret. So if I give him something wrong or if my algorithm is such that it interprets it the way it understands and not the way I understood then there is a problem. So we have to ensure that the computer faithfully reproduces the original analog signal which is there out of this machine and then computer of course store this data in digital form digital data. Now once this signal is inside the computer I can use many mathematical algorithms to analyze the signal. In fact the subsequent lectures on FFT analysis, modulation side bands, envelope analysis, system analysis, order tracking etcetera they are all signal processing parameters which will be used by the computer to understand or find out the features from the signal in a much much better way in a very convenient and easy method so that we can interpret the fault in that machine. Modulation and side bands are just telling you about the case of the pump which is driven by a motor. The motor current is no longer a sinusoidal wave it is no longer a pure tone or a pure tone but it becomes amplitude modulated and if I do an analysis an FFT analysis of such modulated signals I will get certain side bands and this is very powerful way of understanding the signal. And envelope analysis is trying to find out the low frequency modulated wave which is carried by this high frequency modulated signals. Substum analysis is very very powerful in the form of finding out the families of side bands and gear boxes. Right now I just wanted you to know about this terms we will be discussing about them in detail but being an introductory class I will just telling you certain terminologies which we are going to study in this class in subsequent classes. And order tracking is very powerful when the machine speed change the phenomena if you are doing a frequency analysis the frequency is going to change. So if I am doing any signal averaging by if I am not doing order tracking I will lose the identity of the machine. So order tracking is useful to find out certain important characteristic frequencies in a condition where the machines are changing having a speed change. So these lecture 8 to 15 will be covered mostly in the module 2 and that will be mostly towards signal processing. And when we do this signal processing I will be using like I was telling you certain I will be using taking the help of excel to draw certain curves and then I will be taking the use of using a MATLAB to large extent in this course on signal processing. And I understand in our campus we have licenses for MATLAB so you all can install them in your laptop and then use them and wherever possible I will try to give a small examples on MATLAB to do such analysis. Now of course we have talked about signal analysis in the previous module. But module 4 tells us about how do I capture the signals it is basically nothing but mostly regarding the equipment and how to install this equipment how to instrument the machine because CBM always requires an extra investment on this transducers. We have talked about vibration excuse me vibration monitoring. So vibration monitoring how do I measure the vibration out of a machine? What kind of sensors have to be put on the machine where they have to be put on the machine? What kind of other sensors and transducers are available to do vibration? And in CBM a very very important parameter is also the rotational speed. How do I measure the rotational speed of a machine? And then of course many a times this analysis is not done in real time. Sometimes the technician or the engineer will go around the shop instrument and record it and then later on in your lab or in your analysis unit we can always pull up data from the recorded unit and do an analysis. So how do I record it? Once I record it am I able to store the data in the exact form it was while it came out of the machine? I am sure all of you must have recorded your voice signals and you hear it out and then sometimes it sounds different because the frequency bandwidth. Sometimes CD sounds better than an cassette tape it is only because of the frequency bandwidth. So there are issues of the media, the frequency bandwidth of recording, the resolution of the recording etc. So these are issues which have to be understood and I was just telling you about the remote monitoring. Suppose I need to transfer the data from place A to place B I can always carry the data in a CD or I can always transmit the data over the internet and there are limitations of speed at what rates I can send the data and that will affect my machine. Suppose my machine is a very fast moving signal has produces a very high speed signal say for example a gas turbine rotating at 30,000 rpm and in one second there is lot of data which has to be put. Obviously my internet may be I have only 2 gigabits per second is the maximum rate I have over the internet. So what is the size of data which I can send in a packet? There are limitations earlier we used to have about 2 mbps lines about 5 years back and now we have 2 gigabyte per second transfer rates. So there is always a challenge as to the rate at which I can transmit the data and the present day infrastructure available to transmit such high speed data in real time or of course I can always record the data but then recording the data means do I have enough bandwidth store the data what kind of media do I need and if the data sizes increases we move from a CD to a DVD to a Blu-ray player Blu-ray disk all because of the data size because we need to store lot of data. Lot of data means lot of time. So we have to be very careful into what kind of format has to be used and what kind of transmission architecture has to be used for doing CBM and then of course we will be talking about vibration transducer accelerometer seismic transducer LVDT etc. EDICAR improves and then how do I do vibration monitoring? In fact what are the minimum equipment required to do vibration monitoring in a machine? It could be a pump, it could be a generator, it could be a turbine. So what are the equipment required? How do I install? Where can they be subjected to harsh environment? Can they be subjected to high temperatures, nuclear radiations and there are issues and then of course we will talk about in module 4 about basics of noise and noise monitoring because predominantly again in CBM noise is not used as a monitoring parameter because of the fact that noise from machine A gets easily contaminated by noise from machine B. But noise only gives the operator indicator that something is gone wrong with the machine. A loud noise, suddenly a machine makes a strong impact noise that means something is wrong with the machine, a rattle noise, a tic-tac noise, a big bang noise. So these are parameters which give us a warning as to something is wrong with this machine and once we have such an indicator as to as an alarm indicator as to what is wrong with the machine, we can always go back to the machine, instrument it with a vibration transducer and record it, analyze it and try to know more about the machine. Now at the conclusion of module 4 by now we would have, in fact I would have a half the semester would be towards module 4, we should have covered about 22 or about half the lectures by module 4 and that would be toward the mid semester of this course. And then comes actually the application of all the signal processing techniques, the instrument technique which you have studied in this course using it for practical analysis of particular defect like defect in a heart, defect in a lung, defect in a liver. Similarly, unbalance, once we have detected unbalance in a rotating shaft, how do I do a field balancing in situ or suppose there are two shafts which are supposed to be perfectly aligned by a coupling but they will not be aligned and one of them there will be small angular or a small offset misalignment and of course you know they will be all supported on bearings. So, if there is a misalignment which is there in fact in all machines, what are the characteristics of such misalignment vibrations, how do I reduce them and by no, what are their characteristics in the vibration signal? Suppose I have a vibration spectrum, how can I just say if I know the running rpm? By the one thing I should tell you, when I write 1 x that means it is at the rotational speed, 2 x means twice the rotational speed, 3 x means twice the rotational speed, x is the rotational speed, a rotational speed is rpm. So, rpm by 60 will be your in hertz. So, when I say a 1 x component that means a rotational speed, 2 x means twice the rotational speed, 3 x and so on and they are all harmonics. So, x means the rotational speed and subsequently we are going to use this terminology in the class, I thought I should tell you right now. So, misalignment does occur in plants and we have to know how to identify misalignment between meeting shafts. Next is you know there are lot of applications wherein cracks developed in shaft. So, how do I detect the crack in a shaft? Can you imagine I mean here you are having this given example, looking at this drawing you all can understand what I am coming at. So, this is a railway axle. Suppose, there is a crack on this axle which goes unnoticed and the train is running, you know the consequences now. Now, this is one way of it looking at it. Suppose this crack is inside this axle, the crack is developing. What could go wrong with it? How do you detect it? First is how do you detect it? Crack is not visible to the outside, crack is another case, crack is there on the shaft. So, how do you detect such cracks by vibration analysis and another is by ultrasonics? How do you use ultrasonics to detect cracks which occur inside the shaft? And many a times, next is lot of components on a shaft where there are pulleys, gears, there are cases wherein things will be loose. Once things become loose how do I detect looseness in the system? Suppose a pulley is loose on a shaft, how do I make sure that the components are not loose? And then looking at the different other components like ball and journal bearings, you know I was telling you about every component there is a ball bearing or an anti-friction bearing. How do you detect the defects in such ball bearings, defects in gears, compressors, pumps and carbines? So, these are cases which we will be seeing subsequently in module 5. And then the module 6, I mean till module 5, module 5 predominantly all the topics will be through vibration monitoring. And the module 6 is the remaining 30 percent of the CBM which I was telling you about wear debris analysis, NDT, ultrasonics, motor current analysis and this is what we are going to study about. In fact, the lecture number 32 and 33 is on a contaminant analysis and oil analysis. So, we will be studying what is known as the wear debris analysis in chapter or lecture 32 and 33. And then 34 and 35 is towards fault detection itself in electrical motors, transformers and how do I use motor current signature analysis for fault detection in mechanical systems as I was just telling you the case of a submersible pump. And the lecture on 36 is on thermography because as you will know if there is loose contact between conductors there will be arcing and there will be heat generation. If there is a misalignment because of friction there will be heat generation. So, by detecting the heat generated at a bearing location at a coupling location at a contact point of an electrical switch gear I can tell what is wrong with that machine. So, thermography is a very very powerful non-contacting method of finding out fault in a mechanical system. Ultrasonics again as I was telling you the example of this shaft which has a crack in the inside which is not visible from the outside. How by using an ultrasonic probe we can detect the crack in say for the case of a railway axle. We have a railway workshop next door in Kharagpur where in every train axle comes for a periodic ultrasonic inspection. They have a particular schedule at which every train every each and every train axle railway axle is tested for ultrasonic floor detection or crack detection menu. In fact I know in Sweden there are places wherein they have camera installed in the platform itself below the platform so that they will take the image of the bearing thermal image of the bearing and then transmit to a central database to tell you the condition of the bearing. I am sure you all would have noticed even in the railway stations nowadays you know you will see that class is going with an infrared temperature detector. Sometimes they used to have a rod they used to feel if the bearings are hot or not I know if you have seen that, but you will see that in Kharagpur railway station when you wait for the trains there will be a guy who will be coming and checking all the temperature of the bearings by infrared thermometer and that is the lot of practical applications are there in thermography. And then we will talk about ultrasonics acoustic emissions radiography and I will conclude this course with a few case studies from the industry from steel plants, paper mills, ships etc. And with this I would close this introductory lecture on machinery for diagnostics and signal processing. Thank you.