is more involved and more interesting and at the end i'll talk about applications various applications applications where this guided web based entity can be used so there are many methods so this is the standard definition which i have taken from one of the literature Non-destructive testing used for testing materials, components or assemblies for discontinuities or differences in characteristics without destroying the serviceability of the part of the system. So that is the central idea that we don't disturb the serviceability of the component. Unlike the UTM where we find out the ultimate tensile strength and we have to break the specimen So that is a kind of destructive testing So non destructive testings are very important as far as industries are concerned So these are the various methods like magnetic particle testing liquid penetrant testing visual testing radiography ultrasonic Okay And out of this ultrasonic methods are largely used. These methods which are mentioned here like magnetic particle, liquid penetrant, visual testing, these are quite used for finding the damages which are on the surface. Of course radiographic methods can be used to find the damages.
out damages or defects which are inside the component. However, this method is very costly and it is not so handy. Ultrasonic methods are very handy.
You can carry a simple equipment with you anywhere and you can do the ultrasonic testing. But there are some issues with ultrasonic testing. Ultrasonic testing uses bulk waves.
These bulk waves are the waves which pass through the bulk of the medium. And these are very high megahertz frequency waves. And most important that ultrasonic testing transducers need coupling.
Now you can see one of the test setup I have shown here. This is a transducer which emits the wave and senses the wave. So this is how the...
pulse is emitted and it reflects from the end and it goes back to the transducer. So this is the initial pulse which you see which is entering the specimen and this last echo is the echo that corresponds to the end of the specimen. If you get any crack you will have the pulses reflecting from this crack too.
So this is a small peak which you see which is a crack echo. So it is very easy to locate the damage. However, if you have some damage very close to the surface, just imagine this crack echo will be just masked by the initial pulse and you won't be able to see.
So there is always a dead area for the ultrasonic testing. Okay, so that is the area over which this initial pulse lie. And whatever damage lie in that particular area cannot be detected.
Even the components which are very thin or very small in size, which have very rough surfaces, cannot be easily detected by the conventional ultrasonic testing. So under such circumstances, guided wave based methods are very useful. So guided waves are excited at a point and they...
move across the entire component and they are basically guided by the boundaries. You can see they are continuously reflected at the boundaries. And that mainly happens because of the change in the density. And of course...
Why do we use this coupland? Coupland we use at the interface of the transducer and the specimen mainly to reduce the impedance. Because there is a small air gap, if you don't have a coupland you will have a small air gap, then the waves will find large change in density.
Because they have to enter from metal to air and from air to metal. pulses will be just reflected back from the boundaries. If you have a liquid penetrant here, the change in density from metal to liquid is comparatively lesser than from metal to air and the pulses can easily enter.
So here in this case, these pulses are simply guided by the boundaries and waves can travel a long distance. So you just excite at one point and a large area can be scanned. If you do C-scan using the ultrasonic testing, in conventional ultrasonic testing, you have to move this transducer from one place to other place and across the entire area.
In this case, it is not required. You just excite waves at one place and the wave will just go to all the places across the component and it can detect the damage. And how that detects the damage, that I will be telling during the presentation. Now there are different types of guided waves depending upon the components and how they are excited. So this is first type which I want to tell here is the Rayleigh wave.
These waves are named after the people who actually invented those, I won't say invented, discovered by those particular scientists. So Rayleigh was almost the first person who talked about these guided waves and he talked about this surface wave. So you can see this is the Rayleigh wave which Which propagates over the surface okay only on one surface so specifically the components which are very thick in that case if you Excite the waves they propagate over the surface one surface only they don't reach to the another surface Okay, like this case And you can see the waves are like there is a particle motion and the motion is just like a caterpillar whereas the low waves low was a scientist you can see they are also surface waves but here these waves are sheer horizontal they are moving like a snake okay.
and the largely used waves are the Lamb waves. Lamb waves basically propagate in thin plates and they need both the boundary conditions for their motion. So basically the way the waves are reflected, they combine together and they form different mode shapes. These modes are different than the modes what we see in the vibration.
These are the modes of the wave propagation. This is a symmetric lamp wave mode. This is an anti-symmetric lamp wave mode.
You can see here with respect to the center. this is symmetric motion of the particle and this is quite anti-symmetric. This is coming down, this is going down.
So, we can easily make out the difference. However, there are even n number of symmetric Lambeau modes and there are n number of anti-symmetric Lambeau modes depending upon the configuration what they form. Even we know for a continuous system we have n number of modes in case of vibration.
Okay similarly here also we have n number of modes. How to identify that modes that is very important and that can be easily done through the experiments. These are the two equations which govern symmetric modes and antisymmetric modes.
Here P and Q are the Lammertz constants which are expressed here. K is the wave number. CP and CS are the longitudinal and transverse velocities of the bulk wave. Okay. Omega is the angular frequency and D is the half thickness of the plate.
Okay. If you solve these two equations, you are going to get this dispersion. curves what are these dispersion curves you can see here this is between the frequency thickness product and the phase velocity and this is with respect to the group velocity so the velocities of the lambda is dependent on the frequency thickness product okay you can see here this is a naught this is s naught a naught means the fundamental anti-symmetric mode this is fundamental symmetric mode this is the first then a1 s1 a2 s2 a4 s4 a3 s3 and so on so a indicates anti-symmetric s indicates symmetric and this one from zero onwards one two three they indicate the number of that particular mode so phase velocity is the velocity with which the wave propagates if it has a single frequency content. But generally it doesn't happen. What happens when we excite the wave, we don't excite a certain 300 kilohertz wave.
We excite the wave in a certain frequency bandwidth and therefore we have a group of waves which travel together and that is what we get this group velocity okay so there are mathematical equations to get the phase velocity and group velocity group velocity is something which can be easily find out found out through the experiment so most of the studies you will find they find out the group velocities because it can be tallied through the experiments and also through the simulation So one thing is clear here that the velocity of the waves is dependent on the frequency also. If you say the thickness is a constant, it depends upon the frequency. If you have more number of frequency content in a pulse, you will have more number of pulses moving with different velocities.
And that is really a problem. So I will tell you what is that problem in the subsequent slides. But one thing you remember that velocity... of the wave is always dependent on the frequency.
Now how do you do the damage detection? I have here a simulation result. You can see this is the plate and at this point of at this particular point I am exciting the pulse.
Okay so I can see these are the repulse. So same type of repulse you see when you put a stone in water. Okay still water.
And if you sense the displacement of the particles at this particular point, you can see the wave passing through that particular point. And that is an A-scan you can easily get. Now see this particular plate, you have a small hole here and you can see this wave front is now totally disturbed.
And further you can see the clear disturbance here. If you take a point here and if you find... the data if you get the a scan at this particular point and to compare the data what you get here and in this particular case you can easily make out that those two data sets are totally different and hence you can make out that there is a damage present here because with respect to pristine specimen you always have a data with you and that is why these methods are called as baseline methods you If you don't have a pristine data with you, you cannot make out whether there is damage or there is no damage. However, there is one baseline free method, which I'll be explaining shortly, which does not need baseline data.
So I hope it must be clear to you that how we can find out the damage. Now further, this is the basic philosophy and further what you do is always you solve the puzzles. How do you get the data?
How do you process it? And what indices you show which are proportional to the damage, damage location, damage size? That is all the game of research. This is the simple experimental setup, the schematic.
You have a function generator here. This is a plate, let us say blue color. There is a function generator.
function generator creates a tone burst a pulse because you need to pass a pulse which has a voltage sinusoidal voltage let us say as you give the sinusoidal voltage if this is a piezoelectric material it will expand and contract accordingly because you are giving a pulse with positive voltage and negative voltage with positive voltage it will expand with negative it will contract and that contraction and expansion will exert stresses on this particular base plate because this transducer is bonded to this particular plate and those stress waves pass through the specimen and that is what we call bills okay and those waves are sensed by this sensor this sensor is connected of course it could be a piezoelectric sensor with the same reverse mechanism it senses the voltage because you know it can work both ways as actuator and sensor and that signal goes to the digital oscilloscope further goes to the computer for offline signal processing or maybe online signal processing so this is simple experimental setup This is how it looks further, function generator to amplifier. Amplifier you need sometimes when you want to have a very high voltage. And that is specifically required in case of non-linear ultrasonics. Now why that I will tell you in the next lecture. Okay, that is on Thursday so this is the setup where we have an amplifier this is a NF made this is of course the Amplifier which I have used for my research which is there in IIT Bombay aerospace department so this is nf made japan made amplifier there is oscilloscope functional data these are the things which we all know okay this is a piezoelectric wafer actuator this is a piezoelectric wafer sensor go to the oscilloscope and then to the computer so this is a very simple set up so anyone can start this basic experimentation even without the amplifier just need a function generator near oscilloscope even you don't need a computer from oscilloscope you can you just use the pen drive and you can extract the data and do the signal processing These piezoelectric actuators, the cheaper ones are available even for 10 rupees.
You can get one piezoelectric actuator and sensor. And if you want it a little sophisticated, in Pune there is one sparkler ceramics industry which can supply at 450 rupees per transducer. Okay, so very less amount is involved for starting up the research in this particular field. Now, you give something to the piezoelectric actuator, that is what I told, it is a tone burst and I will talk about that.
What actually you create in the function generator? Now function generator, basically you have some inbuilt pulses like sinusoidal, square pulse, but we cannot use them, we want a tone burst like this. So what we do is that we make such a tone burst in MATLAB.
okay then we uh if you are using tectronics amplifier you have a different uh electronics function generator you have a different procedure to take this particular pulse in the uh function generator it depends upon the uh what function generator you are using it every function generator has a different procedure too take the data from the computer to its own memory and then pass it on to the transducer. So for Tektronix there is an ARB express software I think where we can take this pulse and we can store in the function generator. so for that purpose you have to use arbitrary function generator where you can store an arbitrary signal now this is a sinusoidal pulse okay we don't use a continuous pulse we use a tone burst and most important we use a windowing function to window this particular tone burst you can see this shape now what that shape has to do here that shape is important to reduce the frequency bandwidth Different windowing functions have different frequency bandwidths. You can see here, these are the different windowing functions.
And you can see every windowing function has different frequency bandwidth. Frequency bandwidth is the range of the frequencies. Okay? Now you can see here high amplitude frequencies are present.
Now what do they do actually? What actually we need? We actually need a very narrow frequency bandwidth pulse.
Why? Because if you have more number of frequencies, what is going to happen the velocity of the wave is going to change okay because velocity is dependent on the frequency if you have more number of frequency there will be more number of pulses with different velocities and finally shape of your pulse will totally change so we choose what we choose actually generally we use hand window or a Gaussian window which is not here so these two windows are preferred if you see most of the research they will find you will find either hand or hamming or Gaussian so this is the pulse in time domain and this is the pulse in the frequency domain okay now how many cycles you should have in a pulse Okay, so that is shown here, a small study shown here. If you have less number of pulses, it looks very good in time domain, but in frequency domain, it has very high frequency bandwidth.
And there are more number of sidebands. If you increase the number of pulses, you can see here. It is getting little bad in the time domain but in frequency domain it is improving.
You can see this black has very less number of sidebands compared to the blue, green and red. So what should we do? We have to do a compromise. Okay? You have to carry out some experiments and you have to see how do you get the groove velocities.
Whether if you get the groove velocities with certain number of pulses very accurately, you can stick to that. So generally it comes between these. Don't go for very less number of cycles, very high number of cycles, go in between.
Okay? So generally I use 8.5 cycles. Of course different people use different cycles. If you are using in the linear time domain you can go for the lower numbers also.
But of course it depends upon to what level you are making a compromise with the group velocity. what transducers we use i already told you that visuality transducers we largely use pzt lead z conic titanium and so this is basically ceramic type of transducers and pvdf polyvinylidene fluoride these are very like flexible So PVDF are generally used as sensors. I mean to say they are not good to use as actuator.
Whereas PCAT can be easily used as actuator and sensor both. These are some new transducers, macro fiber composites. However, if you see the performance, what I have found out is that performance of PCAT is much better than MFC and performance of MFC is much better than PVDF.
So, PZT what costing I told you on the previous slide, those were for PZT. So, they are available cheap. There are few more transducers.
This is the transducer which you will find in some paper. This is a normal transducer and there is a wedge. So these are used where you need a very high amount of energy. thousand volt. So these transducers are capable of handling that much voltage and giving that much power and exciting the waves with that much intensity.
These wedges help you to direct the energy. okay you know there's a diffraction so you find out the angle use the snail's law and find out the angle at what angle you want the incident beam to penetrate or to go into the specimen and accordingly you decide this particular angle that is a wage angle and then this is very important as it is a laser Doppler of my bro better laser Doppler vibrameter Of course, one thing I forgot to tell here, in the PZP, these are the transducers which are generally bonded to the specimen using some adhesive. It may be epoxy adhesive or it may be cyanoacrylate adhesive. But it is always bonded to the specimen.
So you really don't need any coupland. Whereas in these transducers, this particular transducer, you need a coupland because this is like a very handy transducer which you can move from one place to another. to other place but every time you have to clamp it and you have to use the couplant okay for easy penetration of the waves from the transducer to the specimen. Laser Doppler Vibrometer, this is a single head Laser Doppler Vibrometer and this is the three head Laser Doppler Vibrometer. This is a sensor, this is not an actuator.
Of course, there are laser actuators too available in the market. But this is what I am talking about is the sensor. So this is a sensor with single head generally senses only Z displacement that is out of line displacement of the waves.
It senses basically first as a velocity and then you can expect acceleration displacement from that. And this three head. laser doppler vibrometer, sensors, displacement or velocity in all three directions, X, Y and Z. So this is very costly, whereas this is comparatively cheaper and of course it is very effective. But there are some issues with this.
you know this laser vibrometers basically works on the principle of reflection so you reflect the laser it incidents on the specimen which is vibrating and then some part of laser is reflected back and whatever reflected back is sensed by sensed back by this particular hair okay and then it makes a comparison what what has come back and what has been sent and how they are different and based on that they find out the velocity so there is some internal mechanism mechanism to do that. There are many reflective mirrors inside. But the thing is that since it works on the principle of reflection, if the surface of your specimen is not smooth, I mean to say if it is too rough, in that case the reflections will go elsewhere and the sensing will not happen properly.
Even if the specimen is too glossy, in that case also there is a lot of light. you can see and it is very difficult for the ldb to read so even so what we do is that for the even aluminum specimen we use retro reflective tapes okay which use fairly well reflection and in some cases where a large amplitudes are given in that case we can even spray paint the aluminum specimen even same thing can be done with the composites and that is what we have done it So there are some issues with LDB of course and here you just get only out of lane displacement with single head and that is why you miss the in-plane displacement that is some of the biggest and disadvantage of the single head laser doppler vibrometer and that can be easily overcome by the three head laser doppler vibrometer but it is too costly. Now how do you select the frequency for your experiments? Okay, so this is an input pulse and when you pass the pulse, this is the input pulse, which it is given to the actuator and this is what you receive at the sensor.
Now this is a single pulse and which gets disintegrated into symmetric mode and anti-symmetric mode. Of course there can be n number of modes but in certain range you get in the low frequency range you get only fundamental symmetric and fundamental anti-symmetric mode. Now what we do is that we find out the amplitudes of this S0 mode and A0 mode and we can plot it with respect to frequency. Now these are the amplitudes of the symmetric mode and anti-symmetric mode.
Now this is called as amplitude tuning or frequency tuning. Means now I want, now can you guess? Where can you get this response or at what frequency you can get such a response?
You just see the amplitude of S0 and amplitude of A0. You can see the right place is here near 100 kilohertz. Here the amplitude of S0 is lower because this is symmetric module.
Blue color is symmetric, red is anti-symmetric. So this is the frequency where this experiment is carried out because this is the amplitude of S0 and this is the amplitude of A0 which is bigger. So you can easily make out.
If you want in your experiment both amplitude to be same, you can choose this particular frequency where you get same amplitude, symmetric and antisymmetric mode. In some of the experiments we need only symmetric mode. In that case you choose this 300 kHz where you can get very high amplitude of symmetric and very low amplitude of anti-symmetric which is very negligible compared to S0 mode.
So this is the first step of choosing the frequency, what actually you want. And second point is that this selection of frequency has some wavelength connection. There is a thumb rule that if you want to find some particular damage and if that size of the damage is too small compared to the wavelength, it cannot be detected. That is a thumb rule which many researchers have proposed. So what is one point I want to say is that the wavelength should at least be same as that of the size of the damage.
So you said first of all what is the size of the damage you are wishing to find out or you are wishing to identify. And accordingly you can select the wavelength and from wavelength you can of course calculate the frequency. But one contradicting point I will tell you, this is not always true. Many scientists or many researchers, in fact in my own studies also, we have found out that even the damages which are much smaller than the wavelength have been found out in literature.
Because the... Because this reflection, because what actually, what damage does in the wave? Damage actually changes the amplitude phase of the wave.
So damage size should be large. enough to create a considerable change in the wave characteristics so that you can find out some characteristics which are because of the damage. If damage is not capable of making any change in the wave then the damage will remain undetected. So the impedance also makes a lot of difference.
If you have a very high impedance between the damage and the surrounding material, you can detect those damages and with this principle you can even detect the damages which are much smaller than the wavelength of the wave now what are the different strategies with which you can find out the damage so there are two main strategies pulse echo method and peach gas method in pulse echo method the transmission or the actuation and reception is on the same side okay so this is the only one transducer which actuates the wave and senses the wave so this is the pulse echo method and this method is very useful for locating the damage okay in peach catch you have an actuator and sensor as two separate transducers and which are located on two different sides of the damage so with peach catch it is quite difficult to locate the damage with the help of only two transducers So here you actually get an echo like conventional NDT. and you can find out the damage that is what is called as pulse echo in this case pitch catch you don't get echo what happens you pass the pulse it encounters with the damage and what happens there is a loss of energy because there is a distortion of the wave there is a loss of energy and there is a reduction in the amplitude and change in the phase and based on that change you make out whether there is damage or there is no damage But it is difficult to locate damage with only two transducers. However, if you use an array of transducers, then you can easily locate the damage with each catch. And I will tell you this is really a fun. It is just a puzzle.
You can use three transducers, you can use four, you can use six, you can use eight, you can use ten transducers to improve the accuracy of damage location. And it is really a puzzle. You can see there are many number of algorithms which have been developed in the industry. the literature which help in identifying the damage location okay so it is again I'm telling you it is as good as solving a puzzle to locate the damage because once this is actuator then these two are receiver this is an actuator this to a receiver this is actuator and these two are receiver and you get lot of data with you and then you play with the data and you then finally find out what is the possible damage location but of course this is a baseline method means you always have to have the data when there is no damage then only you can find out the damage location so this is the baseline method this is also baseline method whereas pulse echo is quite a baseline free method because you will get pulse echo only when there is a damage if there is no damage you won't get any echo because already you know the distance because one echo you will getting at the end of the this particular component that is about from the boundary so that is known okay so this is also baseline free method there is one more baseline free method okay i'll talk about after this so this is uh continuing with this uh locating damage of course as i told you there are different number of arrays you can use and you can find out the location of the damage This is another algorithm, one of the algorithms which can be used to locate the damage.
It is a computer tomography. The CT scan what you do in the laboratories, medical laboratories, the same way you can do the implement a CT scan for the locating the damage. So you can see these are all transducers and these are what these lines you see these are basically the paths of the waves.
This is the actuator, these all are the sensors. That's why you can see from one transducer you will find lines to all the sensors. And same for each transducer.
And you can see you get a dense grid. And by using appropriate algorithm, computer tomography algorithms, you can easily localize the damage. You can also use artificial intelligence. This is the artificial neural network.
Even you can use genetic algorithms with the same philosophy and you can localize the damage. So this is one ANN case I have shown here. See in ANN you have to put one damage and then you have to get the data.
You change the damage location and you keep on getting the data. So because you have to train the algorithm. that if my damage is at this place, this is what data I get. You have some index. If the damage location is here, this is the data what I get.
If damage location is here, this is data what I get. Once you train the algorithm, then you put a... randomly a damage okay and you get the data you send it to the ANN and then ANN will keep on trying different cases and then it will find a certain case that matches closely with the present case and it will say this is the damage location okay so this training is done in a grid manner and this is a very randomly of course if you do a random training it works very well This is the baseline free method which is also widely used. This is called as time reverse stability method.
So this is very interesting. What is done is that there are two transducers. From one transducer the wave is sent.
It is received at the second transducer. At second transducer this wave is reversed in time domain. and send back from PZT B to PZT A and from there you get your conclusion. Now how that happens?
So just take a case, I am passing this particular wave, this is a single pulse which I have excited in a plate from PZT A, okay, and it has entered now the specimen and it is going towards the PZT B. So when it is received at PZT B, what happens? This wave gets disintegrated into S0 and A0. Okay, now what I do at PZT B, whatever I get, I reverse them in the time domain, means You can see here S0 is first, A0 is second. Now what I do, I reverse the signal.
So I pass first A0 and then S0. You can see this scale and this scale. This is reversed.
So first I am passing A0 and then I am passing A0 at PZTB. Now these two pulses now enter the specimen and they start propagating. Now what happens?
This is anti-symmetric mode. A0 is also a pulse. So what it will do? It will disintegrate and S0 will also disintegrate. So if only A0 had been there, what would have happened?
You would have got the S0 of A0 and A0 of A0. just see the nomenclature s naught of a naught and a naught of a naught because as you see a single pulse always get disintegrated into two modes s naught and a naught so this a naught also gets disintegrated into s naught and a naught similarly s naught also gets disintegrated into s naught and a naught okay now you can see the velocity of a0 of a0 and s0 of a0 is same and they overlap because their velocities are same so what happens you get here s0 of a0 you get a0 of s0 and this a0 of a0 plus s0 of s0 now these two pulses are added so when you pass this a naught and s naught together this is the response what you get these two responses are just for sake of explanation This is the response which you get when you pass A0 and S0. Now what you have to do after this?
You take this particular pulse and this is the pulse which resembles the original pulse. That is this one. And you compare these two pulses.
If they are exactly same, means there is no damage. If there is a deviation, then there is a damage. So you can see this is a signal result which I have taken from a paper and you can see this continuous line is the input signal and this dotted line is the reconstructed input signal.
Reconstructed signal is this one. This is the reconstructed one. and what you have to do you just have to match this point of both the pulses okay and then you find either root mean squared deviation or you find percentage correlation so percentage correlation for undamaged specimen you will get very high level of percentage correlation 98.3 ideally it should be 100 but because of some errors you get 98.3 and if there is a damage you will get correlation very less because reconstruction won't be so accurate if there is a damage and therefore this signal will differ a lot from this particular signal and you will get less amount of percentage correlation or higher amount of root-bin-squared deviation so you can use either of these as the damage index so this is a very easy method of finding the damage but it is not so easy in experiments to reverse the signal in time domain and to pass and remember you have to have only symmetric mode and anti symmetric mode if you have some higher modes involved into it then it is very difficult to get this reconstructed signal Now, as I told you previously that when you do this experimentation, you need oscilloscope, you need function generator, you need amplifier in some of the cases, but these things are quite bigger and not handy to take them to the places where you want to do the NDP. So this is the setup which I have developed along with my students and you can see this setup is capable of generating a wave, then giving it to the transducer, then it takes the data from the sensor. It processes the data and it tells you whether there is damage or there is no damage.
So this is a small setup and it costs just between 5000 to 10,000. Okay, not more than that. So this is actually a video which I will show you.
So before that, I will explain here. Okay, there are... lights here. So this is the light is glowing means this is the idle mode.
Then you will have here one more light which glows when it goes into the detection mode. Green light will be on when there is no damage and the moment you create a damage red light will be on. if you see the name RTDDS it is real time damage detection system this is a system which can detect the damage 24 by 7 it is continuously on the moment damage happens in the system you will come to know okay and it uses very simple philosophy okay it just find out the root mean square deviation we store the baseline data and then we keep on getting the data after every 5 minutes the moment a damage happens okay you will find out that it is indicating that damage has happened i hope you can see this video so i have switched on it has gone into now detection mode is detecting so green light is on now there is no damage now here we are creating the damage and the moment we create the damage you can see the red light is on okay so this is how we detect the damage in real time of course we're in the process of putting iot into this so that on mobile you can get the message so this particular technique you can use for pipes you can use for rails and so many applications and that is what the applications which i have to tell you further so guided evo based linear methods for pipes you know pipes they are like veins for the industries okay specifically the chemical industries and a small leakage can lead to a huge accident so you can keep detecting the health of each and every pipe there are so many pipes and generally what they do in industry there is a scheduled ndt after certain time they do the ndt and of course if in between there is a leakage that comes as an accident and the company has to bear those expenditure maybe in terms of monetary expenses or in terms of life also so if you use such a technique what i have just shown you you can Basically save a lot of money. You have to just excite at a point and the entire pipeline can be scanned using the waves. You can see here there is one arrangement.
This is a schematic of an underground pipe. There is a transducer here. It is exciting the waves in both sides.
and these are the waves which are propagating so even an underground pipe can be scanned very easily you have to have just a small pit where you can access this transducer and you can get the data from here you can also use this method for detecting the uh damages in the pipes which carry the uh lpg this is the lpg line which comes to our houses And we have done a small study in Hong Kong. What, of course, that is applicable in India too. You see, the pipes come from outside in your house through the wall.
There is a small hole what they make. Okay, after that, they pack the hole. Now in the rain, the water seeps into that particular place from outside and there is a lot of water accumulation. You don't get it inside because that is somewhere inside the wall.
And what happens because of there is an accelerated corrosion. And what we found many places that those pipes are because most of these pipes are metallic, MS pipes and they get corroded. Sometimes the corrosion is so deep that there is a leakage of the gas. You don't come to know because it is inside the wall and maybe the gas may find a way outside and it may not come inside and you will never come to know there is a leakage of the gas until there is a smell inside. So first there is a loss of gas, you won't come to know, you will keep on getting the bill and at times there can be accident if the gas finds way inside your house.
So you can put the transducer, one transducer on this side, one transducer on the other side and you can easily make out whether there is a considerable corrosion or not. Because the damage characteristics will keep on changing based on the extent of corrosion. So there are different strategies when you use this method for pipes.
And this is something very complex you see. This is the dispersion curve when you deal with the pipes. In plate you have only anti-symmetric and symmetric mode. Whereas in pipe there are many number of modes.
So you have some axisymmetric modes and some non-axisymmetric modes. So axisymmetric modes are the longitudinal modes and the torsional mode. Whereas flexural mode Flexural mode is something which is non-axisymmetric.
I hope you can guess. Longitudinal is something like a to and fro. Okay.
That is how the sound waves propagate. Okay. Torsional, there is a twisting mode. Okay. And with the flexural, flexural, it goes like a waves.
Okay. Like your anti-symmetric mode. Okay. That way it goes.
Okay. And that way it is, of course, non-axisymmetric. and you can see there are few numbers after it L01, L02, L01, F11 so these numbers correspond to each specific mode so this 0 indicates that it is an axisymmetric and if this is not 0 it means it is non axisymmetric so this you can see F81 so this is a non axisymmetric but this 81 is a different family member of the flexural modes okay so this is how this nomenclature goes so you can see this L and T mode will always have 0 first where F will never have 0 first this is F 10 1 not 0 1 F 10 1 okay these are different families of the modes and you can make out actually you can see to if you this is a group velocity okay uh the group velocity of these wave brackets you can easily found out through the experiments how you can find out you can locate this particular point okay you get the time okay this time you can subtract from the time corresponding to the input pulse once you get that time you know the distance between the receiver and the exciter and distance divided by time you can get the velocity and that velocity you can search for here if at this particular frequency this is the velocity I am getting means that particular mode is F11 if I am getting this velocity it means it is T01 if I am getting this velocity at this particular frequency it means it is L01 so it is very easy to find out modes and you can also plot all these modes through simulation So this is L02, L01 and F81.
And there is a change if you concentrate on this particular mode, then you get this mode without damage, you get this mode with damage and find the RMSD or percentage correlation or you use some other index. And if those two pulses are exactly the same, you can say there is no damage. If there is a considerable difference, you can say there is a damage.
the experimental setup for pipes is also is almost same you can see here this is the exciter here this is a pitch catch method where you have exciter connected with the amplifier and arbitrary function generator okay so your amplifier is must because generally thickness of the pipes is higher so you have to use an amplifier then there is a receiver the signal goes to this is a receiver amplifier and goes to the digital oscilloscope because if the signal is too small which is sensed then for signal processing you need a considerable amplified data so for that you have a receiver amplifier of course this amplifier is of not that capacity as that of this So this is a pitch catch method. You can also deal with pipes using a pulse echo method. So this is a pulse echo method.
This is a transducer location here. Okay. There is a welding here. There is a trub. This is a bigger notch.
There is a weld here. Okay. And this is the data what we get here.
This is the transducer which act as actuator and also as a sensor so I get here first this is the echo corresponding to the way this way the first one the second echo is corresponding to this Trump third echo is corresponding to this world and there is a more conversion I will tell you maybe in the subsequent slides there's a more than origin means the one more gets converted into other mode so that is something typical happens with the guided wheels Okay, and this is the reflection corresponding to the end of the pipe so you can easily make out so that is the advantage of the pulse echo method that you can easily at one place you just actuate and sense and you can make out the location of the damage itself so pulse echo methods are largely used in the pipes so you can easily get into the research with the pipe so once you are aware of the guided waves you can go for any structure which is elongated okay the first example is pipe the second one i will tell is the rope ropes are very critical structures and mostly we don't maybe even you know we think of some critical parts rope actually we don't think of but rope is also very critical part in mechanical structures and civil structures oh but they are used in bridges then this overhead cranes now we have the entire like your life is dependent on the rope itself if rope breaks then it's gone then the ropes of the lift so this is one small damage which is shown for the rope okay so there are different sensors and actuators available for ropes so one example i'm showing which i've taken from this particular paper now uh this is a this is a like a specially made transducer for exciting the waves and receiving the waves this is yoke this is a magnet it's the electromagnetic transducer okay which passes the waves through the uh Rope and the same transistor is on the other side which receives the waves So, electromagnetic induction is also similar because what happens is that you are basically supplying a voltage which is varying sinusoidally. Accordingly, your voltage will be developed sinusoidally and whatever displacement which is imposed because of this electromagnetic induction is also sinusoidal in nature and that is what induces stress wave in the ropes. and the third one the last application which I'm going to tell is the guided view based linear methods for rails so for rails it is not being used currently okay still we use very conventional methods specifically in India we use very crude method use whenever you travel the trains now what is it not possible but whether you almost have private you must have seen a person which just goes along the rails and he keeps seeing this some whether there is a damage on the rail surface or not and then he keeps on tightening some bolts okay so that is the primary inspection and of course there is a special cart there is which actually do the inspection okay but still we don't have an inspection which is 24 by 7 You come to know some damages when they exceed their limit.
When it becomes a macro damage, a big damage, then only you see that even in conventional NDP. But when the damages are very small, they cannot be detected using conventional methods. However, guided wave based methods can detect such small damages.
so in the rail you have this is a head then this middle part is called as web and this is the foot which is fixed to the slipper okay and these are the possible damages what we can see here and this is how you can excite the waves in the web or also on the head you can use piezoelectric transducers and this is the possible location of course this is the actual something but this is signaling system this is not actually the damage detection system but this is a way which you can like connect some producer to the rail and you can excite the wave and if it is very high voltage and there is a lot of electricity available already so if you use that you can excite a wave which can travel really long One can also use laser. From this particular paper I have taken this image. So they are trying to find the damage, sub-surface damage by exciting waves in this particular head. so this is a line arrayed pattern which is generated by this particular signal this is a sagannac interferometer based optical system okay it creates a uh line arrayed pattern okay and uh what happens is that this is this impinges on the rail head intermittently okay and that creates the disturbance in the rail head which is nothing but the wave that propagates and the wavelength of that particular wave depends upon the spacing of these lines and then you use a normal laser head for measuring those waves. There are two lasers you can see.
This laser is the laser system which is actuator. You can see here there is a single head and that laser enters this entire lens system and it creates the line array pattern. This is a triaxial laser head.
okay and that is actually sensing the waves ok so this is how the laser actuator looks like now What we talked so far about, I did not mention anywhere that this is a linear method and this is non-linear. So whatever we talked was all linear. Because if you remember, I did not say anywhere that the frequency of the input pulse and the frequency of the output pulse, what we get, are different.
Because the frequencies are same. whatever frequency with which you excited the pulse and whatever output you got the symmetric and anti-symmetric mode if you find the frequencies of them they would be same as that of the input pulse okay so frequencies of input and output signals are same and these methods what we have talked about they are basically sensitive to bigger damages or like open cracks where there is an effective barrier to transmission if the barrier is not adequate to create some changes in the wave characteristics then that damage will remain undetected So, there has to be an effective barrier, there has to be an effective change in the wave characteristics because of the damage. So, what are the disadvantages or limitations we see here that we cannot detect small close damages and material degradation. Material degradation means whenever a damage is formed, before that so many activities happens in the bulk of the material.
that is a degradation some dislocations are formed then there are micro cracks formed and slowly those micro cracks get converted into micro cracks and then you see a bigger crack and final failure so such smaller cracks cannot be detected using the linear methods okay you can see here this is just specimen here this is an actuator piezoelectric web for actuator i am passing this anti-symmetric and symmetric mode let us say because i am passing only the pulse and that gets disintegrated into a naught and s naught Now there is a dismount present here. So at this particular point you get a reflection of the waves that is R and there is a mode convergence. Means the symmetric mode S0 will create its own A0 and S0 and A0 will create its own S0 and A0.
Why so? Because This is one thickness and the moment it enters the dismount, the thickness changes. The lower thickness is a new thickness for the waves which go below and this upper one is for the waves which go up. So the velocity changes drastically. because you know that the velocity is dependent on the frequency thickness product not only on the frequency but it is frequency into thickness okay so thickness matters a lot and at this point you are going to have reflection mode conversion mc and mt is the mode turning the mode that travels on the upper side takes a 180 degree u-turn and goes back and the mode which goes from bottom half or bottom part takes a 180 degree u-turn and goes on the upper half upper part that is called as mode turning and these are the mechanisms which have been proved in the literature that this is the thing which happens in the way of propagation so finally whatever you get at the whatever you get at the sensor is really chaotic because you have so many things which happen with the way propagation you don't get only an orderness not you get probably these things also and if you have a stringer some extension some protrusions on the specimen you have more conversion reflection turning and there is a junction where different wave packets are combined together and at the end whatever you get at the sensor is really chaotic and sometimes it is really difficult to make out anything about the damage and that is also a big problem with linear methods.
and for that purpose we use non-linear methods so non-linear methods are the methods which we will be discussing on thursday okay so i think Yeah, that is what I have to tell. So this is my email ID. So all of you can get into touch with me if you have some more questions or if you are aspirants of the research into this particular field, you can get into touch with me. Yeah, thank you very much.
So if you have any questions. Participants can ask the questions if you have some doubts. Hello, participants can ask a question. Excuse me, sir. Please show me last slide for a mail id.
Hello, sir. Yeah. Hello.
Yeah. So I was a good learning in by knowing by attending your session and knowing about this method of testing. So because I did not know about this method of testing, so just wanted to a background check that is this a this is not a conventional method of testing.
Right. So if people if I'm into that kind of research and need to get this done. So how can this be?