Experimental Modal Analysis A Simple Non-Mathematical Presentation Peter Avitabile, University of Massachusetts Lowell, Lowell, Massachusetts Often times, people ask some simple questions regarding responds to the frequency where peaks in the frequency re- modal analysis and how structures vibrate. Most times, it is sponse function reach a maximum (Figure 4). So you can see impossible to describe this simply and some of the basic un- that we can use either the time trace to determine the frequency derlying theory needs to be addressed in order to fully explain at which maximum amplitude increases occur or the frequency some of these concepts. However, many times the theory is just response function to determine where these natural frequen- a little too much to handle and some of the concepts can be cies occur. Clearly the frequency response function is easier described without a rigorous mathematical treatment. This to evaluate. article will attempt to explain some concepts about how struc- Most people are amazed at how the structure has these natu- tures vibrate and the use of some of the tools to solve struc- ral characteristics. Well, what’s more amazing is that the de- tural dynamic problems. The intent of this article is to sim- formation patterns at these natural frequencies also take on ply identify how structures vibrate from a nonmathematical a variety of different shapes depending on which frequency is perspective. With this being said, let’s start with the first ques- used for the excitation force. tion that is usually asked: Now let’s see what happens to the deformation pattern on the structure at each one of these natural frequencies. Let’s place 45 evenly distributed accelerometers on the plate and Could you explain modal analysis for me? measure the amplitude of the response of the plate with dif- ferent excitation frequencies. If we were to dwell at each one In a nutshell, we could say that modal analysis is a process of the frequencies – each one of the natural frequencies – we whereby we describe a structure in terms of its natural char- would see a deformation pattern that exists in the structure acteristics which are the frequency, damping and mode shapes (Figure 5). The figure shows the deformation patterns that will – its dynamic properties. Well that’s a mouthful so let’s explain result when the excitation coincides with one of the natural what that means. Without getting too technical, I often explain frequencies of the system. We see that when we dwell at the modal analysis in terms of the modes of vibration of a simple first natural frequency, there is a first bending deformation plate. This explanation is usually useful for engineers who are pattern in the plate shown in blue (mode 1). When we dwell new to vibrations and modal analysis. at the second natural frequency, there is a first twisting defor- Let’s consider a freely supported flat plate (Figure 1). Let’s mation pattern in the plate shown in red (mode 2). When we apply a constant force to one corner of the plate. We usually dwell at the third and fourth natural frequencies, the second think of a force in a static sense which would cause some static bending and second twisting deformation patterns are seen in deformation in the plate. But here what I would like to do is to green (mode 3) and magenta (mode 4), respectively. These de- apply a force that varies in a sinusoidal fashion. Let’s consider formation patterns are referred to as the mode shapes of the a fixed frequency of oscillation of the constant force. We will structure. (That’s not actually perfectly correct from a pure change the rate of oscillation of the frequency but the peak force mathematical standpoint but for the simple discussion here, will always be the same value – only the rate of oscillation of these deformation patterns are very close to the mode shapes, the force will change. We will also measure the response of the from a practical standpoint.) plate due to the excitation with an accelerometer attached to These natural frequencies and mode shapes occur in all one corner of the plate. structures that we design. Basically, there are characteristics Now if we measure the response on the plate we will notice that depend on the weight and stiffness of my structure which that the amplitude changes as we change the rate of oscilla- determine where these natural frequencies and mode shapes tion of the input force (Figure 2). There will be increases as well will exist. As a design engineer, I need to identify these fre- as decreases in amplitude at different points as we sweep up quencies and know how they might affect the response of my in time. This seems very odd since we are applying a constant structure when a force excites the structure. Understanding the force to the system yet the amplitude varies depending on the mode shape and how the structure will vibrate when excited rate of oscillation of the input force. But this is exactly what helps the design engineer to design better structures. Now there happens – the response amplifies as we apply a force with a is much more to it all but this is just a very simple explanation rate of oscillation that gets closer and closer to the natural fre- of modal analysis. quency (or resonant frequency) of the system and reaches a So, basically, modal analysis is the study of the natural char- maximum when the rate of oscillation is at the resonant fre- acteristics of structures. Understanding both the natural fre- quency of the system. When you think about it, that’s pretty quency and mode shape helps to design my structural system amazing since I am applying the same peak force all the time for noise and vibration applications. We use modal analysis to – only the rate of oscillation is changing! help design all types of structures including automotive struc- This time data provides very useful information. But if we tures, aircraft structures, spacecraft, computers, tennis rackets, take the time data and transform it to the frequency domain golf clubs . . . the list just goes on and on. using the Fast Fourier Transform then we can compute some- Now we have introduced this measurement called a fre- thing called the frequency response function (Figure 3). Now quency response function but exactly what is it? there are some very interesting items to note. We see that there are peaks in this function which occur at the resonant frequen- Just what are these measurements cies of the system. And we notice that these peaks occur at fre- that are called FRFs? quencies where the time response was observed to have maxi- mum response corresponding to the rate of oscillation of the The frequency response function is very simply the ratio of input excitation. the output response of a structure due to an applied force. We Now if we overlay the time trace with the frequency trace measure both the applied force and the response of the struc- what we will notice is that the frequency of oscillation at the ture due to the applied force simultaneously. (The response can time at which the time trace reaches its maximum value cor- be measured as displacement, velocity or acceleration.) Now SOUND AND VIBRATION/JANUARY 2001 1 8 Output Response 5 2 8 Input Force 1 2 3 0 Figure 1. Simple plate excitation/response model. −3 8 7 Increasing Rate of Oscillation −6 Figure 6. A 3 DOF model of a beam. the measured time data is transformed from the time domain to the frequency domain using a Fast Fourier Transform algo- rithm found in any signal processing analyzer and computer software packages. Due to this transformation, the functions end up being com- Figure 2. Simple plate response. plex valued numbers; the functions contain real and imaginary components or magnitude and phase components to describe the function. So let’s take a look at what some of the functions might look like and try to determine how modal data can be extracted from these measured functions. Let’s first evaluate a simple beam with only 3 measurement e ud locations (Figure 6). Notice that the beam has 3 measurement nit g locations and 3 mode shapes. There are 3 possible places that a M forces can be applied and 3 possible places where the responses can be measured. This means that there are a total of 9 pos- sible complex-valued frequency response functions that could Frequency be acquired; the frequency response functions are usually de- scribed with subscripts to denote the input and output loca- Figure 3. Simple plate frequency response function. tions as h (or with respect to typical matrix notation this out,in would be h ). row,column Figure 7 shows the magnitude, phase, real and imaginary parts of the frequency response function matrix. (Of course, I am assuming that we remember that a complex number is made up of a real and imaginary part which can be easily converted to magnitude and phase. Since the frequency response is a complex number, we can look at any and all of the parts that can describe the frequency response function.) Now let’s take a look at each of the measurements and make some remarks on some of the individual measurements that could be made. First let’s drive the beam with a force from an impact at the tip of the beam at point 3 and measure the re- sponse of the beam at the same location (Figure 8). This mea- surement is referred to as h . This is a special measurement 33 referred to as a drive point measurement. Some important char- acteristics of a drive point measurement are: Figure 4. Overlay of time and frequency response functions. (cid:1) All resonances (peaks) are separated by anti-resonances. (cid:1) The phase loses 180 degrees of phase as we pass over a reso- nance and gains 180 degrees of phase as we pass over an anti- resonance. (cid:1) The peaks in the imaginary part of the frequency response function must all point in the same direction. Mode 1 Mode 3 So as I continue and take a measurement by moving the impact force to point 2 and measuring the response at point 3 and then moving the impact force on to point 1 to acquire two more measurements as shown. (And of course I could continue on to collect any or all of the additional input-output combi- nations.) So now we have some idea about the measurements that we Mode 2 Mode 4 could possibly acquire. One important item to note is that the frequency response function matrix is symmetric. This is due to the fact that the mass, damping and stiffness matrices that describe the system are symmetric. So we can see that h = h ij ji – this is called reciprocity. So we don’t need to actually mea- Figure 5. Simple plate sine dwell responses. sure all the terms of the frequency response function matrix. 2 SOUND AND VIBRATION/JANUARY 2001 One question that always seems to arise is whether or not it 0 is necessary to measure all of the possible input-output com- B binations and why is it possible to obtain mode shapes from e, d only one row or column of the frequency response function ud matrix. gnit a M Why is only one row or column −800 of the FRF matrix needed? B d a de, It is very important for us to understand how we arrive at nitu g mode shapes from the various measurements that are available a M in the frequency response function matrix. Without getting −80 mathematical, let’s discuss this. 0 B Let’s just take a look at the third row of the frequency re- e, d sponse function matrix and concentrate on the first mode. If I ud look at the peak amplitude of the imaginary part of the fre- gnit a quency response function, I can easily see that the first mode M shape for mode 1 can be seen (Figure 9a). So it seems fairly −80 straightforward to extract the mode shape from measured data. 180˚ A quick and dirty approach is just to measure the peak ampli- tude of the frequency response function for a number of dif- se a ferent measurement points. Ph Now look at the second row of the frequency response func- tion matrix and concentrate on the first mode (Figure 9b). If I −180˚ 180˚ look at the peak amplitude of the imaginary part of the fre- quency response function, I can easily see that the first mode e s shape for mode 1 can be seen from this row also. b ha P We could also look at the first row of the frequency response function matrix and see the same shape. This is a very simple −180˚ pictorial representation of what the theory indicates. We can 180˚ use any row to describe the mode shape of the system. So it is e very obvious that the measurements contain information per- s a h taining to the mode shapes of the system. P Let’s now take a look at the third row again and concentrate −180˚ on mode 2 (Figure 9c). Again if I look at the peak amplitude of the imaginary part of the frequency response function, I can 0.5 easily see the second mode shape for mode 2. And if I look at the second row of the frequency response eal R function matrix and concentrate on the second mode, I will be a little surprised because there is no amplitude for the second −0.5 mode (Figure 9d). I wasn’t expecting this but if we look at the 0.5 mode shape for the second mode then we can quickly see that this is a node point for mode 2. The reference point is located c eal R at the node of the mode. So this points out one very important aspect of modal analy- −0.5 sis and experimental measurements. The reference point can- 0.5 not be located at the node of a mode otherwise that mode will not be seen in the frequency response function measurements al e R and the mode cannot be obtained. Now we have only used 3 measurement points to describe −0.5 the modes for this simple beam. If we add more input-output measurement locations then the mode shapes can be seen more 1 clearly as shown in Figure 10. The figure shows 15 measured y frequency response functions and the 3 measurement points nar gi used in the discussion above are highlighted. This figure shows a m the 15 frequency response functions in a waterfall style plot. I Using this type of plot, it is much easier to see that the mode −1 1 shapes can be determined by looking at the peaks of the imagi- nary part of the frequency response function. ary d n Now the measurements that we have discussed thus far have gi a been obtained from an impact testing consideration. What if m I the measured frequency response functions come from a shaker −1 test? 1 y What’s the difference between a nar gi shaker test and an impact test? ma I −1 From a theoretical standpoint, it doesn’t matter whether the 0 400 400 40 measured frequency response functions come from a shaker Frequency, Hz Frequency, Hz Frequency, Hz test or an impact test. Figures 11a and 11b show the measure- ments that are obtained from an impact test and a shaker test. Figure 7. Response measurements of a 3 DOF model of a beam: a) mag- An impact test generally results in measuring one of the rows nitudes; b) phases; c) real components; and d) imaginary components. SOUND AND VIBRATION/JANUARY 2001 3 a 1 2 3 b 1 a 2 3 c h d 33 ? ? 1 2 3 Figure 9. a) Mode 1 from third row of FRF matrix. b) Mode 1 from sec- ond row of FRF matrix. c) Mode 2 from third row of FRF matrix. d) Mode 2 from second row of FRF matrix. 1 Mode #1 b 2 Mode #2 Mode #3 3 DOF #1 DOF #2 DOF #3 h 32 1 2 3 Figure 10. Waterfall plot of beam frequency response functions. of the frequency response function matrix whereas the shaker test generally results in measuring one of the columns of the frequency response function matrix. Since the system matri- 1 ces describing the system are square symmetric, then reciproc- ity is true. For the case shown, the third row is exactly the same as the third column, for instance. c Theoretically, there is no difference between a shaker test 2 and an impact test. That is, from a theoretical standpoint! If I can apply pure forces to a structure without any interaction between the applied force and the structure and I can measure 3 response with a massless transducer that has no effect on the structure – then this is true. But what if this is not the case? Now let’s think about performing the test from a practical standpoint. The point is that shakers and response transduc- ers generally do have an effect on the structure during the modal test. The main item to remember is that the structure h under test is not just the structure for which you would like to 31 obtain modal data. It is the structure plus everything involved in the acquisition of the data – the structure suspension, the mass of the mounted transducers, the potential stiffening ef- fects of the shaker/stinger arrangement, etc. So while theory Figure 8. a) Drive point FRFs for reference 3 of the beam model. b&c) tells me that there shouldn’t be any difference between the Cross FRFs for reference 3 of the beam model. impact test results and the shaker test results, often there will 4 SOUND AND VIBRATION/JANUARY 2001 be differences due to the practical aspects of collecting data. The most obvious difference will occur from the roving of accelerometers during a shaker test. The weight of the accel- 1 2 3 erometers may be extremely small relative to the total weight of the whole structure, but their weight may be quite large rela- tive to the effective weight of different parts of the structure. This is accentuated in multi-channel systems where many ac- 1 celerometers are moved around the structure in order to acquire all the measurements. This can be a problem especially on lightweight structures. One way to correct this problem is to a 2 mount all of the accelerometers on the structure even though only a few are measured at a time. Another way is to add dummy accelerometer masses at locations not being measured; 3 this will eliminate the roving mass effect. Another difference that can result is due to the shaker/stinger effects. Basically, the modes of the structure may be affected by the mass and stiffness effects of the shaker attachment. While we try to minimize these effects, they may exist. The purpose of the stinger is to uncouple the effects of the shaker h h 33 from the structure. However, on many structures, the effects 31 of the shaker attachment may be significant. Since an impact test does not suffer from these problems, different results may h be obtained. So while theory says that there is no difference 32 between a shaker test and an impact test, there are some very basic practical aspects that may cause some differences. What measurements do I actually make to compute the FRF? The most important measurement that is needed for experi- b mental modal analysis is the frequency response function. Very h simply stated, this is the ratio of the output response to the 13 3 input excitation force. This measurement is typically acquired using a dedicated instrument such as an FFT (Fast Fourier Transform) analyzer or a data acquisition system with software 1 that performs the FFT. Let’s briefly discuss some of the basic steps that occur in the acquisition of data to obtain the FRF. First, there are analog 2 signals that are obtained from our measuring devices. These h analog signals must be filtered to assure that there is no aliasing 23 3 of higher frequencies into the analysis frequency range. This is usually done through the use of a set of analog filters on the front-end of the analyzer called anti-aliasing filters. Their func- tion is to remove any high frequency signals that may exist in the signal. The next step is to digitize the analog signal to form a digi- h tal representation of the actual signal. This is done by the ana- 33 log to digital converter that is called the ADC. Typically this digitization process will use 10, 12 or 16 bit converters; the more bits available, the better the resolution possible in the Figure 11. a) Roving impact test scenario. b) Roving response scenario. digitized signal. Some of the major concerns lie in the sampling and quantization errors that could potentially creep into the pletely removed. digitized approximation. Sampling rate controls the resolution Once the data are sampled, then the FFT is computed to form in the time and frequency representation of the signals. Quan- linear spectra of the input excitation and output response. tization is associated with the accuracy of magnitude of the Typically, averaging is performed on power spectra obtained captured signal. Both sampling and quantization can cause from the linear spectra. The main averaged spectra computed some errors in the measured data but are not nearly as signifi- are the input power spectrum, the output power spectrum and cant and devastating as the worst of all the signal processing the cross spectrum between the output and input signals. errors – leakage! Leakage occurs from the transformation of These functions are averaged and used to compute two im- time data to the frequency domain using the Fast Fourier Trans- portant functions that are used for modal data acquisition – the form (FFT). frequency response function (FRF) and the coherence. The The Fourier Transform process requires that the sampled coherence function is used as a data quality assessment tool data consist of a complete representation of the data for all time which identifies how much of the output signal is related to or contain a periodic repetition of the measured data. When the measured input signal. The FRF contains information re- this is satisfied, then the Fourier Transform produces a proper garding the system frequency and damping and a collection of representation of the data in the frequency domain. However, FRFs contain information regarding the mode shape of the when this is not the case, then leakage will cause a serious system at the measured locations. This is the most important distortion of the data in the frequency domain. In order to measurement related to experimental modal analysis. An over- minimize the distortion due to leakage, weighting functions view of these steps described is shown in Figure 12. called windows are used to cause the sampled data to appear Of course, there are many important aspects of measurement to better satisfy the periodicity requirement of the FFT. While acquisition, averaging techniques to reduce noise and so on, windows greatly reduce the leakage effect, it cannot be com- that are beyond the scope of this presentation. Any good refer- SOUND AND VIBRATION/JANUARY 2001 5 ence on digital signal processing will provide assistance in this Analog Signals area. Now the input excitation needs to be discussed next. Basically, there are two commonly used types of excitation for Input Output experimental modal analysis – impact excitation and shaker excitation. Antialiasing Filters Now let’s consider some of the testing considerations when performing an impact test. Autorange Analyzer What are the biggest things to ADC Digitizes Signals think about when impact testing? Input Output There are many important considerations when performing Apply Windows impact testing. Only two of the most critical items will be mentioned here; a detailed explanation of all the aspects per- Input Output taining to impact testing is far beyond the scope of this article. First, the selection of the hammer tip can have a significant Compute FFT Linear Spectra effect on the measurement acquired. The input excitation fre- Linear Linear quency range is controlled mainly by the hardness of the tip Input Output selected. The harder the tip, the wider the frequency range that Spectrum Spectrum is excited by the excitation force. The tip needs to be selected Averaging of Samples such that all the modes of interest are excited by the impact force over the frequency range to be considered. If too soft a Computation of Averaged tip is selected, then all the modes will not be excited ad- Input/Output/Cross Power Spectra equately in order to obtain a good measurement as seen in Fig- Input Cross Output ure 13a. The input power spectrum does not excite all of the Power Power Power frequency ranges shown as evidenced by the rolloff of the Spectrum Spectrum Spectrum power spectrum; the coherence is also seen to deteriorate as well as the frequency response function over the second half Computation of FRF and Coherence of the frequency range. Typically, we strive to have a fairly good and relatively flat input excitation forcing function as seen in Figure 13b. The fre- quency response function is measured much better as evi- Frequency Response Function Coherence Function denced by the much improved coherence function. When per- forming impact testing, care must be exercised to select the Figure 12. Anatomy of an FFT analyzer. proper tip so that all the modes are well excited and a good frequency response measurement is obtained. a 40 The second most important aspect of impact testing relates Coherence to the use of an exponential window for the response trans- B ducer. Generally for lightly damped structures, the response d e, of the structure due to the impact excitation will not die down d nitu FRF to zero by the end of the sample interval. When this is the case, ag the transformed data will suffer significantly from a digital M Input Power Spectrum signal processing effect referred to as leakage. In order to minimize leakage, a weighting function referred −60 0 Frequency, Hz 800 to as a window is applied to the measured data. This window b is used to force the data to better satisfy the periodicity require- 40 Coherence ments of the Fourier transform process, thereby minimizing the B FRF distortion effects of leakage. For impact excitation, the most e, d common window used on the response transducer measure- gnitud Input Power Spectrum mtateinotn i os ft hthee e wxpinodnoewnt itaol lmy idneimcaiyzien lge awkiangdeo iws .s Thohwe nim inp lFeimguenre- a M 14. Windows cause some distortion of the data themselves and −600 Frequency, Hz 200 should be avoided whenever possible. For impact measure- ments, two possible items to always consider are the selection Figure 13. a) Hammer tip not sufficient to excite all modes. b) Hammer of a narrower bandwidth for measurements and to increase the tip adequate to excite all modes. number of spectral lines of resolution. Both of these signal processing parameters have the effect of increasing the amount of time required to acquire a measurement. These will both tend to reduce the need for the use of an exponential window Actual Time Signal and should always be considered to reduce the effects of leak- age. Sampled Signal Now let’s consider some of the testing considerations when performing a shaker test. Window Weighting What are the biggest things to think about when shaker testing? Windowed Time Signal Again, there are many important items to consider when performing shaker testing but the most important of those cen- ter around the effects of excitation signals that minimize the Figure 14. Exponential window to minimize leakage effects. need for windows or eliminate the need for windows alto- 6 SOUND AND VIBRATION/JANUARY 2001 gether. There are many other considerations when performing shaker testing but a detailed explanation of all of these is far Autoranging Averaging with Window beyond the scope of this presentation. One of the more common excitation techniques still used today is random excitation due to its ease of implementation. However, due to the nature of this excitation signal, leakage is a critical concern and the use of a Hanning window is com- monly employed. This leakage effect is serious and causes dis- tortion of the measured frequency response function even when windows are used. A typical random excitation signal with a 1 2 3 4 Hanning window is shown in Figure 15. As seen in the figure, the Hanning window weighting function helps to make the sampled signal appear to better satisfy the periodicity require- ment of the FFT process, thereby minimizing the potential distortion effects of leakage. While this improves the distor- tion of the FRF due to leakage, the window will never totally Figure 15. Shaker testing – excitation considerations using random remove these effects; the measurements will still contain some excitation with Hanning window. distortion effects due to leakage. Two very common excitation signals widely used today are Autoranging Averaging burst random and sine chirp. Both of these excitations have a special characteristic that do not require the need for windows to be applied to the data since the signals are inherently leak- age free in almost all testing situations. These excitations are relatively simple to employ and are commonly found on most signal analyzers available today. These two signals are shown 1 2 3 4 schematically in Figures 16 and 17. In the case of burst random, the periodicity requirement of the FFT process is satisfied due to the fact that the entire tran- sient excitation and response are captured in one sample in- Figure 16. Burst random excitation without a window. terval of the FFT. For the sine chirp excitation, the repetition of the signal over the sample interval satisfies the periodicity Autoranging Averaging requirement of the FFT process. While other excitation signals also exist, these are the most common excitation signals used in modal testing today. So now we have a better idea how to make some measure- ments. Tell me more about windows. They seem pretty important! 1 2 3 4 Windows are, in many measurement situations, a necessary evil. While I would rather not have to use any windows at all, the alternative of leakage is definitely not acceptable either. As discussed above, there is a variety of excitation methods that can be successfully employed which will provide leakage free Figure 17. Sine chirp excitation without a window. measurements and do not require the use of any window. How- ever, there are many times, especially when performing field 2 testing and collecting operating data, where the use of windows is necessary. So what are the most common windows typically Input h61 used. 1 4 Basically, in a nutshell, the most common widows used to- day are the Uniform, Hanning, Flat Top and Force/Exponen- 6 tial windows. Rather than detail all the windows, let’s just sim- 3 Output ply state when each is used for experimental modal testing. hout,in The Uniform Window (which is also referred to as the Rect- angular Window, Boxcar or No Window) is basically a unity 5 gain weighting function that is applied to all the digitized data Figure 18. Input-output measurement locations. points in one sample or record of data. This window is applied to data where the entire signal is captured in one sample or eral field signals usually fall into this category and require the record of data or when the data are guaranteed to satisfy the use of a window such as the Hanning window. periodicity requirement of the FFT process. This window can The Flat Top window is most useful for sinusoidal signals be used for impact testing where the input and response sig- that do not satisfy the periodicity requirement of the FFT pro- nals are totally observed in one sample of collected data. This cess. Most often this window is used for calibration purposes window is also used when performing shaker excitation tests more than anything else in experimental modal analysis. with signals such as burst random, sine chirp, pseudo-random The force and exponential windows are typically used when and digital stepped sine; all of these signals generally satisfy performing impact excitation for acquiring FRFs. Basically, the the periodicity requirement of the FFT process. force window is a unity gain window that acts over a portion The Hanning window is basically a cosine shaped weight- of the sample interval where the impulsive excitation occurs. ing function (bell shaped) that forces the beginning and end The exponential window is used when the response signal does of the sample interval to be heavily weighted to zero. This is not die out within the sample interval. The exponential win- useful for signals that generally do not satisfy the periodicity dow is applied to force the response to better satisfy the peri- requirement of the FFT process. Random excitations and gen- odicity requirement of the FFT process. SOUND AND VIBRATION/JANUARY 2001 7 Mode 1 2 1 4 aij1 ζζ 3 6 aij2 1 ζζ2 aij3 ζζ3 5 ωω ωω ωω 1 2 3 Figure 21. Breakdown of a frequency response function. Figure 19. Plate mode shapes for mode 1– peak pick of FRF. MDOF SDOF Mode 2 2 1 4 Figure 22. Curvefitting different bands using different methods. 3 6 How Many Points??? 5 Residual Effects Residual Figure 20. Plate mode shapes for mode 2 – peak pick of FRF. Effects Each of the windows has an effect on the frequency repre- sentation of the data. In general, the windows will cause a degradation in the accuracy of the peak amplitude of the func- tion and will appear to have more damping than what really How Many Modes??? exists in the actual measurement. While these errors are not totally desirable, they are far more acceptable than the signifi- cant distortion that can result from leakage. Figure 23. Curvefitting a typical FRF. So how do I get mode shapes from the plate FRFs? So now that we have discussed various aspects of acquiring measurements, let’s go back to the plate structure previously discussed and take several measurements on the structure. Input Time Force Output Time Response Let’s consider 6 measurement locations on the plate. Now there f(t) y(t) are 6 possible places where forces can be applied to the plate FFT and 6 possible places where we can measure the response on IFT the plate. This means that there are a total of 36 possible input output measurements that could be made. The frequency re- sponse function describes how the force applied to the plate causes the plate to respond. If we applied a force to point 1 and measured the response at point 6, then the transfer relation between 1 and 6 describes how the system will behave (Figure Input Spectrum Frequency Response Function Output Response 18). f(jω) h(jω) y(jω) While the technique shown in Figures 19 and 20 is adequate for very simple structures, mathematical algorithms are typi- Figure 24. Schematic overviewing the input-output structural response cally used to estimate the modal characteristics from measured problem. data. The modal parameter estimation phase, which is often referred to as curvefitting, is implemented using computer quency, damping and mode shapes – the dynamic characteris- software to simplify the extraction process. The basic param- tics. The measured FRF is basically broken down into many eters that are extracted from the measurements are the fre- single DOF systems as shown in Figure 21. 8 SOUND AND VIBRATION/JANUARY 2001 Figure 25. Measured operating displacements. These curvefitting techniques use a variety of different meth- Mode 1 Contribution Mode 2 Contribution ods to extract data. Some techniques employ time domain data while others use frequency domain data. The most common methods employ multiple mode analytical models but at times very simple single mode methods will produce reasonably good results for most engineering analyses (Figure 22). Basically, all of the estimation algorithms attempt to break down measure- ments into the principal components that make up the mea- sured data – namely the frequency, damping and mode shapes. Figure 26. Excitation close to mode 1. The key inputs that the analyst must specify are the band over which data are extracted, the number of modes contained ample here, we are only going to consider the response of the in these data and the inclusion of residual compensation terms plate assuming that there are only 2 modes that are activated for the estimation algorithm. This is schematically shown in by the input excitation. (Of course there are more modes, but Figure 23. Much more could be said concerning the estimation let’s keep it simple to start.) Now we realize that the key to of modal parameters from measured data, the tools available determining the response is the frequency response function for deciphering data and the validation of the extracted model between the input and output locations. Also, we need to re- but a detailed explanation is far beyond the scope of this ar- member that when we collect operating data, we don’t mea- ticle. sure the input force on the system and we don’t measure the All structures respond to externally applied forces. But many system frequency response function – we only measure the times the forces are not known or cannot be measured easily. response of the system. We could still measure the response of a structural system even First let’s excite the system with a sinusoid that is right at though the forces may not be measured. So the next question the first natural frequency of the plate structure. The response that is often asked concerns operating data. of the system for one frequency response function is shown in Figure 26. So even though we excite the system at only one frequency, we know that the frequency response function is the What are operating data? filter that determines how the structure will respond. We can see that the frequency response function is made up of a con- We first need to recognize that the system responds to the tribution of both mode 1 and mode 2. We can also see that the forces that are applied to the system (whether or not I can majority of the response, whether it be in the time or frequency measure them). So for explanation purposes, let’s assume for domain, is dominated by mode 1. Now if we were to measure now that we know what the forces are. While the forces are the response only at that one frequency and measure the re- actually applied in the time domain, there are some important sponse at many points on the structure, then the operating mathematical advantages to describing the forces and response deflection pattern would look very much like mode 1 – but in the frequency domain. For a structure which is exposed to there is a small contribution due to mode 2. Remember that an arbitrary input excitation, the response can be computed with operating data, we never measure the input force or the using the frequency response function multiplied by the input frequency response function – we only measure the output forcing function. This is very simply shown in Figure 24. response, so that the deformations that are measured are the The random excitation shown excites all frequencies. The actual response of the structure due to the input excitation – most important thing to note is that the frequency response whatever it may be. function acts as a filter on the input force which results in some When we measure frequency response functions and esti- output response. The excitation shown causes all the modes mate modal parameters, we actually determine the contribu- to be activated and therefore, the response is, in general, the tion to the total frequency response function solely due to the linear superposition of all the modes that are activated by the effects of mode 1 acting alone, as shown in blue, and mode 2 input excitation. Now what would happen if the excitation did acting alone, as shown in red, and so on for all the other modes not contain all frequencies but rather only excited one particu- of the system. Notice that with operating data, we only look at lar frequency (which is normally what we are concerned about the response of the structure at one particular frequency – when evaluating most operating conditions). which is the linear combination of all the modes that contrib- To illustrate this, let’s use the simple plate that we just dis- ute to the total response of the system. So we can now see that cussed. Let’s assume that there is some operating condition that the operating deflection pattern will look very much like the exists for the system; a fixed frequency operating unbalance first mode shape if the excitation primarily excites mode 1. will be considered to be the excitation. It seems reasonable to Now let’s excite the system right at the second natural fre- use the same set of accelerometers that were on the plate to quency. Figure 27 shows the same information as just discussed measure the response of the system. If we acquire data, we may for mode 1. But now we see that we primarily excite the sec- see something that looks like the deformation pattern shown ond mode of the system. Again, we must realize that the re- in Figure 25. Looking at this deformation, it is not very clear sponse looks like mode 2 – but there is a small contribution why the structure responds this way or what to do to change due to mode 1. the response. Why does the plate behave in such a complicated But what happens when we excite the system away from a fashion anyway??? This doesn’t appear to be anything like any resonant frequency? Let’s excite the system at a frequency mid- of the mode shapes that we measured before. way between mode 1 and mode 2. Now here is where we see In order to understand this, let’s take that plate and apply a the real difference between modal data and operating data. simple sinusoidal input at one corner of the plate. For the ex- Figure 28 shows the deformation shape of the structure. At first SOUND AND VIBRATION/JANUARY 2001 9 Experimental Finite Modal Element Testing Modeling Modal Perform Parameter Eigen Estimation Solution Develop Modal Mass Rib Stiffener Model Spring Repeat Figure 27. Excitation close to mode 2. Until Structural Desired Changes Characteristics Required No Are Obtained Yes Done Dashpot Use SDM to Evaluate Structural Changes Figure 30. Schematic of the SDM process. Frequency Finite Response Corrections Element Measurements Model Figure 28. Excitation somewhere between modes 1 and 2. Parameter Eigenvalue Output Spectrum Estimation Solver y(jω) Modal Model Modal Parameters Validation Parameters Frequency Response Function Synthesis of a Dynamic Modal Model Mass, Damping Structural Forced Real World Stiffness Changes Dynamics Response Forces Modification Simulation Modified Structural f(jω) Modal Response Data Input Spectrum Figure 31. Overall dynamic modeling process. Figure 29. Broadband plate excitation. glance, it appears that the deformation doesn’t look like any- thing that we recognize. But if we look at the deformation pat- tern long enough, we can actually see a little bit of first bend- ing and a little bit of first torsion in the deformation. So the operating data are primarily some combination of the first and second mode shapes. (Yes, there will actually be other modes but primarily mode 1 and 2 will be the major participants in the response of the system.) Now, we have discussed all of this by understanding the fre- quency response function contribution on a mode by mode basis. When we actually collect operating data, we don’t col- lect frequency response functions but rather we collect output Mass Rib Stiffener spectrums. If we looked at those, it would not have been very clear as to why the operating data looked like mode shapes. Spring Figure 29 shows a resulting output spectrum that would be measured at one location on the plate structure. Now the in- put applied to the structure is much broader in frequency and many modes are excited. But, by understanding how each of the modes contributes to the operating data, it is much easier to see how the modes all contribute to the total response of the Dashpot system. So actually, there is a big difference between operating de- Figure 32. Modal model characteristics. 10 SOUND AND VIBRATION/JANUARY 2001