VIBRATION THEORY WwW 12 1.3 1.4 1.4.1 14.2 1.4.3 1.4.4 15 1.5.1 1.6 W7 Why Measure Vibration Sources of Vibration Effects of Vibration Nature of Vibration Periodic Functions Random Vibration Transient Vibration Stationary Random Vibration Measurement Parameters Relationship between Displacement, Velocity and Acceleration The Decibel Scale Peak, Average and RMS Values1.1, WHY MEASURE VIBRATION? It is necessary to measure vibration for several reasons. Whilst sometimes the vibration picture is obvious, there are many occasions when it is not, and it would be senseless to design some means of reducing for example a harmonic component frequency while vibration continues at the fundamental frequency. From the design point of view, the magnitude and frequency of the vibration need to be known to ensure that the stresses induced are not too great for the material to withstand. Another important reason is that if a certain piece of machinery has been found to resonate at a given frequency, running that machine at a speed which will excite that resonance can often be avoided. For vibration damping or isolating, it is necessary to know the amplitudes and frequencies involved in order to be able to select the correct damping materials. Preventive maintenance is another area where vibration monitoring is useful, since many faults (like worn teeth on a gear wheel or the start of a roller bearing’s failure) can be detected long before failure merely by noticing changes in the vibration spectra. 1.2, SOURCES OF VIBRATION Common causes of vibration are some part of moving machinery being out of balance, turbulent fluid flow, rattling of loose objects, impulses and shocks etc.: the list is endless. Vibration is normally undesirable, but it can often be reduced by careful design or by development ‘modifications. Sometimes, however, vibration is introduced to make things work, such as vibrating conveyor belts, mechanical hammers and even musical instruments, and in such cases resulting noise is unavoidable. It is then important to isolate it as much as possible, even to the extent of not letting your hi-fi loudspeakers rattle your neighbour's best ornaments on a shelf connected to your adjoining wall! 1.3. EFFECTS OF VIBRATION The effects of vibration are often serious. Humans subjected to vibration can be affected by blurred vision, loss of balance and consequent lack of ability to do their job properly. For machine ‘operators, lorry drivers or airline pilots etc., it is very important that this does not happen. In some cases, certain frequencies and levels of vibration can permanently damage internal body organs. Machinery can also be damaged by vibration. Obviously sensitive instruments etc. are best isolated from vibration, but more robust machinery can also be affected; if the vibration occurs at the resonance frequency of some component it can often be cracked or broken by fatigue, and nuts, bolts and rivets can be shaken apart. in some cases, say in an aircraft, this can be disastrous. Noise resulting from vibration is also often a serious problem, and can be a health hazard to people exposed to it for long periods. One really nasty thing about vibration is that it will not stay in one place unless special steps have been taken to isolate it. Vibration is transmitted through any solid object in contact with it, including the floor, walls, pipes, electrical conduit and any other mechanical linkage, which in turn cause anything in contact with them to vibrate and give out noise. Even if a particular frequency from one machine does not cause any of its own components to resonate, it could quite possibly excite a resonance in a connected machine. It is thus important that we should be able to measure and control vibration (and noise). 41.4. NATURE OF VIBRATION Vibration may be considered as an oscillating motion of a particle or body about a referenée position, Traditionally vibration signals are divided into these groups: A. Stationary Vibration B. _Non-Stationary Vibration 1, Periodic vibration 3. Transients 2. Stationary random vibrations 4, Shock 5. Non-stationary random vibration Simple examples of each are given here. 1.4.1, Periodic Functions Fig. 1.1. Example of a Simple Harmonic (Sinusoidal) vibration signal. (f = 1/T) The sinusoidal signal of Fig, 1.1 is by definition at one discrete frequency, which is given by 1/T, and has the units of Hz (Hertz) which is the number of cycles/second. T is the time taken for the wave to perform one complete cycle and come back to the same point again. This motion is periodic in that it repeats itself at regular intervals of time (T). The nearest that one would ‘come to a pure tone in vibration is the movement of the legs of a tuning fork. RAP Fig. 1.2. Example of Compound Harmonic periodic motion, (f+ 2f) The vibration of Fig, 1.2 is still periodic, because it repeats itself at regular intervals, but it is now compound harmonic (i.e. it is not purely sinusoidal). In fact, this particular example isa combination of two sine waves, as we shalll see later, and is the sort of vibration produced during the piston acceleration of an internal combustion engine. 12Thus a periodic signal need not be symmetrical, or even a nice geometric shape, and it can be made up of many combinations of frequencies of different amplitudes, as we shall now see. (N.B. The vertical axis of Figs. 1.1, 1.2, and 1.3 could equally well be labelled displacement, velocity or acceleration, since all these parameters vary in a similar manner with time.) VASA . Ror Fig. 1,3. Waveform of Fig. 1.2 split into its components The signal of Fig, 1.2 is made up of two components of different frequencies and amplitudes, and these are shown in Fig. 1.3, It can be seen in this case (although not always) that one sine wave has a frequency which is twice the other, and that its amplitude is much smaller. ‘The accelerations are summed arithmetically to obtain the combined signal level. An important and useful way of portraying the amplitudes and frequencies of all the different. components of a complex signal is the Frequency Spectrum, which is a plot of amplitude against frequency, and a few simple examples are given in Fig. 1.4, ‘As can be seen from Fig. 1.4 A, the simple sine wave has a spectrum consisting of one discrete frequency (f= 1/T), and is represented by a line whose height is related to the amplitude, The waveform of Figs. 1.2 and 1.3 has two components, each having a discrete frequency. Since one frequency is twice the other (f and 2f) the periods will also be in a ratio of two. [if the lower frequency f has a period of T: seconds (f = 1/T) then the higher frequency will have a period of T/2 seconds (2f = 2/T), and viceversa.] The two frequencies each have one line on the spectrum, whose height is related to its amplitude (Fig, 1.4 B). In this case, the frequency f would be called the Fundamental Frequency, since it is the main one, and the frequency 2f is its Second Harmonic, since it is twice the frequency of the fundamental. ‘A "square" wave is shown in Fig. 1.4.C, which is yet another example of a periodic function. This consists of a fundamental frequency f, and odd harmonics only (i.e. 1, 3, 5, 7), which drop off in level at a known defined rate. All periodic functions, such as those shown above can be defined by precise mathematical equations, which can make their analysis much simpler. Such functions are termed “Deterministic”. 1-3A Aeceteation (ams [pAceteraton (pAccetecaion (nu) f Aeseren : prceetcetion (ams) ret + c so Tee 7 fof ee oe n Fawn Fig. 1.4. Examples of (a) periodic signals and (b) their frequency spectra 1.4.2, Random Vibration The most commonly encountered type of vibration in everyday life is Random, i.e. it is continuous, but non-periodic, and contains many frequency components. Whilst a lot of these components will be related in the form of harmonics of certain frequencies due to machinery movements, many components will be entirely independent of these. saseran sce = ™ a t a Fig, 1.5. Example of Random Vibration and a typical Frequency Spectrum 1-4A random vibration could well look similar to that shown in Fig. 1.5, which shows its random nature, and lack of repetition, As can be seen, the spectrum (Fig. 1.5(b)) is “Broad Band”, i.e. it is continuous, since it contains all frequencies. It could also have some spikes, as shown, due to resonance and harmonics of certain vibrating components. These spikes are often caused by ‘the main source of vibration and once they are discovered from frequency analysis they are relatively easy to isolate. A random vibration is by definition non-deterministic; it can be represented mathematically only by a series of probability statements, since to obtain a complete description of the vibrations an infinitely long time record is theoretically necessary, which is obviously not Practical. A random vibration is called “Stationary” if successive samples taken of it are essentially the same in character, although the rigorous mathematical definition of this is somewhat more complicated. A stationary random vibration is normally easier to analyze than a non-stationary one. 1.4.3 Transient Vibration A transient vibration is one which is non-continuous. This type of vibration is found to occur due to impact, or during the starting up of a motor etc., or anywhere that the exciting force is not continuous. A transient vibration is often deterministic. Acceleration Aceleration ‘im Fig. 1.6. Example of a Transient Vibration and a typical Frequency Spectrum The sort of transient vibration shown in Fig. 1.6 would have a spectrum with many “lobes”, as shown, which would contain a whole range of frequencies. The size of the lobes {in terms of frequency range) will depend on the duration of the transient, and their relative amplitudes will depend on the shape of the transient itself. 1.4.4, Non-Stationary Random Vibration From a theoretical standpoint, all random vibrations are non-stationary because the statistical Parameters are changing over the time. In practice, however, these changes can be small or of a non-important character, that the vibrations may be considered stationary, 15Non-stationary random vibrations therefore, in practice, have to be described as being vibrations with significant changes in the statistical parameters during the relevant observation time (see Fig. 1.7). a Fig. 1.7. Non-Stationary Random Vibration 1.5. Measurement Parameters There are three main parameters used to describe vibration; they are Displacement, Velocity ‘and Acceleration, and all have their uses. Displacement is the distance moved by the measu Point from its natural position, velocity is the speed at which that point moves, and acceleration is the rate of change of its speed with time, Displacement is proportional to strain in the material, and acceleration to the force acting on the object. These parameters are measured in the normal units of distance, velocity and acceleration, which are, metres (m), metres per second (m/s) and metres per second squared (m/s) respectively. The units are in accordance with ISO R 1000 (SI units). Acceleration is also often measured in terms of gravitational constant (g), and it is common to see vibration ranges or levels quoted in terms of numbers of g. This is because it is easy to calibrate measuring equipment at levels of 1 g, and also because g is an internationally understood symbol, regardless of units. Historically, displacement was the first parameter to be measured because with slow moving machines and large displacements it was the easiest to see and the easiest to measure by optical ‘means. (It is not possible to measure velocity and acceleration optically.) Then, displacement limits were set by the law of elasticity, the fatigue of materials or by mechanical clearances. With the introduction of higher speed machines, the displacements became smaller and more difficult to see, but breakdown could still occur, so a move towards velocity measurements was made. Many rotating machines have a vibration frequency spectrum with a fairly even distribution 16of energy across the frequency range up to 1 kHz, (Fig, 1.7), thus making it reasonably easy to prescribe a velocity limit for a particular type of machine. (It is even easier to Prescribe for single resonance frequencies.) Standards organisations in several countries now recommend this type of limit expecially for electric motors, but also for some other types of machines. Measurement of displacement emphasises the Very low frequencies. Important vibrations at higher frequencies, which could play a major role in the safety, function or gustomer appeal of a product are not taken fully into consideration. Measurement of acceleration gives the reverse picture. More weight is put onto the higher frequencies, where it is often most ‘important to understand the vibrations. Measurement of velocity pays, in most cases, approximately equal attention to the vibration level at all frequencies. It is emphasised that these comments apply principally to rotating machinery over the frequency range up to 1 kHz. O no le | Fig. 1.8. Spectra showing the effects of measuring Displacement, Acceleration and Velocity on rotating machinery in the frequency range up to 1 kHz However, for high frequency vibration (above about 1 kHz), it is necessary to measure acceleration, because displacement and velocity drop off rapidly at high frequencies relative to acceleration. ‘The reason for this will be seen later. Since high frequencies play a large part in preventive maintenance, early detection of breakdowns, and vibration monitoring acceleration measurements are becoming very important. For example, many faults in machine elements like roller bearings can be detected in the frequency range 20 kHz — 50 kHz long before failure occurs, ‘thus making it possible to plan repairs and shutdown times. 71.5.1, RELATIONSHIP BETWEEN DISPLACEMENT, VELOCITY AND ACCELERATION There is a well defined relationship between these three parameters, which will now be discussed. (see also App. 1.) Fig. 1.9. Example of Simple Harmonic Motion Fig. 1.9 shows a wheel turning anticlockwise at a constant speed of w radian/s. (There are 2n radians in one revolution, so the wheel is turning at «/2n rev/s, or 30 «/ rev/min.) 6 is measured anticlockwise from the bottom x is measured to the right. The wheel drives a reciprocating mechanism B by means of a protruding stud P which is free to move in the slider S. B then moves backwards and forwards with Simple Harmonic Motion (provided there is no mass or friction in any of the sliding contacts) in the horizontal plane, Thus the displacement (x) of B will be as shown on the curve, and one cycle will be completed with each complete revolution of the wheel, In this case revolutions/second = cycles/second = Hz, and the standard relationship between angular speed w and frequency f is obtained: w = nf, The displacement x of B from its central position is given by: x=rsinwt (when @ = 0° or 180°, x = 0; 8 = 90°, x= +r; 6 = 270°, x =—r) The maximum displacement is given when sin wt = + 1, and is thus equal to r. Differentiating displacement with respect to time, we obtain the velocity v of B: ve9k= x=res coset (when 9 = 90° oF 270°, X= 0;0 = 0°, K= 41 0; 6 = 180°, x=—rw) The maximum velocity, V, occurs when coswt = + 1,and V = rw. Differentiating again with respect to time, we obtain the acceleration a of B: (When @ = 0° or 180°, X = 0; 6 = 90°, 2 sk =r? sinot % “ = 102; 6 = 270°, K = tw?) The maximum acceleration, A, occurs when sinwt = #1, and A = rw? 18The following diagram (Fig, 1.10) illustrates the effects of these relationships, Tine Fig. 1.10. Phase and Amplitude Relationships between Displacement, Velocity and Acceleration 't can be seen from Fig. 1.10 that there is a phase difference between these components. The velocity leads the displacement by 90° (or one quarter of a cycle), the acceleration leads the velocity by 90°, and so the acceleration leads the displacement by 180°. However, in terms of measurement this phase difference does not matter when the reading is averaged over several cycles (as it normally is in measurements; this will be mentioned later). It can also be seen that each differentiation has multiplied the signal by w. This does have some prac- tical significance. It means that at higher frequencies, the acceleration signal is the largest by a factor Of w or w?, and since w = 2nf, this often makes the acceleration signal the easiest one to measure, or sometimes the only one that can be measured, since the other signals drop into instrumentation noise. The main significance of the relationship between these parameters is that at any given frequency one ‘measurement can provide the other two. Differentiating the displacement will give velocity, and differentiating velocity will give acceleration, although these two are not normally performed. Integrating acceleration will give velocity, and integrating again will give displacement, and this is a fairly standard procedure. This integration is performed by use of an electrical integrating circuit, The emphasis of different parts of the spectrum can be seen quite clearly from the spectra of Fig. 1.7. In this case we have a fairly “flat” velocity spectrum. The division by w to obtain displacement attenuates high frequency signals (at high frequencies w is large), and in this case the signal dis- appears. The multiplication by w to obtain acceleration emphasizes high frequency signals. Thus, whilst acceleration may not be the parameter we wish to look at, at high frequency it is the one we would measure in order to obtain, say, velocity. When drawing spectra, we normally plot the frequency as a logarithmic scale, as it has the effect of expanding the lower frequencies and compressing the higher frequencies, and allowing us to have a reasonable resolution without large sheets of graph paper. Logarithmic plotting also ties in well 19with the concepts of an Octave (doubling of frequency) and a Decade (multiplying the frequency by 10) as one Octave is now a constant width on the paper, as is one Decade, regardless of the fre- ‘quency range it is covering. Linear scales are useful sometimes, however, particularly for sorting out harmonics. We also commonly use a logarithmic scale to plot vibration amplitudes. This gives us the Decibel (dB) Scale, which is also commonly used in acoustics, electronics and control theory. The Decibel will be introduced formally in the next paragraph, but part of the use of the two logarithmic scales of level and frequency can be seen in Fig. 1.11, where all the parameters now have straight line relationships with each other for one constant parameter. ée | Acceleration» 100% 9} 1% 20 Veocty ve B= ot * 0 ose 50 ~6.d8/0cve 01% 60 0.00% 70 NOSE LEVEL on 1 0 100 1000 He Freaveney asst Fig. 1.11, Relationship of levels of Displacement, Velocity and Acceleration In this case (Fig, 1.11) we have a flat acceleration spectrum and the velocity and displacement drop off with frequency at rates of 6 dB per octave (20 dB per decade) and 12 dB per octave (40 dB per decade) respectively. The displacement drops into instrumentation noise very rapidly, and the velo- city signal a little later. At very low frequencies, distortion and cut-off is introduced by the limi- tations of the measuring and integrating equipment. 1-101.6 THE DECIBEL SCALE Because of the extremely wide range of vibration levels to be measured, a relative logarithmic scale is used, which gives levels in Decibels (dB). For acceleration the definition is: a Pa = 10 logo (5) 8 re where a? is measured power, and 2,4” is a reference power. It will be readily seen that the following relationship also holds la = 20 en0(52-) a8 re where a and a, ¢ are now in terms of amplitudes of directly measurable quantities. The reference amplitude ag, is normally chosen to be 1g (or to coincide with the amplitude the system is being calibrated with). Other reference amplitudes are sometimes chosen for specific purposes, but when quoting absolute levels in dB it should be stated what the referer.ce level is, as the decibel is only a dimensionless ratio, not a unit itself. The output of, say , an accelerometer connected to an analyser and a meter can be read off directly in dB if required and there are no calculations to perform to obtain it. The absolute level of the measured signal is then obtained by simple subtraction of the level indicated by the reference signal. The biggest advantage of the dB is in keeping numerical values within reasonable bounds. Supposing for instance we had two power levels and that one was 100 times the other. Then 2 /a,42 = 100, so their ratio would be 10 log;9(100} = 20 dB. If they were in the ratio of 10°68" then we would have 10 logo (10'°) = 100 dB. Thus it can be seen that a ratio of 1010 in power levels can be contained in 100 dB. Care should be taken to ensure that we are certain whether amplitude or power is being used, because a ratio of a/aet = 100 in terms of amplitude gives us 20 log 49 (100) = 40 dB. Similarly, an amplitude ratio of 101° gives 20 logyg (10 '°) = 200 aB. Again, it will be clear that doubling the amplitude or the power level does NOT double the numerical value in dB, it adds 6 and 3 dB respectively. Consider the following examples. a) We want to add one power level Pa, to another power level Pa,, both 80 dB above a reference level. a a Pa = Pa; = 80dB = 10105» (;25) . 10 00,9 (23) ONG, re 2 a,” = {ajg4)? antilog, o sah = (qq)? 108 ‘Adding the powers, we get a,? + a)? = (a9)? - 2 - 108 112 Converting back to dB we get Pa, + Pa; =10 log, (at 10 logy (210°) = 10 log, 9 2+ 10 log,» 10° = 3+80 = 83 dB above the reference level. b) We have two amplitude levels (La, and La, ) of 80 dB above a reference level, La,=La,= 80¢dB = 20 05, (32) = 20 ono (=i,) ef a Aref = a, = arg antilog, g {3|- ye 10% Adding the amplitudes, we get a, +a, = a,¢¢° 2° 10* Converting back to dB we get La, +La, = 20 logy, (24) = 2 log, (2°10) = Fret 20 log, ) 2+ 20 log,, 10* = 6 +80 = 86 dB above the reference level. For the two other parameters velocity and displacement, the same definition is valid. For velocity (amplitude): L, = 20 log () dB ¥ Vo where v is the measured amplitude of the velocity and vz is the reference velocity amplitude. For displacement (amplitude): d Ly= 20 log q dB where d is the measured displacement amplitude and vy is the reference amplitude of the displacement. 112Although at first the decibel scale appears very awkward to handle and understand, its advantages the long run will become obvious. The effects of this logarithmic scale are to compress the higher levels and expand the lower ones, so that they can be represented on reasonably sized pieces of paper with reasonable accuracy, and the numerical values do not get so small or so farge that they get out of hand, either on paper or in a machine, (01.000 1.072 1.023 1.035 1.047 1.059 1.072 1.084 1,096 1.109 11.922 1.435 1.148 1.161 1.175 1.189 1.202 1.216 1.230 1.245, 21,259 1.274 1.288 1.303 11318 1.334 1.349 1.365 1.380 2.396 31.413 1.429 1.445 1.462 1,479 1.496 1.514 1.531 1.549 1.567 41,585 1.603 1.622 1.641 1.660 1.679 1.698 1.718 1.738 1.758 5 1.778 1.799 1.820 1.841 1.962 1.884 1.905 1.928 1.950 1.972 G 1.995 2.018 2.042 2.065 2.089 2.113 2.138 2.163 2.188 2.213 7 2239 2.265 2.291 2.317 2.944 2.371 2.399 2.427 2.455 2.483 8 2.812 2.841 2.570 2,600 2.630 2.661 2.692 2.723 2.754 2.786 9 2.818 2.851 2.884 2.917 2.951 2.985 3.020 3.055 3.090 3.126 10 3.162 3.199 3.296 3.273 3.311 3.950 3.388 3.428 3.467 3.508 11 3.548 3.589 3.697 3.673 3.715 3.758 3.202 3.846 3.290 3.996 123.981 4.027 4,074 4.121 4.169 4.217 4.266 4.915 4.265 4.416 13 4467 4.519 4.571 4.626 4.677 4.732 4.786 4.842 4.898 4.955 14 8.012 5.070 5.129 5.188 5.248 5.309 5.370 $.433 5.498 5.569 15 5.623 5.689 5.754 5.821 5.888 5.957 6.026 6.096 6.165 6.237 16 6.310 6.383 6.457 6.531 6.607 6.683 6.761 6.839 6.918 6.998 17 7.079 7.161 7.244 7.328 7,413 7.499 7.586 7.674 7,762 7.852 18 7.943 6.095 8.128 8.222 8.318 8.414 A611 8610 8.710 B910 19 8913 9,016 9.120 9.226 9.33 9.441 9,860 9.661 9.772 9.886 In practice, the conversion of ratios to dB and vica versa is made easier by means of a table ‘such as that shown in Fig. 1.12. Here ratios from 1 to 10 (9.886) are shown, Further multiples of 10 are made simply by adding 20 dB to the result. ‘As an example a ratio of 13,7 = 10 * 1,37 corresponds to 20 + 2,7 dB, or 22,7 dB (still ina voltage ratio). Similarly 33,2 dB, or 20 + 13,2 dB corresponds to a ratio of 10 * 4,571 = 45,71. Also by using a pocket calculator the conversion is easily made, 1.131.7 PEAK, AVERAGE & RMS VALUES There are several different ways of quantizing the level of vibration. The first is to use the peak (or maximum) value, as was done with the analogy of a turning wheel earlier (Fig. 8). This peak value is shown in Fig, 1.13, It is useful for simple harmonic vibration (such as the one shown), but for other types it is not so good because it depends only on an instantaneous vibration magnitude, and takes no account of the time history producing it, Displacement Fig. 1.13, Peak, Average and RMS values for a Sine Wave The peak to peak value, or the magnitude of the positive and negative extremes of the motion is also commonly used. For a symmetrical signal (usually the case) it is twice the peak value. Another quantity, which does take into account the time history is the average absolute value, which is defined as Ix| dt 1 x ei ‘laveragel Tt en Even though this quantity takes into account the time history over a period T, it has been found to be of limited practical interest, because it has no direct relationship to any useful physical quantity. A much more useful descriptive quantity, also taking the time history into account, is the RMS {root mean square) value, defined as 1.14\ fit Xams = Tg ita ‘The major reason for the importance of the RMS value is its direct relationship to the energy content of the vibrations, The relationship between these values can be expressed as 1 Xams = Fe Xlaveragel = Fy “peak Where Fy and F,, are called form factor and crest factor respectively, and give some idea of the waveshape of the vibrations being studied. For a sine wave, this will be © 1 Xams = Xlaveragel ~ 7% “peak 2/2 V2 * motion, F, “Wt = 1,11 (= 1dB) and F, = 4/2 = 1,414 (=3 dB) Most vibrations encountered will not be pure harmonic waveforms, and in general an RMS Measurement is to be preferred. 115