ORBIT Machinery MESsages Housing measurement transducers By Merk Gilstrap, Technical Ser In a previous columa (Orbir, June 1083), we discussed the considerations for shafl versus housing measurements on rotating machinery. We also evaluated wo salt measurement techniques, the pio imity probe and the shaft sider (Orbit, March 1983), ‘This column completes the series by evaluating the wo most fre quently used housing measurement tears ducers, the velocity, Seismopzobe"” and the aeeeleramieter. This discussion is restricted to standard, commonly available vibration measure ‘ment qansducers. We will consider only these transcucers that have a broad range Of applications on various types of rotating ‘machinery. Specifically, see mean velocity ttansducers of the magnet and coil spring. mass-damper type and compression-type piezoelectric accelerometers We will discuss those transducers that are typically employeil for singular over- all machinery evaluation in monitoring and preventive maintenance programs Granted, there are special-purpose ceaus~ ducers that may he excellent for detecting anid diagnosing contain specific mactinery 1cs Manager problems. Bul they may olfer am insu Ficiont level af overall protection for monitoring purposes unless the moniter. ing system is augmented by otlier cans ducer types Differences between velocity and acceleration measurements Any vibratory motion can be measured simmltaneously by hoth a velocity and acceleration rinse, butte sien = puts of each can be signitcunty different. Velocity Is the time rate of hang of cis- placement, Acceleration isthe tine rte of change of velocity. Velocity is a direct function of vibration frequency, while acceleration is a function of frequency seared. atiempt to charaeterize the forces acting tipon a machine, it srculd be ideal to prove thateither velocity oreceleration fe more representative of those forces and thustake 2 major step toward corset tcanscacer seloction. In Fact, both sides of this ous tion have ben debated for years, with no clear winner Velocity Transducers: Suitable for overall monitoring of cortain machine types. Accelerometers: Mostly suit- ‘cble for hich frequency mec surements with some overall monitoring applications in cortain situations. July 1984 Proponeets of velocity stale that “con siant force prexhices constant velocity.” Proponents pf acceleration elaim that New- {on was right and force equals mass times acceleration." Actually, vitvation is the esulant inter action of Aaemfbl forces operating om the ‘machine versus the sell (dlesignedio} dynamic sitlness forees in the system, In terms of the pormal rotorbeuting support system. dynantic stifness has duce components; systenn spring coef cient is related to displacement. darnping {0 veiocky, and mass to acceleration. ‘The- oretcally since spring eoetficiont, camp ing, and mass contribute (o the total pictore, all shrce engincering units of vibration measurement can be valid for certain applications In some situations, an segument ean be made for one vibration parameter (is: placenteat, velocity, orazcoeration) being superior to the others, if it ean be shown that the Forces aeting on the systom are more dependent upon one dynamic sill ness term than te others. For example, vibration below a resonance freeuoney: is generally dependent upon the spring Covffivient (machine acts Tike a pure spring). Forthese vibrations, displacement may be the mest meaningfal parumctee. ‘Vibrations atand above a resonance Ire- «quency are dependent upon damping and mass, respectively. In these cases. velocity and acceleration, respectively, may pro= vide more valttble data, From purely theoretical viewpoint, however, mostcom- mion machinery problems span the [re quency spectrum from below 10 above resonances in the system, Consequently, displacement, velocity and acceleration ean all be sig eer for ov sant, ancaningtul parann- all machine evaluation, Transducer Characteristics Signal-to-noise ratio All theory aside, and since the mechan ical theory doosn't clearly favor one mes surement technique over another, we must look for a more significant determinant Transducer sensitivity and signal-to-noise ratio are important considerations in any selection pracess Since velocity and avceleraiion sepre- sent two different measurements, vibra tion amplitudesexpressed in both units canJuly 1984 be significantly different for the same vibratory motion, A uansduces's sen- sitivity (Scale Facror) determines its signal output level. This, combined with che actual vibration motion amplitude, defines the signal-to-noise ratio. Forexamnle, asinasoidal vibration with a frequency of 100 Iz (6000 epm) and displacement amplitude af 1 mil (5 ‘icrons) peak-to-peak has an equi velocity amplitude of 0.3 ia /see. (8 mms) (0-to-peak and an acceleration amplitude of 0.5 Gs O-to-peak, Considering the seo sitivity of typical transiucers—S00 mV per in.fsee. (0 mV per mms) O-1o-peak Tir velocity and 100 m¥ per G O-te-peak for acceleration (also the American Petroleum Institute (API) 678 Standard) — ‘these amplitudes would yield signal ouiput Tevels of (50 m¥ and $0 m¥ Oto peak, respectively. Both of these amplitudes represent good signal levels, well above the noise floor, i.e., @ good signal-to-noise ratio, How- over, futher comparisons will show that as vibration frequency decreases. the signal- to-noise ratio for an acccleromet creases more rapidly than far a velocity ‘uansduces Asanexample, fora I mil peak-to-peak displacement at 50 Hz. (3000 epml, the velocity and acceleration amplitudes and transducer output levels would he 0.16 in.isee. (4 mueys) oF 80 mV O-t04 0.13Gsor 13 mY -to-peak, res Boh signal levels are still above the noise: floor, but the margin is less for the accelerometer. Finally, fora | mildisplacement ata low frequency of 10 Hy (600 cpm), the nun. bets would he 0.03 in./see. (0.8 au’s) or IS mV for velocity, and 0:005 Gs or 0.5 mY for acceleration. [m this case, che velocity output is still barely satisfactory, bul the acceleration output is clearly net This data is summarized in Table 1 This companison was based on a con- stant displacement amplitude, Table U1 summarizes the data using a constant velocity amplitude, and the signal-to- noise comparison would yield the same conclusions, Thus, using a standard accelerometer for low frequency measurements may be difficult, particularly if the accelerometer is simultaneously measuring vibrations at significantly higher frequencies. In this case, the low amplitude low frequency ORBIT components may go unnoticed in the rela ively higher overall si Insummary, transducer selection based bon sixnal-io-poise ratio requires. (1st an identification of the anticipated vibration frequencies based on the potential "machine problems ani then an approxitns tion of the vibration amplitudes. at those Frequencies. This discussion is based on sensitivities of standard Bently Nevada producis, Most competitive units. tw tbe ‘marketplace havea similarrelationship. In fact, only a few standard seismic trans ducers from other manufacturers have higher seusitivities than Bently Nevada seismic transducers, snd those ate mostly velocity pickups, not accelerometers, Frequency response in, wih reference to Berly Novas prodiciy te frequency response of one the sarward velocy scismapmaes f4S Miz to Kile. nd for an acceltometer $ Hz to 20 Kil, both at-3 UB. Thos from this data alone, i would seem that the TABLE ‘ureur Fnow ‘OUTPUT AGM INPUT VIBRATION velocity Pronu AGCELEAONETER FREQUENCY [DISPLACEMENT] AMPLITUDE | SIGNAL LEVEL*| AMPLITUDE | SIGNAL LEVEL" Gon AWPLITUDE || WCHSECOND | mvoPcak | ‘seoPcaK | nvoPcAK IWS PEARPEAK| "OPER Ban 00 Dat ie oa ET a0 00 16 30 oo 1s foe 409 3 15 ons os and 100 pert areleraton Comparison is “Signa ieesarshases on varednet senses of 00 mY zovich'secrel (OY oe rans] Oger ecoey Comparison of Velocity and Acceleration Levels 1.09 4 constant vibyation displacemen’ amplitude of 10 mils pos peak, English unis only are shown for simplicity, Melek conversions ate: (ils — 25.4 mictons 1 inehisecond = 25.4 mmisecond TABLE IL ‘oureur ERO UTPUT Fi INPUT MBRATION veo PRUE ASGELENOMEER FREQUENGY [DISPLACEMENT | AMPLITUDE | SIGNAL LEVEL"] AVPLITUDE [SIGNAL L=VE CPN ANBLITUDE. | HICHSECOND | mvopEsK | ‘eaPEae | mVOPEAK MWS PEAKDEAK| ODER eo, 06 020 77 os B ‘ac00 127 220 “00, 016. 6 <0 6a7 920 00 008) 3 and 100 porate acceleration “Signa ievesarabases on xanedcer senses ofS00 a geri seca (Oe pers] peak orweloey Comparison of Velecity and Acceleration Levels Comparison i based on 2 constant vibration velocity amplitude of (2 inchisecond (9.08 men’s) perhaps Eglsbunison sexton fvanplcs Ne cece 1 Dee" 381 icons 1D Insocond = 254 mmsccond TABLE IL [GER craRATERSTIGl ——verooi’ ——[-—sceceamrica —) ropes Unt ‘Exe Rpiei Seat aos ea tans Sigal ate Good Oe Resid Poo aw Freon fang Fas Raa ned SBT Tay sme RE = rere ea eae Frat Meevane aay Fate Sg or Coe tog em Fcomtcs Cares: | snc) eae tae oa Cees ea Seeeuue ms aa rhe |accelerometer is superior w the velocity pickup for all applications But the previous section ilustrates that this may notbe (rue for the lower end of the Hrequeney spectrum. IL miay be true only ‘sometinies inche middle Frequency region. ‘where apparently both velocity and weve! cecation cou! be used. Act rased in this frequency’ applications Although spec! hnigues can extend the range of some velocity transclucers to well above KH. er is the usual choiee for ‘applications above | KHz. Tn general, acceleration measurements, emphasize fhigher tequeney signal com ponenis, Displacement will emphasize the ower frequencie. Installation and reliability For a certain mildie frqoney range perhaps 20 030 Hew | KH, onc of the Trandacer charactersties discussed thos farhas revealed an obvious choice. Ibeth types have acceptale pertormanee in this Foeion, we must eviluate some otter Tnstallation requirements and trans diucer reliability bring out advent al disadvantages of both teumslucs Installation raquiremenis would be similar for both velocity and avcelor lion transducers, Both are usually nnoumted an abearing housing. The etach ment must be reasonably perpendicular dnd provide goon! fash contact withthe mounting surface. However, if the acceler Gomer is intended «0 meesuse high I jquencies. the method of tachment, 40 Particular, becomes mone era ground smeesh enough For tat surtave-t0 this ease, an interrnettate mounting block will be necessary (APL 78 requires this) For most applicutions, a combination atachment works best, A stad (with proper torque) plos a thin film athesive {epoxy moantine al produce the desired high Irequency response. In effect this means that most accelerometer installa tions should be considered permancat Usually, magnetic base and hand-held techniques provide litle more frequency range than with a velocity transducer Tn general, the mounting requirements for velocity Seismoprobes are less sin gentthan for acelerometers. although low ial construction tech ORBIT Frequency models have mounting, restrictions a6 to orientation angle. Reliability, on the other hand, scems 1 Tavor the accelerometer Since the velocity transducer is a spring-mass-damper mechanical sysiom, sonic degradation in performance under normal use can be expected ever a period! of time, Mos! manufactzers recommend 2 eal itration check sammually. for the velocity transducer, and warrant their specified per Tormanee tor no more than two years. War ranjies for accelerometers may not be any longer, but their recost of long-term relia bility in the field, ussumiag ne ebase, is heres On the uther hand, a velocity transducer cam ttke more abuse on a routine daily basis than am accelerornews An acceler fometer may be specified for high shock leads, but the rating usally isonly forthe sensitive measurement axis. A velocity transducer can usually withstand more abuse in the ecoss-axis direction, Environmental factors Both transducer types can be applied in similar indusivial shvironnients, ‘This is especially true if the installation is good, iLe., a surrouiding junetion box (not con- reeling the transducer) providing environ ‘mental as well us mechanical protection Temperature may be the only deciding, environmental factor Here agin, theres au compromise ‘Accelerameters (with external change amplifiers) can be designed 10 meet higher temperature environments, but suffer more from thermal shock and ambient tomperature gradients. A high temperature application generally means that the rnachine easing is extremely ot éue tothe internal gas oF liquid), not necessarily the ambient natebine environment Unacceptable signal noise may be pro- duced by a significant temperature ditTer- ‘ential from the top of an secelerameter %0 Us mounting surface, Also, thermal waves, radiating from the hot mounting surface may cance aclditionsal noise Cost Hardware cost is never an unimportant factor, but inthis evaluation, itis nota very decisive one, either, In general, scecl fomaters, particularly the high frequency and high temperature medels, ae more costly than veloeity transducers, But they also generally require less frequent July 1984 replacement, a leas! ftom the standpoint cf normal pertormunes degradation. And in most systems, transducer costs ate relatively low compared © the typical in- vestment in monitoring and analysis instrumentation: Signal conditioning Some oF the compromises snake in ce transducer selection process can be some: what tempered by certain types of signal conditioning circuitry, The eleetroniesean be ineither an intetice module oF mont tor'eacout instrument, Three signal con citioning techniques used often are signal integration, filiering, and amplitude! phase compensation, All ace available in certain monivor systems ancl are ustally incorporated in portable insteuments used for vibration analysis Signal integration An integration circuit is used to convert either a velocity signal o displacement or tn eceleration signal velocity. If fora given application, 2 particular vibration neering unit is desirable, but dhe ces cluver that measures direc in those writs ‘cannot be used (for whatever eeason), then signal integration isa viable alternative ‘Some users preter accelerameters over velocity pickups as a fandamental heusing measurement transiucer, hut place more value in velocity than acceleration engineering Units a8 an overall indisstor oF machinery condition, In this case, the accclecomiete” signal would be integrated to velociy Integration also is used when it s desi= ableto measure housing vibration displace nent, Since no commonly used tans: ducers measure seismic (absolute) dis- placement, a velocity pickup is normally employed with its signal integrated t dis, placement. Double integration ofan aeeel ration signal to displacement is also pos sible. But, because acceleration units caphasize the higher frequency compo- nents, very high gain is required to sive value to the lower frequencies. Oftentines this grin aruplities the noise level as much asthe real signal. producing a meaningless doable-intezeated signal Signal integration has definite value for machine analysis and monitoring pur- poses. Asexplined above, since displace- ‘nent, velocity, and acceleration each mea- sure different characteristies of the same vibratory motion, evaluating hat motionJuly 1984 in two, oF all sheee. of the units ean be more informative than wsing only one The dual probe is combination trans- ducer systees using « proximity probe for shatLrelative vibrationand a seismic irans- ducer for beating housing absolute vibra tion. In order to compare shaft relative ard housing ubsolute motion, the housing rmeasirement signal must he iniegeated 10 dlsplavement. Also, wo derermine shat ribsolute mation, the instantaneous time ‘summation of the shart relative and hous ing absolute displacement waveforms is required. If the scismic tranducer i a velocity seismoprobe, single integration is required: double integration is require if itisanaceelerometer In thisapplication, a velocity sensor is preferred, since single integration is less noise susceptible and isa amore reliable process than double integra- tion of an acceleration signal Often, fltsing othe acceleration signal prior to double intogration is requined to provide adequate sixnal quality. However, such additonal eicuity adds an urneces- sory degree of unrelibility for monitoring upplivetions. Inaddition, the dfieulies of using the accelerometer in the low fre queneey region will limit the use of the Complete dual prohe assembly (For shat absolute measuresents). Finally, ia some applications, housing measurements are used fr reasons of ira- dition, economy, andor installation diff- culiy on machine types that technically ‘would be best measured by shaft displace- ment transcucers. For these situations, housing displacement measurements derived trom integrated velocity signals ray be the next est alternative Filtering Sometimes optimum tiansducer per formance and usefulness is limited because of erronecs signal noise oc the apwuze of unwanted vibrations. TP the tele information is contained within « paricular frequency domain, then filters in the signal conditioning iruits may help. For anssante signal frequencies that ie entirely in a range above or below the range of interest, Iow-poxs and high past thers can be used, When erioncous sig- nals Hie within the desiced frequency range. the application of ban reject (notch) or combination low-pass! high- pass fillers cin be consicered Tn all cases, the exact filer charac- tecstos (bandwidth, Q. tate of roll of ORBIT ste.) must be known before they are used. When fillers sre used correctly. they may Imprave signal-to-noise ratio and over come some of the difficulties of signal integration, But when used without com plots knowledge of ta filter's performance and specific machine behavior, some meeningful information may be lost in the process: Amplitude/phase compensation For most velocity transducers. the spee tied lower frequency limit is that which the ample response is attemated by a 40.7% «own by 29-38%). At that point, there is also some phase tlag or eal) ero Belov this point, although the translacer's response is even ss tect. the signal aay all be weak if the exact seaports nown and repectable. A signal edition: ing eieaic dat is maxched” tothe ampli tule and phase response can be used to iminimize these errors in ths frequenc This circuitry can be incorporated in a snonitor system or may be provided in portable ead! insiniments. Tis essen tial for monitor systems on Low speed machines, such as cooling ower fans tis also useful for obtsining sceurate cata dur ing star-ap und shutdown, and For lov speed balancing sitvations Conclusions When choosing any monitoring system, portable or periodic, some basic questions iuist be answered in the Following osder: 1. What is the machine (ype and the 4esign of its components, and what are the process conditions? 2. What are the most Likely problems that could occur on this machine’? 3. How will the machine's vibration ccharacteristies change if each ofthe above problems occur’ 4. What systems of transducers and sig, fal conditioning citcuiiry will best mea sure the above vibration changes and mect the mechanical and environmental requirements of the installation? 5. Which system provides the hest cost Denefit ratio? Equally important to consider are the basic objectives of the menitoring sysiem, ‘What level of information is expected from it? Who in the plant will need this informa- tion te make decisions about che machi? Is the system simply w proteetthe machine binst very sovere damage or is ite pro- vide diagnostic level information 9 the plant's rotating machinery specialist? Any one of several systems might he adequate for providing some warning prior to total machine failure, But the selection, process must be more careful for a system intended to detect machine problems early in their occurrence and provide might into the mature of the problera as well The characteristics of velocity and acceloraticn transducers are summarized in Table IHL, For certain applications of Jrousing meususcmnent tanscucers, there is a clear choice, The velocity transclucer is generally more vsefitl for very low fre= quencies and the accelerometer alone is applicable tor very high frequencies. In the middle frequency doxnain, either type might be used with success for particular applications. Since the corsideration of all her factors does not reveal an obvious supeticrity of one over the other for some applications, choice hecomes as much one 6} personal preference andor specific ‘pplication as anything else. We welcome your comments on this subject. We would like 1 know if you agree with our observations, or if in your experience, you have gained some krowl- edge which provides a more decisive choice between velocity and aeceleration transducers for general or specific applications. Pleave write Mark Gilsteap in Minden, For more information check the follow fing L numbers on the return car ‘Acceleration Transducer System, ovat Low Frequency Scismoprabe Velocity Transducer, DISA. ‘Velocity Transducer System, L393, General Purpose Velocity Transducer, Loss. Machine Protection System for Various, Types of Rotating L:quipment, Par Two, L235; une 1983 Orbit, L8009. March 1983 Orbit, L8008, Ill