MEI Back-to-Basies The importance of transient data analysis by Richard Thomas Senior Engineer Machinery Diagnostic Services Bently Nevada Corporation Baton Rouge. Louisiana ransient data reveals much about I amachine's condition that steady state data cannot. New technolo- gles make this information much more available Jor evaluating the mechanical integrity of a machine irain. Often, though, machinery auchts donot use ansient daa. Recent suiveys indicate that throughout indlusery, simpliciy is preferred over complevencss in machi This arly shown by the number af rotating machinery predictive maintenance pro grams in which only static, steaely state vibration amplituile is trencled, while important cransiene vibration data fs not even acquired 2 audits, Consider the analogy of pilot, He must be aware af + Instantaneous speed and altitude amplitude), *# Lostantancous direction (phase), ® airplane mated time of arrival (ample tude trend © Heading (phase trend) © Vertical speed, ascending or descending (transtent operatic) Wdte piloc only “wended” his speee, and ignored his direction and his rate of ascentor descent, disaster woukl surely follow, However, in the past decad predictive maintenance programs Increasinglyemphasize "speed of Might analysis, while de-cmphasizing oiler, equally important data, ‘This article will explain what cransient lata is, why its important, andl the types of plots thatare used for its display. 16 onit ‘What transient data is and whyit’s important ransient data is amplitude, phase, frequency, position ancl process data Transient data is acquired from a machine during startup or shutdown, or luring a change in speed or load, Vibration cata cen be convovedd te avo formes: Static and dynamic, Static dias is diserete values thai describe the vibration siqnat For tusiance, a vias tion signal might be described oy the staiie vabies of amplitude, phase lag angle and jrequency, Dynamic data is Whe cactuad eibraion signal Asa rotating machine changes speed the forces and stilftesses tbat aci upon it mange, Although many clements con tribute to these forces and stiffnesses, a Machine's vibration response is simply 270 Force divided by stiffness. A machine's vibration response, as it changes speed, tells Us much about the nature of the Forces and stffoess acting vn iL This, perspective cannot he supplied by steady state dara nt data cay icenuty: * Slow roll spect ‘The maximum speed at which ao dynamic motion can he measured. Shaft bow and runout can be measured slow roll speeds. fow volt rector: The slow roll vec- tor is data unrelated wo a machine's dynamic motion. It can mask dynamic motion data, so we sure il for use in creating eompen- sated plots, + Mode shapes: Laweral svode shape information is valuable for balane 4.0 mil pp Full Scale 180 X to ¥ (CCW) Rotation Figure 1 Uncompensated 1X polar plat September 1995ing and for idendifying faults, such ws shafi cracks, bearing failures and rotor to stator nubs © Heavy spot locations: Essential information for balancing. © Balance resonance spocds: Shak sotative speeds that equal lateral natural frequencies of the rotor § tem. Ator near these speeds, vibea- tion amplitudes are highest. To. axord damage, & machine should tot be operated at or near reso nance spee * Synchronous amplificaiion factor A measure of the 1o10r's suscep. Uubility to vibration when rotational speed is equal 19 a rotor lateral mat ural frequency * Synchronous quadrature dynamic siffness. Inversely velated 10 the synchronous amplification factor, itis an indicator of rotor sys- tem damping. This damping acts a “dam” against forward circular whip type malfunctions. * Load: 4 vnidivectional steady state force acting on the rotor system Misalignment caused by a loud is one of the largest contributors t0 machine fiilure and redveed lifespan. 4.0 mil pp Full Scale © Rubs: Concact beoween a machine's rotating and stationary parts © Instabilities: The most common are the uid induces! instabilities, whirl ancl whip. Instabilities can cause destructive levels of vibration Shaft cracks: & evacked shaft can use the most catastrophic failures, ‘Transient data ean be acquired from both machine startups ane shurdowns, Dawe acquired from startups and shut downs may differ slightly because of dif ferences in driving torque and in the thermal and alignment states of the machine, Transient data formats ‘The most common presentation plots used for transient data analysis are: Polar + Bode * Shait average centerline © Onbit © Spectrum cascade Polar and Bode plots Polar and Bode plots both show the change in filtered vibration vector with changes in shaft rotational speed. On a Amplitude and Dirootion of Y Response at Balance Resonance +34 point X to ¥ (CCW) Rotation Figure ‘Compensated 1X polar plot September 1995. polar plot, the evo vector components, phase and ampliude, are plowed on the same graph, in polar format. On a Bode plot, phase and amplitude are plotted on separare XY (rectangular) graphs. Bach plot fonmac makes certain vibration characteristics easier co identify. ‘They complement each other; therelore, you should always view polar and Bode plots together: Apolar plo (Figures Vand 2) isapolae format presentation of the locus oF the shale 1X (or 25.) llered Wbration vec tor from a single channel asa function of shalt rotative speed, usually acquired during machine startup or coastdown (wansicnt operation). A Bode plot (Figures 4 snd 4) is pair of graphs in XY format displaying the 1X (or 2X, 3X. ) vibration vector from a single channel asa function of shaft rota tive speed, The ¥ axis of the top graph represcnis phase kg angle, while the ¥ axis of the bottom graph represents amplitude, The common X axis repre sents shall rotative speed. A 1X Bode plot ts sometimes called an unbakance response plot Usually, we plot 1X data in these for: nats, altliought we carvalso plot the har- monies of running speed @X, 3X...) Higher-order transient data plots, not discussed in duis article, are valid anc useful wos. Polarand Bocle plots identify the slow roll runout vector, the slow roll speed! Jimit, the balance resonance speeds, the synchronous amplification factor, the synchronous quadeawire dynamic sti ness and the location of heavy spots, Polar and Bode plots are essential for identifying changes in resonance fr quency due to elunetion mechanisens, for instance, rubs and cracked satis The slow roll vector and slow rott speed range The stow roll vector is lata unrelated to a machine's dynamic motion, 1 is, caused by rotor bow, mechanical cunout and clectrical runout, We compensate polar and Bode plots for slow roll runout because ir can mask data Uiat shews the machine’s dynamic motion. Orbit 17Most of the polar and Bode plots we use in machinery diagnostics have been compensated for slow roll runout, Ust= ally, our only use of uncompensated plows is to identify the slow eoll vector and the slove roll speed range limi “the slow rall speed limit is the point, at very low rotational speeds, where vibration amplitucke or phase begins to change. Above that speed, the machine hegins to show shalt dynamic motion, Therefore, slow roll data should be acquired below this speed. On an uncompensated Bode plot, slow roll vecior components are read Irom the vertical axes on the phase and amplitude graphs, The values of the lefl- most, fattest portion of each curve are the slow roll vector’s phase and ampli- tude components. In the example shown in Figure 3, the 1% slow rell vec tor is 1.0 mils pp at 225 degrees phi lag. (Onan uncompensatedl Bode plot, the slow roll speed limit is determined by noting the speed where either vibration phase or amplitude begins to change Draw a vertical line from that point te the horizontal speed axis. ‘That speed is the highest speed at which slow roll data can be acquired. In the example shown, in Figure 4, slow roll data must be acquuired cow 1000 rpm, On an uncompensated polar plot, sead the slow roll veetor by drawing & Jine from the origin to the lowest speed sample on the graph, ‘The vector’s amplitude is the line's lengih measured on the polar plot's concentric amplitude scale, The vecten’s phase lag component ‘s the point on the polar plot's phase scale, measured against the disection of shalt rotation, where that line incersects sf extended. In the example shown in Figure 1, die 1X slow 10ll vector is L0 mils pp st 225 degrees phase lag, On an uncompensated polar plot, ehe slow ral speed range is more dificult 0 ‘dentfy, because ithas rospeedscale or ais, We prefer to use the Boeke plot for this measurement. Balance resonance speeds A balance resonance speed is « shaft rolative speed, oF speed region, which equals « natural frequency of the ro10r 18 obit system. As shaft rotative speed approaches a rotor system natural fee quency, vibration amplitude increases, and vibration phase begins to lag (move against the direction of rotation) the unbalance force that causes it. At the resonance peak, the amplitude of the vibration ig maxirmusn ancits phase lag is approximately 99°. As rotative speed increases past the resonance peak, the phase lag increases until it reaches, approximately 180° acspeeus above the resonance. At resonance, the stiffnesses that restrain shaft motion are at a mini- mum (as explained later for the syn- chronous amplification factor), so vibration amplitude is ata maximumand phase log is changing most quickly. ‘Therefore, 2 machine should never be ‘operated at or neara balance cesonance, We use compensated polar and Bode plots to identify cesonance speeds. On oth plots, halance resonance speeds are identified by the characteristics that define them: a peak in amplituckeaccom panied! by a 90 phase lag. On a Bode plot, as the amplitude plot's graph line Iises to its peak, the phase plot's graph line fall by 80° Gn a polar plot, the amplitude and accompanying phase shift appear as a Circle The amplitude peak is the point oon this ciscle farthest from the polar plot's origin. tis 90° phase lag is shown by the right angle between a ¥ector at resonance and a yector at a speed well below resonance. First, deaw a line from the origin through a point on the response circle farthest from the origin Next, draw a line feom the origin through a point on the circle ata speed well below resonance (very near the ori gin), The angle between the two lines ts approximately 90°. The synchronous amplification factor and the synchronous quad- rature dynamic stiffness ‘Two other, related machinery param: eters can be analyzed from polar and Bode plots. One is the synchronous amplification factor, or Qs. Qe is useful for estimating the vibration amplitudes that occur at rotor system natural fre quencies duc 19 the unbatance force Inversely related to itisthe synchronous, quadrature dynamic stiffness, System dynamic stiffness limits 2 rotating machine's vibration response, because vibration response Is equal «0 input force(3) divided by dynamic stiff ness, Dynamic stiffness is composed of ovo terms, a direct term and a quadra- ture term. The direct term produces motion in line with the force andl con Phase lag, degrees Uncompensated 1X Bode plot September 1995tzins a spring stiffness anda mass stiff ness component. The quadrature term produces motion at 90° to the force (one geiarter of 360°) and contains the damping stiffness terms. Ata mechanical resonance, the spring stiffness and the inertial stiffness of the direct term are equal in magnitude but opposite in direction; they cancel each other. At a ‘mechanical resonance, the only restraint isthe quadnuturedtynamic stifness term, ‘Thos, the synchronous amplification fac- or, Qs, Is 2 measure Of the amount of synchronous quadrature dyoamic sil ness chat is present in the rotor bearing system at a mechanical resonance. A high Qs indicates a low synchronous quadrature dynamic stiffness, and a low Quindlcates 3 high synchronous quadra- ture dynamic stifiness, Although several criteria can he applied to evalume the significance of Qs, the synchronous amplification factor should be less than § for a welkdesignes! system. Qs values between 5 and 8 indi- cate thar the cotor system Tias marginal quadrature stifness at eesorance. Ifthe rotor system is not symmetric, then this ‘measurement cannot be used until the resonances ace properly separated, For split resonance separations, refer 10 BRDRC papers. Q; values in excess oF 8 indlicate that the system has low quada- ture stiffness at resonance and is suscep. Phase lag, degree: tible to large amplitude excursions at resonance, High cuacrature stiftness i lesired because it imits vibration ampli tude at resonance Qs may be calculated in several ways. One is the Half-Power Bandwidth method (See fal Power Bandwidth method for the evaluation of syn- chronous and nonsynchronous quad- ature suffnesses in the June, 1994 issue ofthe Orbit). To calculate Qgby the Hal Power Bandwidth method, divide the frequency at the resonance peak by the resonance bandwideh, ‘Ihe resonance bandwidth isthe differencein frequency between the points on either sice oF the resonance peak that are SdB below (or 0.707 times) the peak amplitude Another methoxt is to ealeulate the ratio of vibration amplitude at resonance to vibration amplicude at high rotauve speed (hut below the nest higher mode range). The following will explain how the values required for the Halt-Power Banclwicth method are obtained from both the Bode and polar plots ‘On the Bode plot (Figure 4), the reso- nance frequency occurs at 90 degrees phase lag, which is near the amplitude peak of the angle of the high spot from the angular location of the heavy spor Yo calculate the resonance bandwith, multiply the vibration amplitude of the resonance peak by 0.707. Draw a hor mn pp Figure 4 Compensated 1X Bode plot zontal line on the Bode plot's amplitude graph at that point. The difference between the speeds at the two points the line intersectsis the resonance band width. In Figure 4, the peak amplitude is, 3.5 mils, and oceurs at 2415 cpm, 3.5 mils, 0.707 = 2.5 mils pp. The two points on the graph with these vibration levels are at approximately 2330 and 2530 rpm, The difference between them (the band- ‘With is 200 1pm. Inthis example, Q, = 2415 #200 = 12. ‘On the compensated polar plot (Fige se 2), the resonance frequency is the point on the plot circle farthest from the origin. To calculate the resonance band- width, multiply the peak amplinude by 0.707. This value will correspond 10 0wo points on the ploc; cach has a speed associated with it. The difference Ddetween those speeds is che resonance bandwidth. Inthe example shown in Fig, lite 2, the peak amplitude is 4.5 mils pp, ancl occurs at 2415 cpm. 3.5 mils pp X 0.207 = 2.5 mils pp. The two points on the graph with vibration levels of25 mils pp are at 2330 and 2530 rpm. The differ- ence between them (the banchwidth) is 200 rpm. In this example, Qs = 2415 / 200= 12, A Qa value calculated froma Bode plot should ageee with that calculated from a polar plot, Small differences are normal, and are due to accuracy in determining values. However, a Qs caleulated from startup data and 2 Q, calculated from shutdown data mey ciffer due 10 ther mal, load and ramp rate differences, as well as for other reasons, Caution should be used when apply: ing Qg calculated by elther the Halt Power Bandwidth method or the ampli tude ratio method, because of potential ‘mechanical nonlinearities end abnor malities which can affect machinery vibration response. Further, the syn- chronous amplification factor does cot define or deseribe the properties that determine rotor stability. ‘The means of determiving operational stabily margin is given in references #23 8 24 Heavy spot location A rotor that is not perfectly balanced will hayc a mass unbalance. The location of the mass unbalance, or heavy spot, is essential information for use in rotor balancing, Orbit 19 September 1995.©n a compensated Bode plot, che value of the lefimost, flat portion of the phase cure ts the angle of the heavy spot location with tespect to the trans- lucer used to generate the plot. In the ‘example shown in Figure 4, the angle of te heavy spot is located at 350 degrees phaselag, swith respect to the transducer used to generate the plot. On a compensated pokir plot, the angle of the heavy spot location isideati- fied by drawing a line from the origin through the lowest speed sample on the graph, to the polar plot's phase scale, In the example shown in Figure 2, the angle of the heavy spot is located at 350 degrees phase lag, with respect t0 the transducer used to generate the plot Be aware that if che probe that sup: plied data for the Bode or palar plot is Jocated near a fui-tubricated bearing or seal, the heavy spot location may not he correct - see Precautions om Polar-plot Balancing in the March, 1995 issue of the Oxbit. The shaft average centerline plot ‘The shaflaverage centerline plot ¢Fig- uuce 5) depicts the average position about which the shaft vibrates within its bearing clearance, in XY coordinates. Both transient (versus speed) and trenel (@ersus time) data can be plotted in this format. The X and ¥ axes Show the aver- age change inthe shales borizoncal ancl vvertival positions with respect co some initial ar reference position, A typical shaft average centerline plot's Kand Y reference points are at the bottom center of the plot, This is because, on a typical, horizoncally- mounted machine with the toc mass beoween its bearings, we assume: that dhe shafi rests i the DouoM ofits bear ing at zero speed, Of course, fa better knowledge of shaft static position is kaown, that should be the stacting point. The shaft average centerline plot begins at either zero or slow roll speed. ‘As the machine 1s beouaht up to speed, the plotshowsshow the shaft rises.0a the bearing's oil wedge. The shaft average centerline plot alse. shows changes in the shafi’s pasttion with changes in load, ‘The shaft average centerline plot is Useful for idenufying changes in load 20 omit and bearing wear, as well as For calculac- ing the average eccentticity ratioand the rotor pesition angle. Averge eccentricity ratio is rela tive measure of the shaft’s position Depween the center of the bearing ancl thebearing wall. tis calculated by divide ing the average position of che shafecen- teriine by the bearing (or seal) radial Glearance, A sha with a zero eccen- tsicty ratio is concenuric with the bear ing or seal, while shaft with an Ccccentriciy rato of 1s in contact with the bearing or seal. Adeercasing ccccn- tricity ratio ean inclicate a potenval sta Dilty problem. ‘The rotor position angle is the angle between an arbitrary reference through the center of a bearing (asually vertical down) and a line connecting the bearing and shat centers, measured in the direc: tion of rotation, It can indicate the pres- cace of abnormal loads acting on a machine ‘A usual shafi average centerline posi- tion of a horizontal machine has an eccentricity ratio greater than 06, and a rotor position angle between 20° andl 50°. A.smaller eccentscity ratio indicates that the shaft centerline is approaching the bearing centerline, which usualy Jeads to a deerease in dynamic stiflness and an increase in the bearing’s fluid average citcumterential velocity ratio, 1 may Because the system dynamic stiffness redluced, the cotor has a tendency t9 be easily excited by the dynamic and static forces which act on it. Conversely, an eccentricity ratio greater than 0.6 ined= cates that the shat is approaching the bearing surface. where destabilizing forces are minimized. The orbit plot ‘An orbit plot (Figure 6) shows the dynamic, two dimensional path of the centerline motion of machine compo- nent observed by XY teansducers, i the plane of those transducers. When the transducers are XY shalt proximity probes, the orbit is the shaft centerline lateral vibration, called precession. An orbit can also be displayed on an oscillo- seope in the X versus ¥ mode. An orbit pio clearly shows: * the maximum vibration amplitude the direction of maximum amiplitucke * the influence ofan asymmetric sup por stiffness + the presence ofany load forces * frequency of vibration versus rota- tive speed! and direction of preces- sion (when a Keyphasor® signal is present) ‘The orbit is a plot of the shaft cen- cerline’s path as the shaft vibrates; there- fore, many vibration characteristics and Figure 5 ‘Shaft average centerline plot September 1995Imalfunctionsare casy to denny using it ‘The maximum vibration amplitude, arid its direction, are measured on the long. cestaxis of the orbit. Asymmetric supnort stiffness autcl unidirectional Iyads are densified by the shape ofthe orbit. Ifthe orbit is unusually far, oreven “pinched im one direction (Figure 6), ts indica tive of one of these conditions. Multiple orbit plots are often pre: sented alone or in conjunction with other transient data plots to. facilitate data interpretation (Figure 7) The spectrum cascade plot Asspeccrum cascacle plot (Figure 7) is @ graph im XY format displaying. ire- qqueney spectra versus shaft rorauive speccs, Shalt rotative speed anc vibra tion amplitucie are usually presented! on two Separate vertical (Y) axes. Fre- queney is displayed on the horizontal () axis, This data formatis used to eval urate the change in vibration amplitucle and frequency charactenstics during machine transient conditions. The orbit plots shown in Figure ? show the forms of| the vibration at various speeils. These obit plots can be correlated to the incl vidual spectrum plots that comprise the cascade pl Rently Nevada's diagnostic systems also generate an enhanced spectrum cascade plot, the full spectrum cascade plot. It displays the data from owe probes, rather chan one, anelshiows neg- ie frequency compenents useful for identifying the direction of precession of individual vibration components, Systems for transient dala capture and analysis Benily Nevacla miakes owo excellent systems for iansient data capture and analysis, Transient Daca Manager ®2 and ADRE® for Windows, The Transient Data Manager 2 (eDM2) System automatically collects and processes data during both tran- sientand steadystate operatinn,and dis. plays iin several plot formats A TDM2 system consists of a Monitor: ing Rack, a Communications Processor, and a host computer that uns TDM2 Software. The Monitoring Rack is Bently Nevada 3300 continuous mon- itoring rack, which has buitin dynamic and static dats ports from which the Communications Processor acquires vibration data, ‘The Communications Processor is a Bently Nevada Transient Data Manager or Transient Data tater face External, dedicated computers that alcquire ancl terporariy store dite and communicate with the host TDM2 com puter. The host computer isa PC G860r beter) running Transient Data Manager 2Sofiare Figure 6 Compensated 1X orbit plot September 1995 “TDM2 Systems are usually permanent installations. A single TDM2 hast com- puter can track 12 monitoring racks, each with a maximum of 24 points TDM2 computers can be networked, ‘with other TDM2 computers, with plant and OCS computers, andl with Bently Nevada's Engineer Assist™ expert sys- tem, With networking, a TIM2 System ccan track up 10 120 Monitoring Racks of up to24 pointseach. TDM2is another of Bently Nevada's systems that “move data not people Like the TDM2 System, the ADRE (Automated Diagnostics for Rotating Equipment) for Windews System col- lects and processes data daring both transient and steady state machine oper- ation, However, ADRE for Winklows is a porcable system An ADRE for Windows System con sists ofa 208 DAIU,a computerinterfsce card, and a host computer running ARE for Windows Software. The 208 DAIUis a small, portable dedicated cor puter which collects, stores and pro- cesses up (08 channels of cata. TWo 208 DAIL can be conneeted toa ADRE for Winclows computer for 16 channels of simultaneous data collection. The inter- face care! conacets the 208 DAIU 1© the host computer. The host computer is a PC (486 oF better): it can be a desktop computer, but 4 notebook computer is more appropriate for this portable system ADRE for Windows runs uncler Micro soft Windows, which gives it enormous, power and flexibility, You can view data from several transclucers, each in its own window, view data from one transducer in several different plot formats, each in is own window, or both, Efficiency 1s importane in any data analysis, transicnt or stcady state. A plant’s vibration specialise is most pro- ductive when he at his desk analyeing data, not walking a rouce collecting it. All Bently Nevada online diagnostic systems acquire vibration data automatically, freeing machine spectalists from that repetitive task, The data is also available on another computer via a modem and phone line, soin most cases machinery specialist need noc be onsite to analyze a machine. All Bently Nevada peersa Orbit 21nently-installed diagnostic systems, inchuding Transient Data Manager® 2, support network communications. In a network, machine data is always avail able t operators and managers. They can use it to make better-infocmed oper- ating decisions that extend machine availability andl bfe. Bently Nevada sys tems communicate with each other and ‘with plant control and DCS computers. For more information on transient data analysis, see the references for this article. For more information on Bently Nevada's systems for transient data cap- cure and analysis, systems with remote access and systems that network with your plant’s process or DCS system, con- act your nearest Bently Nevada Sales and Service representative I References: ‘Transent daiaaralssis abroad subject, and space permite ste expla only sores sppiiations Rertly New has published many ‘icles weltedtosinsene dara analy, in Booty Rotor Dynamics Research Comoro papers, it Benily Nevada Corporation Apgcatin Notes, anxinGxbit magazine ace reprinisSomeot ‘he papers ar listed here. They are avaible to you contzet the Orbit ecitr, at theaudess ele ‘phone ot fesnunber listed in dere of hs magne 1 Soseynska, A, “Intredsion to Balancing, ‘ently Kotor Dynamics Research Corporation, Report No. 187, Minden, Nevada, 1987, pp. 146 2ileatly, DE., Hated, C1, “Precautions on or por balancing" OF, Volume 14, No.1, March 1993, 18141 Bently, DIE, "Nearing aia sufess versus cceentcity." Orbit, Volume 9, No. 1, Septeer ber 19, rs120, 4 Southaiek, D, “Using Full Spestiam Phos,” Osbit, Volume 14, No. i, December 1993, 158) 5 Southaic, D., “Using, Fall Spectrum Ps {oatt2),” Orbit, Volume 15, No.2, June 1995, 15im 6 "Transtnt eta tnterface External" Ori, ol une 4, No Mare 1998 7 Bead, DIE, Maseynska, A. “Role of Creu ferent Flowin the Stbiey of Hcl Mand Machine Koors,” The hh Worshop on Rotardinamics Tassahiley Trablems in High Performance Turhomachinery, Teas ARM Unyersty, College Staion, Teras, NASA CP 3026, May 1988, pp. 415-430. S Musnyrsks, A, “The Role of low-telited Tame [gentil Forees in HotorleatingSeal Sistem Slit,” The Thi faecnational Symposia fon Transpor: Fhevomend anid Dyan of Howting Machinery (SROMAC3), Honolulu Hawai Ap 1990 9 tusayrska, A, Gran, -¥“Sablny and inst Fal a Tvo-dtede Boror Supposed by Two Tluic-Lubeicited Beatings," Benily Rotor Dynamies Research Corporation Report No. 1WS0, Trane of tho ASME Joumal of Wstion aiid Acoustics. 113, No.3, pp. 16224. The STLE Annual Neoting, Markee, Canada, 29 Spril-2 ay 2991 ap " Frequency: 1 Hertz/dlv Figure 7 Spectrum cascade plot with “plus” orbit plots 22 orbit 10 Nusearaka, A, Bendy, BE, “Fuic-geverated Instahiites of owoss," ODL, Volume LO, Xo. 1 Apsil 186, 18172, 11 Bently, DE, Muszynsha, A, Life Extension of Rotating Machinery with Vibration Monitor ing" Proceedings of FPEI Rotating Machinery Dytamics, Bearings, andl Seas Symposium, St Luis, Missout, September 1986, 32 Lame, WiC, Suse Ay “Pecidic and Con. tinuous Viiraion Mooring for Preventive Predictive Maintenance of Rotating Nachiry ‘ony Tans. ASME, Journal oF Engineering for ‘Gas Tatbines and Power ¥. 109, pi 1987, pp. 159.182. 13 Muszinaka, A., “Wisraional Diagnostics of Rolatng Machinery Malfunctions,” the course fa "Rotor Dynamics znd Vibration co Tucho- fnachinery,” Yen Ragman taste for Fluid namics, Belgaum, Seprember 1992. Submit. ted to the Letermtional journal of Rotting Machinery, 1995, 14 Tortl, DE, “Polr esting Aplications for Rottsg Machine." aroconted at Vibrzions Insets, Machinery vibrations 1¥ Sema, Cherry itil, New Jersey. Novernber 1980, aod 15 Bently, DE, "Shak Vibration Measurement and Analysis Techaiques,” Benily Nevada Reprint Astle, Apo 1962, LOG 16 Bently, DE, Nuszynska, A. “Detection of Rotor Gracle” Hrcceeding: af Tease ASM Uni verity Sih Turbemachinery Syinposiue snd Short Courses, Cexpus Chris, Tesas, Novem ‘ber 1985, pp. 129-199 17 Laws, 6. Tayon a, "bration aralysseet niques use to devet a rotor erarkoonaboler feedwates pump,” Orbit, Volume 7, No. 3 ‘Gerobe 1986, L817, 18 Nasznska, A, "Rovor-t0-Stalionary Bement Ruled Vibration Phenomena in Rowing ‘Machinery. Literature Sorvey, The Saock and beatin Digest, v 21, Mo. 3, Maech (588, pps 19 Nuszinska, A., Franklin, WD, Hayashi, RD, "Retoro Stator Paral Rubbing ard It ects Rotor Dynaenic Response" The Sit ‘Workshop on Foiordmamic instability Prob- lems in Iligh Lexfoenance Turbomachinery, CClege station, Texss, NASA CP 5122, ay 1960, pp. 453302 20 Schuilbess, 5s, "Diagnostic techniques using ADAE® 3 fr evaluation of radial rs in rota Jngmachinery Crit, Mole 12, No. 3, Octo. ber 1991 18136, 21 “Shalt Centesne radial position, sumo 10, Nod, Apel (989, LBL28. 22 Jorian, Na, "The mporance of moni ‘Shaft Cemesne das,” Ort Volume 13,No.2, Nay 1992 18152. 23 Nusnska, 4. "Frequency Swept Rowing Input Pension Terbnigues and Klenifise tion of the Tluid Foose Medele in Rotor? Bearing/Seat systems and Fuld Mending Machines,” Journal of Scund snl vibration, Volume 12, No.1, 2990, pp 108-124 24 Nuszyrska, Ay "Mod testing oF Rotors with Fluid Interaction ” International Journal of ovating Machinery, Volume 1.No.2,1995, pp, 3-116. 26 ently, Donal “How w handle (a seme cases f) forward zd reverse o1bis,” Oxbt, ‘lume 19,No.2, Jane 1995, A088, thie Yo! September 1995