Vibration Monitoring of Turbomachinery by Alex Lifson: Section Engineer Chandler Evans West Hartford, Connecticut

View metadata, citation and similar papers at core.ac.
uk brought to you by CORE

provided by Texas A&M University
VIBRATION MONITORING OF TURBOMACHINERY

by
Alex Lifson
Section Engineer
Chandler Evans
West Hartford, Connecticut
and
Harold R. Simmons
Principal Engineer
Southwest Research Institute
San Antonio, Texas
ABSTRACT
Alexander Lifson is currently The overall concept of vibration monitoring is presented
employed by Chandler Ev ans Corpora based upon theory and the practical constraints of machinery
tion as a Section E ngineer. In his present construction , transducer characteristics and the cost/benefit
position, he is responsible for directing ratio of available systems. General features of velocity, accelera
analytical research activities in the area tion, and displacement measurements are considered. M achine
of aircraft fuel pump performance. malfunction characteristics , mechanical impedance ratio , and
Prior to joining Chandler Evans, he force versus restraints consideration are assessed \\ith respect
worked for Southwest Research Institute to selecting vibration transducers for measuring housing vibra
in the areas of acoustics, rotordynamics, tion, relative shaft vibration , and absolute shaft vibration . The
compressor flow simulation, fluid me construction , application , limitation s , and comparison of avail
chanics, and vibration/stress analysis of able vibration transducers is discussed.
machine parts and piping components. Typical vibration monitoring protection systems for various
B efore joining Southwest Research Institute, Mr. Lifson was types of rotating machinery (turbines , compressors , pump s ,
employed by Exxon Research and E ngineering Company, where fans , electrical motors and gears) will be discussed . Data acquis
he was responsible for performing rotordynamic studies, for ition, data management and analysis , cost, and general capabil
simulating compressor operation, and for t he acceptance of ities of periodic and computerized on-line vibration monitoring
mechanical and performance tests of compressor trains. systems are analyzed . A presentation of vibration data in the
Mr. Lifson received his B . S. degrees in Mechanical Engineer form of overall vibration, vibration frequency spectrum , bode,
ing and Mathematics from Leningrad Polytechnic Institute waterfall and trend plots is addressed.
( 1978), and his M. S. degree in Mechanical E ngineering from Description and identification of typical sources of vibration
MIT (l981) . such as unbalance , misalignment, rubs , resonance , subsyn
chronous instabilities , and electrical problems is provided . De
tection of blade related problems through the use of advanced
Harold Simmons is a Principal En techniques for defining vibration related machine malfunctions
gineer in the Mechanical and Fluids E n such as modal testing, strain gage , and radio telemetry are
gineering Division of Southwest Research briefly discussed.
Institute where he is responsible for man Vibration severity limits for measurements on both machine
aging machinery v ibration investigations housings and shaft relative to the housing are presented; appro
for power utilities and other industrial priate correction factors are introduced to accommodate differ
clients. Mr. Simmons is also responsible ent machine designs , installations , and vibration sources . The
for developing innovative computer use of these limits , and examples of vibration monitoring succes
aided approaches for machinery diagnos ses , is supported by reviewing the actual field case histories. The
tics such as balancing, modal analysis, significant concepts presented are supported by the actual case
and turbine blade v ibration detection. histories data.
Prior to joining Southwest Research Institute, Mr. Simmons
was employed by Pratt & Whitney Aircraft where he was respon
INTRODUCTION
sible for the design and dynamic analysis of jet engine compo
nents for the SR 71, F14/F15 and other high performance air M odern monitoring systems require three major aspects in
craft. order to effectively protect the machinery from damaging
Mr. Simmons received a B . S. M. E . degree in 1963 from the dynamic actions . The system must bring together in an or
University of Florida and is registered as a Professional E n ganized manner an accurate means for measuring the critical
gineer in the States of Texas and Florida. motions and forces within the machine indicative of its health;
an effective means for collecting, analyzing, condensing, and or
ganizing the signals acquired; and a realistic means for interpret
ing the collected, analyzed data.
M easuring the motions within turbomachinery requires a
considerable extension of our natural s enses . For example, if we
101
102 PROCEEDINGS OF THE NINETEENTH TURBOMACHINERY SYMPOSIUM
only have X-Ray vision , we could see the rotor elements spin
ning inside as illustrated in figure l. \Ve could not, however, see
the very minute motions that are sufficient to cause damage; mo
tions which must be contained within the thickness of one page
(about 4 mils) of this proceedings for most machinery!
Oldtimers often tell of balancing an "Indian Head ='Hekel" on
edge to indicate a good running machine in years past. Although
this remains a cute trick for demonstration purposes , science
and technology have provided us with much better tools for
properly measuring machinery vibration . Accelerometers and
velocity transducers are now relatively inexpensive measure
ment tools for accurately sensing vibration of bearing housings
and cases . Proximity probes or shaft riding velocity sensors have
been readily available for 20 or more years to measure journal
motions that critically affect the life of hydrodynamic bearings .
The journal excursions shown in Figure 1 are examples of the
orbital motions that can be accurately sensed and recorded using
a pair of proximity probe s ; in this cas e , gear box pinion orbits
during a surge event of the compressor train shown in Figure 2 . Figure 2. Rotating Elements Inside a Typical Turbine Driven
The pinion, spinning at about 6000 rpm , is acted upon by Compressor Train.
dynamic forces that are created when the surge momentarily un
loads and then suddenly reloads the gear mesh; whirling out properly process vibration signals to assure adequate precision,
ward the pinion journal momentarily touches the bearing. For fi·eedom from errors , and rapid analyses .
tunately, surge events are rarities in well controlled compressors Interpreting the data is almost always a matter of comparison
and other forces predominate to stabilize journal rub . of recent results and trends with historical data bases collected
internally, with published criteria, or with expert opinion . Data
Note: Journal
Impact• Bearina bases developed , internally, on ones own machinery provides
the real benefit of including the impact of plant operating and
maintenance procedures and can be interpreted with plant pro
duction requirements in mind .
GENERAL VIBRATION CONCEPTS

It has become common practice to install vibration monitor
ing equipment on critical and costly industrial machines . The
underlying objective , of course , is to detect an increase in vibra
tion level when mechanical problems oceur. Therefore , when
selecting a vibration monitoring system , one of the first ques
tions which mnst be answered is : "\Vhat vibration should be
measured?''
• machine housing vibration,
• shaft vibration relative to hearing cap , or
• absolute shaft vibration

Each of these measurements is currently being used on a variety
of industrial machines and each can claim success in instances
where serious mechanical failure was averted by a warning of ex
cessive vibration . To resolve this seemingly simple question,
both the likelv malfunctions of the machine and its mechanical
impedance s hould be considered.
Figure 1. Pinion Lateral Response During Compressor Surge.
Machine Malfunction
The recent rise in availability of microprocessors has had a pro Important factors determining the machine malfunction are
found effect on collecting and analyzing machinery vibration mechanical design , machinery typ e , and the function the ma
data for monitoring purposes . A great number of engineers now chine pe1forms in a particular proces s .
have PCs or equivalent on their desks with a variety of software Rotor related malfunctions include: mechanical o r acoustic re
tools for analyzing and presenting moderately large data bases . sonance , unbalance (from a multitude of causes), external pre
M onitoring systems have kept pace by developing hardware loads (e . g . , misalignment between machines ) , internal preloads
that either contains built in "PC" type processors or interfaces (e . g . , misalignment of machine components) , bearing failure ,
with the users PC . Acquisition of data varies from hand entry, surge , cavitation , radial and axial rub s , thermally and mechani
to portable spectrum collectors , to dedicated transducer arrays . cally bowed rotors , coupling lockup, gear wear, instability prob
The Fast Fourier Transform (FFT) is the universally adopted lem s , lubrication los s , loose parts , and cracked shafts , blades ,
method for analyzing data as the frequency content of the vibra or impellers . Housing related malfunctions include bearing sup
tion signal is most indicative of detrimental excitation source s . port failure , excessive piping forces , casing and foundation reso
Significant hardware and software resources are required to nance s , insufficient foundation support, loose parts , improper
TUTORIAL ON VIBRATION MONITORING OF TURBOMACHINERY 103
casing tiedown , foundation material failure , and thermal casing 4

warpage .
Generally, rotor related malfunctions can be best detected by
a vibration monitoring system measuring shaft vibration relative 3
to the bearing cap with an exception of malfunctions related to ><,
lfE ..
high frequency vibration . On the other hand, housing and high 0
frequency related malfunctions are best determined by a vibra

tion monitoring system installed on the machine housing.
i
Ill
2
Q
Mechanical Impedance 5
a..
How much motion can be measured at the rotor and how !i
much motion can be measured at the bearing housing is purely
a function of the mechanical impedance characteristics of the
machine . Mechanical impedance (Z) is the ratio ofimposed force 0
divided by velocity and is established by the stiffnes s , mas s , and 0 3
damping characteristics of the rotor, bearings , bearing support "'
systems , machine casing, and foundation . There are two types FREQUENCY RAnG t;;;
of mechanical impedance which are most often associated with
machine vibration: driving point impedance (Z1) and transfer Figure 3. Effects of Resonance and Damping on Mechanical Vi
bration for a Simple System.
impedance (Z2). Z1 relates a forcing function generated by a
machine rotor to absolute shaft velocity displacement, and Z2 re
lates a forcing function to absolute machine housing velocity dis ROTOR ORBITS
placement, or more specifically: ROTOR
MODE SHAPE
/ �
Zz = F/V(housing)
\Vhere both V(shaft) and V(housing) are absolute motions of those

components .
M echanical impedance ratio T is then defined as :
Using this definition , we can say that machines with high

mechanical impedance ratio will not readily transmit vibratory Fi gure 4. Additional Shaft Proximity Sensors Provide E nhanced
motion energy between shaft and housing; while machines with Definition of Rotor Motion Near Shaft Critical Speeds.
low mechanical impedance ratio easily transmit shaft motion to
the bearing housings . Therefore , machines with high mechani
cal impedance ratio are usually monitored by transducers
measuring relative shaft vibration , and the machines with low A word of caution should be offered about establishing a
mechanical impedance ratio are usually monitored by transduc mechanical impedance ratio (absolute shaft amplitude v s hous
ers measuring machine housing vibration . ing motion); the mechanical impedance ratio varies not only
from one machine type to another, but can vary on machines of
All machinery systems have dynamic properties which con
the same design due to the type of machine problem experi
trol their vibratory response during operation; chief among
enced especially if these problems are characterized by widely
these is the system resonance effects . The mas s , stiffness , and
varying frequencies . It is also not uncommon to see a given
damping properties of the various machinery components con
machine undergo mechanical impedance changes due to varia
trol system resonance , as indicated by the amplification curves
tions in lube oil tempereature and pressure , alignment, bearing
in Figure 3. Obviously the closer the operating excitation fre
preload , and machine speed. When there is sufficient cause to
quencies (w) are to the machines natural frequencies (wr) the
believe that the machine to be monitored is in this middle
greater the amplification of excitation forces such an unbalance .
ground, then installing both shaft and bearing housing sensors
Increasing the stiffness of bearings and supports or decreasing
should be considered.
effective mass of rotating components are practical means of in
creasing machinery natural frequencies . Increasing system
damping is an effective means of reducing amplification without VIBRATION MONITORING SYSTEMS
significantly affecting the natural frequency. M ost of the vibration monitoring systems available on the
M achines with high support stiffness and moderate stiffness market can be classified as simple vibration level monitoring for
journal bearings usually exhibit high damping and have a high trip protection, periodic vibration monitoring, and continuous
impedance ratio . Conversely, machines with moderate support on line computerized vibration data acquisition .
stiffuess and high bearing stiffness (especially ball bearings) usu The simplest vibration monitoring systems for trip protection
ally exhibit low damping and have low impedance ratio. respond to excessive levels of overall vibrations . These systems
Machines built \\ith exceptionally flexible shafts will have very first "detect" the signal using some form of rectification and
low damping resulting in very high amplification of excitation amplifying circuitry. The detected signal is then compared \\ith
sources . Flexible shaft machines also may have nodal points setpoints for trip and alarm. Signal detection is the process of
near their bearings such that neither bearing mounted ac establishing the vibration level . Various circuits exists which
celerometers nor journal sensing proximity probes adequately provide the peak-to-peak, zero-to-peak, or RMS vibration level;
detect hazardous vibration levels (Figure 4) . the choice is mostly a function ofconvention for the sensor used.
M onitoring systems with basic trip protection are \\idely de for interactions and inherent tradeoffs are issues to be addres
ployed and are essential equipment for large turbomachinery; sed. Also identified are the level of capability available in system
however, the only diagnostic information considered is overall products currently in the marketplace . The full list of capabili
vibration level . Additional information such as spectral charac ties so identified may not be available from any single syste m ,
teristics , phase relationship s , and dependence of these features but the list should assist the reader t o identifY those features
on operating parameters (speed, load, time) are often desirable which are most important and to form a basis for evaluating sys
for diagnostic purposes, trending, and warnings of incipient fail tems which are available . The discussion is not a product com
ure . This information usually requires.a continuous online com parison , but rather an attempt to present some of the considera
puterized system or, at least a periodic vibration monitoring. tions which should influence a choice between products for a
The newest class of protection systems appearing in the mar particular application .
ketplace uses digital electronics to give a level of protection, An important step in the process of selecting a computerized
claiming to be· similar to or better than the generally accepted monitoring capability is to set functional goals for the system .
analog circuitry, while adding the ease and control of digital Goals may be adjusted as specific products are reviewed, and
programmability. In an analog protective system , most parame budget realities are faced, but the follO\.,ing is offered as a useful
ters such as signal detection circuit, trip and alarm levels , voting starting point in initial goal setting. Major functional areas and
logic, delay times, etc. must be hardwired prior to installation . associated features are sho\\ n in Table 1, along with the detailed
The benefit of digital based systems is that these parameters can discussion following the table .
be established on installation , and changed at a later time if con
ditions or experience warrant. With appropriate design and use Table 1. Ov erv iew of Comput erized Vibrat ion Monit oring
of modularity and redundancy, it is usually possible to achieve Functions.
the same degree of reliability presently associated with the
analog protective syste m s . The potential for convenient integra FUNCTIONS FEATURES
tion of such digital protective systems with various types of pre • Signal Conditioning • Precision
dictive monitoring system is clear. Cost per channel of these DATA
�CQUISITION
• Multiplexing
• A ...,.. 0 Conversion
•
�P�.� � r::onts
Sampling Rate
•
new digital protective monitors is also reduced. • Bufferino � I
The discussion presented hereafter will guide users in select • FFT • Frequency
DATA Programmability • Alarm Driven

ing and assessing the capability of computerized and periodic MANAGEMEN • Diagnostics • Data Frozen at Trip
vibration monitoring. Additional information can be found in • ANALYSIS • Management of Ccnfllcts

• Dlstlibuted Processing
• Multi·Tiered Storage
references [ 1 , 2, 3, 4, 5].
• Spectrum • Waterfall
Computerized Online Vibration Monitoring System DATA
RETRIEVAL
• Waveform
• Orbit
•Trend
• Bar Chart
" • Bode • Tabular
The application of computers to vibration monitoring is a DISPLAY • Polar • Operator Displays
rapidly changing area. Computerization presents both oppor • Comparison • Resonant Parameters
tunities and choices to users ofturbomachinery who desire more • User Access
• Protection System Interfaces
INTERFACES
complete and accurate information about the health of their • Data Ccllector Interface
• General Purpose Data Base lnt-ce

equipment . For maintenance personnel, it presents the oppor
tunity to change the nature of concern from reactive to anticipa
tory. When the normally installed basic shutdown protection
system performs its function and trips the machine on high Data Acquisition. Before a signal is acquired by the syste m ,
vibration, it generates questions that often require considerable i t usually requires some degree o f conditioning . Common signal
action . conditioning includes low pass filtering, high pass filtering, at
tenuation or amplification of the signal . To avoid aliasing errors
• What caused the trip? created by signals above the frequencies of interest, low pass fil
tering is universally applied to cutoff signals above 0 . 4 times the
• \\'hat action must we taken before we can go back on line? sampling rate . Like\\ise , high pass filtering is used to strip off
DC bias or eliminate very low frequency signal components ,
• What is the impact on future operation? leaving only the time varying portion of the signal which can
then be sampled with much higher resolution.
A computerized monitoring system , however, when properly When either high or low pass filtering is used, the effects
used \\ill generate anticipatory questions well in advance of a should be understood . For example , low pass filtering may have
trip. undesirable effects when the time domain signal or when details
of the orbit are of interest. Similarly, high pass filtering may not
• What is causing this change? be desirable if the absolute level of a signal is of interest - for
example , thrust position signals , shaft radial position signals , or
• What action can we take to avoid a trip? thermal growth signals , since the DC signal level is lost.
The number of bits in the analog to digital (A to D) converter
• When can action be scheduled to minimize impact on (precision) has a significant influence on the resolution of small
operation? signals , particularly when superimposed on a large signal, and
on the detection of changes in these small signals . Attenuation
There are a variety of monitoring systems now available . The and amplification of the signal are techniques used to bring the
choices include a number of systems which emphasize vibration maximum range of the signal within full scale for the A to D con
together with general purpose monitoring and control, and sys verter in order to maximize resolution. Some systems provide
tems which provide limited capability to handle vibration infor autoranging, which automatically attenuates or amplifies the sig
mation as one of a broad range of functions . This discussion will nal to bring its range near to full scale .
concentrate on systems whose primary emphasis is on vibration . The filtering situation is complicated if the user desires to rec
Major functional elements of a computerized vibration ord the DC level of eddy current probes with the AC component
monitoring system s , system performance, and considerations superimposed . For example , most eddy current probes have a
nominal gap setting of about 50 mils . To handle this , full scale lX, 2X, etc . ) , blade passing frequency, etc . A technique avail
must correspond to at least 80 mils . Therefore , with a 12-bit A able in some systems is a programmable diagnostic matrix into
to D, a vibration component or change of less than 0 . 02 mils is which the inhouse machinery expert can introduce likely caus es
lost. With an 11-bit A to D, anything less than 0 . 04 mils is lost of measured symptoms; causing appropriate messages to appear
and with an 8-bit A to D, anything less than 0 . 32 mils is lost . Be if these symptoms arise.
cause of the potential dynamic ranging problem with A to D con All the major systems on the market today have some degree
version of eddy current probes , one choice is to treat the DC of distributed handling and processing of the data. This may con
gap signal and the vibration signal as separate channels handled sist of a series of satellite data acquisition systems which accord
separately by the system if both items of information are of ing to some schedule acquire and store a series of sets of time
interest. series data, for subsequent transmission to a host computer and
Although systems can provide up to 100 KHz sampling rate FFT analysis . The benefits of distributed processing generally
on any individual channel, 20 kHz is more widely available . In include more flexibility, more throughput, and more responsive
addition, most s ystems provide some user control over the sam nes s . Some large systems allow for data acquisition capability at
pling rate for each channel. Sampling rate influences the highest remote site s , with modem transmission to a central com
vibration frequency which can be m eaningfully covered by a fre puterized monitoring syste m .
quency spectrum (typically 0 . 4X sampling frequency) . With a
Presentation o f Vibration D ata. Useful displays o f vibration
20 kHz sampling rate , an assessment to about 130X running fre
data greatly enhance a monitoring systems value for defining
quency of a 3600 RPM machine is possible . Thus , a vibration at
machinery problems quickly. Some of the options available are
blade passing frequency on a 70 blade compressor stage should
as follows :
be detectable .
Spectrum Display showing amplitude as a function of fre
An important consideration during machine startup or shut
quency (Figure 5 [6, p. 224 ] ) . Typically, this is a 400 line display,
down is the speed increment between successive data sets . The
though some systems provide for fewer spectral lines with bene
factors influencing this increment are the speed of the machine ,
ficial reduction in system cost.
the number of channels to be acquired, the number of parallel
data acquisition channels available , settling and autoranging
time after switching, and FFT processing time . .10
Data Management and Analysis. One of the most important GENERATOR GAS TURBINE
functions of data management and analysis is to transform data (I) .08
from the time domain to the frequency domain . Almost all sys !!:
tems perform this operation on the digitized data using the AIR INJECllON BLOWER
Cooley-Tukey FFT algorithm . The function may be performed � .06
in specially designed hardware (sometimes called firmware), in
a general purpose parallel processor, or in software .
g� .04
To establish the time required to acquire , process and store
spectra data, users must consider storage volume, speed of data
acquisition and computation. For example , with 500 channels , .02
requiring as much as six seconds per channel to switch, acquire ,
and analyze , the minimum time to for all channels is close to one
hour if each channel is acquired sequentially; parallel processing
can greatly reduce this time . 100 200 300
The volume of data for long term monitoring is usually con FREQUENCY Hz
served by increasing the interval between data sets preserved
beyond the most recent 24 hours . S chemes for greatly reducing Figure 5. Spectrum Display of Machine Casing Velocity Signa
data storage volume , while maintaining equivalent information ture [6, p. 224].
may consist of condensing the data to overall vibration, static
(DC) levels , one or more spectral lines (e . g . , lX, 2X, 3X Waveform Display in which vibration is presented as a func
machine orders) , and relevant process variables (speed, flow, tion of time (Figure 6) .
etc) .
The benefits of more frequent data acquisition for a machine
indicating a potential problem are clear. This feature is quite
widespread among available systems although differences may
exist in the frequency of data acquisition after an alarm .
Some systems are capable of capturing and holding samples
of data obtained for a short period of time before a trip occurred.
This capability is very useful as it allows analysis of events lead
ing to the trip . Some form of continuous acquisition and discard
ing (or overwriting) of data is required during normal operation
so that, when the alarm sounds , the discarding operation is dis
abled and all of the last data is held .
Some systems include trend extrapolation which will forecast Figure 6. Waveform Display of Vibration.
the time to alarm or trip, or can alarm on a specified rate of in
crease along \\ith absolute vibratin levels . A further aid to inci To support orbit display, it is necessary to have X-Y displace
pient problem detection is the availability of different alert and ment probes installed and to have the time-series data stored or
alarm levels which correspond to different frequency bands . recreatable . Orbits may be filtered (synchronous) or unfiltered
With this capability, the user can program the system to indicate (Figure 7) . Some systems can show the change in filtered orbits
specific problems corresponding to running speed orders (%X, during startup or coastdown.
A Bode plot generally shows synchronous amplitude (and 180

often phase) as a function of speed during startup or coastdown I
(Figure 8) .
A Nyquist plot (also referred to as Polar plot) shows synchron +
X
NYQUIST • x.
ous amplitude vs phase , during startup or coastdown, in polar

form (Figure 9) .
A waterfall plot (also referred to as raster or cascade) shows
changes in vibration frequency spectra as a function of either
rotor speed, load, or time (Figure 10) . •2F��-+�+-��+-��
+ +
Trend plots show variation in some measure of vibration as a

function of time covering short, medium, or long time scales
(Figure 11) . +
"
+
"
I
0
Figure 9. Polar Plot Showing Amplitude, Phase, and RPM Dur

ing Startup.
I
+
5x 10x 15x INTEGRAL ORDERS
Figure 7. Unfiltered Orbit Display 1Jipical of Misalignment.
10
BODE
2495rpm
0 1KHz 2KHz
Frequency (Hz)
I�
\
Figure 10. Waterfall Plot Used for Analysis of B lade Resonance
Problem.
\
\)
16.0
........... , ............... , 12.0
9.0
M
I
��o---,-ooo
�----�--�z�o=oo�--�----��3 oo�o�--��
3 o 8.0
L
p
SPEED (RPM) 3.0
p
Figure 8 . B od!le Plot Showing Amplitude and Phase During o.o

Startup. r··········t··············r··········-,···············r············r···········-,··············r············-,
Figure 11. Trend Plot Showing Overall Vibration for O ne Week
Period on a Shaft Rotor in Vertical ( Upper) and Horizontal
Bar charts show single vibration parameters (e . g . , overall or
(Lower) Direction [2].
1 X vibration) at a number of points
on machine train (Figure 12) .
Tabular data provides specific magnitudes of selected vibra
tion parameters for plant or machine train. Normalized bar chmts for selected critical parameters
•
Displays designed for operators including: where full scale is set to trip or alarm level for that channel.
• Machine train configuration showing vibration sensor loca
Periodic Portable Vibration Monitoring
tions (Figure 13) .
• Simulated analog meters which can be viewed at a glance Some machinery often falls into the category for which it is
for conformance with middle of the range nominal values (Fig not possible to justify the installation of a costly, permanently in
ure 14) . stalled, computerized vibration monitoring system . In this case ,
of a portable data acquisition system is that they eliminate the

need for maintenance personnel to generate charts of data, thus
automating the labor intensive process of monitoring.
Data collectors accept a signal from temporarily or perma
....................................... ···· • ······························································································· : nently installed vibration transducers . This data is stored in the
data collector and transferred into a host computer at the end of
the monitoring cycle for storage , analysis , and reporting of data
using predictive maintenance software . The principal purpose
of a periodic vibration monitoring program is to detect changes
in vibration parameters which indicate the onset of problems ,
and analyze vibration signals if a problem arise s . Vibration levels
are monitored when the machine is in good condition to provide
�--············· · ·
···· ··········· .... a benchmark against future changes .
Some systems are designed t o detect the general condition of
a machine (overall vibration, trending) , while others can offer
Figure 12. B ar Charts Showing Amplitudes of Various Train Vi more comprehensive information for diagnosis of a particular
bration Parameters [4]. vibration-related machine fault (orbit display, frequency
spectrum) .
Portable data acquisition systems are undergoing continuous ,
rapid, development . Therefore , in the future , the user would
face an increasing amount of choices available to him . Some cur
rent features which are important in selecting a data collector
and associated software used with a host computer are pre sented
forthwith .
1 CT 2 3 GEN 4 EX 5
All collectors provide overall vibration readings often re
Figure 13. Machine Train Configuration. quired for detection of a gross machine malfunction . M ore ad
vanced collectors can average overall vibration levels obtained
on the machine, thus reducing the possibility of errors as
B MILS B MILS sociated with unsteady effects influencing the quality ofthe vib
R R ration signal.
PK-PK G
G T 30 Some collectors can show the strength of vibration signal vs
15
H frequency and even provide averaged spectrum information,
9 9 15 which are important features for establishing the source of vibra
10 R
u tion problems .
H 0
v Simultaneous recording of two vibration signals and phase is
5 0 s
a useful feature for displaying orbits and performing machine
E
·= : ...- R T ..... : 15 balancing . This feature is especially useful if the machine is
R I
0 30 equipped with X-Y eddy current probes or dual probes measur
T z
ing shaft and bearing housing displacement simultaneously.
Figure 14. Simulated Analog Meters. Some data collectors can accept input from various transducer
types , for example , eddy current probes measuring shaft vibra
tion , and accelerometers measuring bearing housing vibration .
manual data collection using portable data collectors or elec This feature i s especially important if a plant has machines
tronic note pads (Figure 15) is an attractive alternative method . equipped with different transducer types .
The data collectors offered by several major manufacturers of Other features which should be considered for evaluation of
vibration instrumentation equipment are becoming a practical data collectors include : the amount of data which can be stored
tool in many predictive maintenance programs . One advantage before downloading to a host computer (especially important for
complex systems with several hundred entry points), screen vis
ibility, ease of entering data, weight, size , and operating temper
ature limits .
Host computer and associated software is the key to successful
implementation of vibration monitoring programs , especially if
a large amount of equipment is covered under the program . It
organizes and automates the gathering, analyzing , storing, and
reporting of critical vibration data.
Currently, almost all software available on the market can plot
trends . More comprehensive software can compare vibrations at
particular frequencies or produce waterfall display for viewing
historical spectral trends , which is an important feature for de
tecting and analyzing a particular machine fault. Some also per
mit direct comparison of trends between several locations , thus
allowing for comparison of vibration levels on similar machines .
All software have some capabilities to set machine points and
route s , which may include route repositioning, recalling , delet
ing or adding selected points into the route , as well as route
Figure 15. Typical Data Collector. schedule , instructions , and statistic s .
Cost of Vibration Monitoring Programs ing housings , are to detect housing high frequency rotor related
To justify a vibration predictive maintenance program for malfunctions which include : ball bearing fatigue , support loose
machinery, it is necessary to demonstrate that the potential sav nes s , casing or foundation resonance , loose parts , and blade or
ings of the program offset the cost of the program . gear problems (Figure 16).
The potential savings of the vibration monitoring program in
clud e : prevention of a catastrophic failure , reduced mainte ,..........
nance and downtime cost due to early problem identification,

severity assessment of a vibration-related problem, orderly
backtracking after a trip or failure , and reduced insurance costs .
The costs associated with the program include: purchase price
and installation of the vibration monitoring equipment; labor
and material cost for the vibration system maintenance and data
handling; lost production due to nuisance machine shutdowns
attributable to vibration monitoring equipment malfunctions .
It should be noted that the validity of the numbers associated
with the cost/benefits analysis of vibration monitoring program
is often questioned due to variability of the assumptions in their
computation procedure . Some companies rely on several
scenarios to establish the range of potential savings/loss as
sociated with the implementation of a vibration monitoring sys
tem . I n general, the level of vibration monitoring protection
needed will depend on the type and severity of the service , crit
icality of equipment , number of units at one site , availability and
qualification of maintenance personnel, and machinery type and
age .
Some typical equipment costs are shown in Table 2 of vibra Figure 16. Casing Acceleration and Velocity Vibrations Signa
tion monitoring systems , broken into three categories : basic vib tures Recorded on the Compressor Section of T hree Gas Thr
ration monitoring for trip protection, periodic vibration monitor bines [6, p. 145].
ing, and on-line computerized continuous vibration monitoring.
Since every transducer has its merits along with its limita
Table 2. 7l!pical Vibration Monitoring System Costs. tions , no single type can be relied upon to meet all application
requirements . Therefore , many critical installations require
TYPE OF more than one transducer type for complete protection. A sum
i!!IY!lT�U eo!!IT
$60010$900 Condnuoua analog mary of advantages , disadvantages and application of the various
Basic VIbration per channel protection, baaed on transducer types is presented in Table 3, followed by a specific
Monitoring for overall vlbradon level. description of each transducer type and its application.
Trip Protecdon sensor cast not
IMiurioOif.
E ddy Current Probes-Measurements of
$20,000 10 $25,000 for Vlbradon apeclrUm data
Periodic asingle dalll CXIHectot recorded at regular
Shaft Relative Vibration and Position
VIbratiOn and a PC ba88d data lntetvall-typlcally Keyphase measurements and measurements of shaft vibra
Monitoring management lyttem monlhly comparllon
wl1h buellne 1rendl tion or shaft position relative to the bearing clearance and axial
"'dAtil. shaft movement , are normally made using eddy current probes.
40 channel system: Regular storage of For shaft vibration and position measurements the transducer
On-Une $1500 10 $2000 llpeclnlllntormadon ll'ld
is installed in a threaded hole in the bearing (Figure 17) or on a
Computerized per channel lnlndlng; variety of
COntinuous 500 c:hannel system: dlsplaya Including orbits,
Vlbra I $30010$700 Nyqulat, Bode; Watlrfall .
Monitoring per channel COlt dOel not lncluc:le
trip proteCtion, monitor,
and senaora.
The costs represent recent estimates and are presented as

guideline s ; accurate costs can only be developed when quota
tions are obtained from a supplier for a specific application.
VIBRATION TRANSDUCERS
The installation of vibration measuring devices has become
common practice on critical and costly industrial machine s . The
underlying objective is to detect an increase in vibration level
before mechanical problems occur. To achieve this , it is neces
sary to select a transducer type that will measure the vibration
(machine housing or shaft) most likely to reveal the expected fail
ure characteristics [7, 8] . Shaft measurements are recommended
for rotor related malfunctions which include : imbalance , mis
alignment , shaft bow, and fluid film bearing instability. Velocity
pickups or accelerometers , installed on machine casings or bear- Figure 17. X-Y E ddy Current Probe 1nstallation.
Table 3. Comparison of Vibration Transducers. shaft, and its AC component represents the dynamic shaft dis
placement (vibration). Shaft material, changes in cable length,
TRANSDUCER
ADVANTAGES INSTALLATION DISADVANTAGES
and sometimes temperature , influence the probe sensitivity of
TYPE
voltage vs gap.
•Can measwe static
and dynamic shaft
•Mechanical oreleclri
cal nmout DOise.
Popular oo macbioes
with Ouid film bear
To monitor shaft vibration, ideally, two eddy current probes
motioa. •Limited high fre. ings. Usually ooe pai.r are installed 90 degrees apart on each gas turbine bearing. The
•Accurate low fle queocy scositivity. ofX-Yprobes90" apan
EDDY CUilRBNT
queocy....,..... -calibntioo Jeasitive per bearing. two probes together can provide valuable information on vibra
NONCONTACT
DISPLACEMENT
•No wear.
•Small size.
to shaft materials.
•Difficult to install or
tion amplitude levels , shaft orbit, and shaft position. Commonly
PROBPS
•Measures directly in replace. used noncontact high temperature probes are rated up to 350°F
dlspla:emeJltllllits,
•l!asyiOcoli......
• R equires extemal
power source.
(180°C). For higher temperature s , specific probe environmental
limits should be reviewed. The typical probe tip diameter is ei
.Simple to install witb •Tr.msducerteSOIUIDCe Typical on older ma ther 0. 190 in (5 mm) or 0. 300 in (8 mm). By adding two X-Y re
space available. ooise aod pbase Rbift. cbioes. Mou01ed oo
•No exaemal power <ross-axis aoise. machiDecasiagorbear dundant probes at locations which are difficult to access , and op
VELOCITY
""""" ·
•Stroogrespoaseiatbe
oCao be affected by
magocticfielda.
iog bousiog. erate in a harsh environment , the reliability of the monitoring
PICK-UPS mid-liequeocymoge. •Perfonnaocedepada system can be enhanced.
.Can be iDstalled on 1 tioa due to wear.
lemponrybasis. •Difficult to calibn.te. Knowledge of the vibration phase angle enhances diagnosis
of a number of vibration problems. An additional eddy current
•Simple toiDB1all.
•Good bigb liequeocy
•A Jemote dlarge am
plifier requiJed for
Typical oo ain:reft de
rivative ps turbines,
transducer observing a once per revolution event , such as a
...,..... bigb .._fiiUie .... fans, llllial compn:s- notch on the shaft, reliably provides the needed phase refer
•Somemodelssoitable viroomeot (approxi· 1018,aod small pwnps.
forbigblem mlllelyabove260"F). Somelimes added to or ence. For temporary installations , a piece of reflective tape , at
ACCBLER
OMI!TI!RS
eaviroameot. •Possibility of low lie replaced older velocity tached to the shaft surface , observed by an optic sensor, can be
•Small s!... queocy ..... due 10: pict-upo. Occasiooally
•Relalively goochelia iotegratioo to dis addediOexistiogvibra sufficient for a phase angle measurement.
bility. placement, looae coo lioa probes for im
oec:ti.ODS,etc, proveddiagaootics. One of the important parameters obtained with eddy current
probes is a steady state position of the rotor with respect to the
oCombioedadv.,..gea •Combioed disadv��Jw Conunoooosomclaqe bearing clearance. However, this type of measurement should
of eddy cumolprobes tagea of eddy cuneat iDduolrialpiUdlinos.
aod velocity pick-ups probes aod velocity Usually ooe probe per be viewed with caution when applied to high temperature envi
or accelemmeten. pick-ups or acoeler beariq, iusblled iD ronment. Even high temperature probes, which minimize the
�nfluence of temperature on voltage vs displacement curves , can
DUAL PROBPS ·Me- obaft aod omeoen. vertical dire ctioa.
bousias vibration Sometimes two probe&
separaoely, and caD 90" apan perbelllils
measure abs o lute
l .
mtroduce some measurement errors. If the probe gap is set ini
obaft dlsplacemeut. tially with shaft stationary at ambient temperature , the effect of
increasing temperature on these probes makes the apparent gap
rigid bracket so that the tip of the pickup can be brought close less than the true gap. For probe and driver systems which have
to the shaft. The eddy current probe assembly consists of a trans been optimized for high temperature operation the uncertainty
ducer, extension cable, connector, and oscillator-demodulator in gap introduced by increasing temperature can be limited to
(Figure 18). An external oscillator provides a high frequency car 1.0 to 2 . 0 mils at 200°F, and 2. 0 to 5.0 mils at 350°F. These tem
rier signal to a coil in the pickup tip , producing a magnetic field perature optimized probes tend to be of larger diameter (0. 300
radiating from the tip of the displacement probe on to the shaft. in or 8 mm as opposed to 0. 190 in or 5 . 0 mm). Available calibra
The shaft of the machine creates an impedance to carrier signal tion data indicates that the uncertainty can reach 10 mils at
amplitude , and this impedance varies inversely ·with the gap be 200°F if the probes installed are not optimized in high tempera
tween the shaft and pickup tip. As the shaft approaches the ture operation. Because of this temperature sensitivity shaft ab
. solute position signals must be interpreted with considerable
probe tip the magnetic field strength reduces due to generation
of the currents on the shaft with the carrier signal amplitude care and with awarenes s of the effect of temperature on the D C
modulated proportionally to the shaft displacement. Additional signal o f the probe in question. To minimize the uncertainty,
normalization of the carrier signal adjusts its output to a specific calibration curves should be obtained for the specific probe s to
sensitivity usually 100 or 200 mv/mil. The typical linear probe be installed at different temperatures e ncompassing the antici
range extends from 10 mils to 80 mils gap. The signal DC compo pated range; the bearing metal and oil temperature should be
nent represents the average gap between the probe tip and the measured; and the effect of temperature on both level and slope
of the curves should be considered. As an example of the uncer
tainty introduced by temperature sensitivity, under normal
loading conditions , the shaft can appear to be running in the top
half of the bearing. At the same time, properly calibrated and
interpreted probes can give reliable information about shaft
position.
The most common source of noise found in ed dy current
probes is shaft electrical or mechanical runout. Often runout er
rors are created when a journal is flame sprayed or plated to re
build worn surfaces. To prevent the probe signals from becom
ing useless after such an operation, the shaft s urface sensed by
the probe should be properly masked. S urface grinding and rol
ling or shot peening are treatments that improve uncoated
shafts. Alternately, special analysis equipment can subtract the
runout signal. A runout subtractor first records the phase and
the shaft displacement at slow roll when all motion is presumed
to be due to runout and then vectorially subtracts this displace
ment from the displaceme nt readings obtained during machine
Figure 18. Typical E ddy Current Assembly. operation.
Accelerometers-Measurements of Housing Vibration

An accelerometer is an electromechanical transducer which
produces voltage proportional to the acceleration to which it is
THIN FILM OF
subjected. Among many different types of accelerometers , the
SILICONE GREASE
piezoelectric ones are the most common in turbomachinery ap
plication. The piezoelectric accelerometers can be divided into
two major groups: compression and shear types (Figure 19 and
Figure 20) .
PIEZOELECTRIC
CRYSTAL STACK
Figure 21. Installation of a Thread Mounted Accelerometer.
Velocity Transducers-Measurements
-CONNECTOR
of Housing Vibration
Seismic velocity transducers often measure bearing housings
or machine casing vibration. A typical velocity pickup is shovm
Figure 19. Compression 11jpe Piezoelectric Accelerometer [6, p. schematically in Figure 22 [6, p. 35] . A cylindrical coil in the pick
37]. up is attached to a permanent magnet suspended on springs .
The signal is generated by the motion of the coil in the magnetic
field of the magnet. The spring suspension system is designed
to have a very low natural frequency so that the magnet remains
stationary in pace at frequencies above 8 to 10 H z . A damping
medium, typically a synthetic oil, is generally added to damp
critically the natural frequency of the spring mass system and
roll off its response characteristics below approximately 10 H z .
For permanent monitoring the transducer is usually bolted o r
stud mounted directly t o the bearing housing or t h e machine
CONNECTOR
casing . For temporary monitoring or probing s urveys , the pick
ups can be hand held or magnetically secured on the machine
casing .
Figure 20. Shear 11jpe Piezoelectric Accelerometer [6, p. 41].
A typical piezoelectric transducer consists of one or several

piezoelectric crystal elements which are loaded by a mass or
masses and preloaded with a spring. The piezoelectric crystal
produce an electrical charge proportional to the applied force
when loaded in compression (compression type accelerometers)
or shear (shear type accelerometers) . Converting a piezoelectric
crystal generated signal to a signal suited for use with conven COIL ----lit'.'..:::·:
tional vibration monitoring devices requires internal or external
electronics . An accelerometer with built-in electronics can only
withstand temperatures up to 260°F (125°C ) . Therefore , re
motely located charged amplifiers are required for high temper
ature accelerometers which are rated up to 1380°F (750°C ) . The
llf!:
signal strength amplitude of a piezoelectric accelerometer is af
fected by changes inn temperature . This becomes especially im
portant when the accelerometer is subject to varying operating
conditions or temperature gradient across the accelerometer
surface; gas flowing over the accelerometer can be of signifi
cantly different temperature from the temperature of the mount Figure 22. Typical Velocity Pickup [6, p. 35].
ing surface . With a remote charge amplifier, the connecting
cable can be a source of additional noise due to loose connec
tions , ground loops , triboelectric noise (usually used by separa Although some velocity pickups are available \\oith linear
tion of the conductor and shield due to vibration,) and elec amplitude response from 1 to 5000 Hz, velocity pickups are gen
tromagnetic noise due to nearby electric fields . erally used to measure vibration displacement at low and inter
Some special considerations also apply in mounting ac mediate vibration frequencies , typically 10 to 1500 H z .
celerometers to the vibrating surface . Accelerometers are usu High temperature velocity pickups are usually rated up to
ally mounted on a flat, smooth surface by a threaded stud (Fig 600°F to 700°F (315°C to 400°C) , which is generally adequate for
ure 21) . This mounting technique offers significant reliability most installations . Some velocity pickups have a limited longev
and repeatability than alternative mounting techniques such as ity and often require replacement after one to two years of
epoxy or magnet attachments . operation.
D ual Probes-Measurements of Absolute and current probe added to the vibration reading from the seismic
Relative Shaft Vibrations probe . Others use the relative shaft displacement data for de
tecting a vibration problem and utilize the vibration information
Some large industrial rotating machinery installations have a
from the seismic probes to assist in identifying the root cause of
dual probe (Figure 23) which combines an eddy current probe
the problem .
and a seismic transducer (usually a velocity pickup) . With this
combination, signals are available for detecting relative shaft ADVANCED METHODS FOR DETECTING
vibration , absolute shaft vibration , or absolute bearing housing BLADE VIBRATION AND TIP CLEARANCE
vibration . Since the signal from an eddy current probe is propor
tional to dynamic shaft displacement relative to the bearing and Several independent industry surveys show that blade re
the signal from a seismic probe s ignal is integrated to the appro lated problems are the leading cause of turbomachinery equip
priate displacement units and added to the shaft relative dis ment failures . The consequences of a lost blade are usually se
placement signal to obtain shaft absolute displacement . vere , and can cause complete destruction of the unit. Several
techniques are now emerging which show promise for diagnosis
of blade failure causes and for detecting incipient blade failure .
The promising techniques include . strain gage telemetry to
monitor dynamic strain, accelerometers to detect the proximity
VELOCITY
ofblade resonances to excitation frequencies , impact tests to de
SEISMOPROBE fine static blade natural frequencies , and noncontact sensors (op
tical, eddy current, capacitive , magnetic, and acoustic-doppler)
to monitor, tip clearance , blade position , and time of arrival.
Some of these techniques are discussed later in more detail.
Strain Gage Measurement
Advanced strain gage technology have resulted in several ex
periments to measure cyclic stress of blades and impellers . Slip
PROBE rings and telemetry systems may serve as the link between the
SLEEVE strain gage and the data analysis system . Slip rings have rela
tively short life span and are prone to signal noise during rota
tion . In telemetry systems , signals from the sensors are encoded
onto a radio frequency carrier by transmitters. The transmitted
signals are received by a stationary antenna and receiver, which
decodes the signal . S train gages are advertised to survive tem
peratures up to 2100°F and centrifugal force fields in excess of
20, 000 gs . Installation of strain gages is a laborious process re
quiring considerable preparation such as surface degreasing,
bead blasting , and the use of special bonding compounds .
Laser- Optical B lade Tip Clearance Measurement System
Laser-optics systems offer the means to make precision di
mensional measurements of rapidly moving rotating objects .
One of the laser-optics system designs currently employed by
Although the use of a single vertical dual probe is quite com one of the jet engine manufacturers uses a laser generated light
mon two dual probes on each bearing 90 degrees apart are desir source , which is focused on a single optical fiber and is imaged
able (Figure 24) . Some companies rely on the measurements of on the target blade tip through a lens and sapphire prism (Figure
absolute shaft vibration , i . e . , vibration readings from the eddy 25) . The reflected light returns through the prism and lens and
is focussed on the output coherent fiber optic bundle . With air
cooling the probe can function up to 2000°F and has been used
in jet engine tests for measuring average blade tip clearance .
With appropriate signal proces sing, this probe appears capable
of measuring not only rotor motion and individual blade tip
clearances , but blade vibration and rotor torsional vibration.
Typical accuracy of this probe is 0 . 003 in or les s . The probe usu
ally provides from 20 to 50 hours of continuous operation before
being removed for cleaning or other repairs .
Capacitance and E ddy Current Proximity
Sensor Tip Clearance Measurements
Capacitance proximity sensors determine the distance be
tween the sensor and the target (blade tip) by measuring the
capacitance reactance of the sensor/target system . A typical
capacitance sensor assembly consists of a sensor surrounded by
an insulated cylindrical guard (Figure 26) . The guard ring is con
nected to the shield and the sensor is wired to the coaxial cable
electrode . Capacitance sensors are linear over a range of 0 . 4
sensor diameters and are accurate t o 0 . 003 i n for typical appli
cations . Capacitance probes have been reportedly used in tern-
Figure 24 . X-Y Dual Probe Installation [2].
racy of 0 . 002 in . Inductive interference requires that the turbine

casing be pierced by a hole at least six times the desired mea
surement range .
Both capacitance and eddy current probes have degraded
range capabilities for thinly tipped blades . Special tuned elec
tronics are required to provide measurements at high blade pas
sage frequencie s . There is also a need for enhanced signal pro
cessing software to expand the ability of the probes to monitor
individual blade clearance and vibration .
MORE COMPREHENSIVE VIBRATION

LIMITS AND TEST CASES
Due to the large variation in machine design s , vibration limits
are based on statistical or consensus evaluations of vibration and
failure data from many operating units . As such, there are no ab
solute limits that "Will assure successful longterm operation nor
firm upper limit that will cause failure for any specific machine .
The best to be hoped for is to increase the probability of success
ful long-term operation and minimize unscheduled outages due
to equipment failure . U sing statistical criteria, there will always
be unexpected failures and amazing vibration endurance of
some machines . In general, high vibrations are bad, low vibra
tions are goo d , and the line separating the two is somewhat un
certain . The authors know of several plants where a vibration
reduction program resulted in considerable savings due to re
duced machine outages . Thus , lowering the vibration to reason
ably oJ:>tainable levels �hould be encouraged .
Figure 25 . Laser Optical Sensor System. Vibration in rotating machinery may be the result of several
different phenomena and affe ct various machine parts . There
fore , reliable vibration measurements and their assessment
perature environments in excess of 2000°F without external should be based on each particular machine type , its installa
cooling . Capacitance probes calibration has been typically tem tion, vibration sources , and failures that are likely to occur.
perature sensitive and the probe reliability has been hampered Most vibration failures can be classified according to the fol
by the presence of moisture and other contaminate , but recent lowing consequences :
improvements have resulted in significant reductions to these
problems . • Structural fracture due to fatigue or dynamic overload .
M anufacturers of eddy current probes advertise recent ad
vancements that allow uncooled operation to 1100°F with accu- •Wear, fretting, or surface fatigue of bearings , gears , cou
plings , etc .
• Performance loss due to internal machine clearance rubs

(seals , blades , impellers) .
An important factor to recognize in applying vibration limits
is growth of vibration amplitudes with time . All machines , re
gardless of their health , exhibit some form of vibration; how
ever, on machines where a mechanical defect has deteriorated
to an unacceptable level, rapid growth in vibration and conse
quent failure can be expected. In addition to vibration severity
criteria, machine resonance characteristics should be con
trolled. The margin between the rotor bearing resonance (criti
cal speed) of a machine and its operating speed range has a very
strong influence on the resultant vibration sensitivity to unbal
ance . Only a slight change in this margin can have a dramatic
effect on vibration level . This is also true for such sources as
blade resonance problems on fans and turbines .
During the past several years , there have been many vibra
tion severity guidelines proposed by standards organizations [9,
10] , technical societies [ 1 1 , 12, 13, 14] , and equipment manufac
turers [ 15, 16 , 17] , along with experienced individuals [ 18] .
Using these s tandards as a base and adding the experience ac
cumulated by S outhwest Research Institute "With various types
of rotating equipment, vibration limits have been established
for machine housing vibration (Figure 27) and for shaft vibration
relative to bearing housing (Figures 28 , 29) . These vibration lim
its cover a wide range of equipment, installation types , and
Figure 26. Capacttance l'roximity Probe Installation, machine malfunction characteristics .
TUTORIAL ON VIBMTION MONITORING OF TURBOMACHINERY 11.3
1 .0
1-
F
�85 �i
.8
r- D
.6
>d 1-
!� 4
c
�-
.
�i
� <( .2
r-
B
� zs
1- A
0 I I I Ill j_ _l _l_lll I I
100 200 500 1000 2000 5000 10000 20000 50000
SHAFT OPERATING SPEED, RPM
Figure 29. Shaft Vibration Limits Related to B earing Clearance.
ration should be used with a table of appropriate correction fac

tors (Table 4) .
Relative s h aft vibration is measured by proximity probes . Two
vibration severity charts are p rovided for relative shaft vibra
tion . The first chart (Figure 28) utilizes only actual relative shaH:
displacement and requires usage of appropriate correction f�lc
tors . The second chart (Figure 29) establishes journal severity
Figure 27. Ho using Vibration Limits. limits with respect to bearing clearance . The probability of bear
ing metal-to-metal contact and unacceptably high bearing load
increases with the ratio of mtt.'l:immn journal motion to bearing
50
1- clearance . It should be noted that this chart does not require the
1- application of any correction factors .
20
1l1ble 4. Vibratio n Service Fa ctors.

� F SERVICE
FACTO R S A P P L I CAT I O N
D
1-
1- � "'
c
!'.... t"- K, = 1 .0 N O R M A L FACTO R F O R F I LT E R E D V I B RATI O N
�
...... N EA R R U N N I N G S P E E D
B
�1' b-... K, = 0.85 U N F I LTE R E D V I B R ATI O N F O R N EA R R U N N I N G
I; ...... ...... SPEED SOU RCES
r:
A
" �!'- K, = 3.3 A N Y SELF-EXCITED O R U N STA B L E V I B R ATI O N
.......
1- !"--- SOURCE
1- �,....., K, = 1 .4 E Q U I P M E N T R AT E D LESS T H A N 300 H P
K , = 0.7 R I G I D ROTO R M A C H I N ES
0.1 I I 1 11 1 I I I Il l I I J lll K, = 0.6 SOFT FO U N DATI O N M A C H I N ES

1 00 1 00 500 1 000 2000 5000 1 0 K 20 K 50 K 1 00 K
K, = 3.5-10 H I G H C AS I N G -TO- R OTO R M A C H I N ES
VIBRATION FREQUENCY, CPM
K, = 3.5 H I G H F R E Q U E N C Y B L A D E V I B R AT I O N S O U R C E S
Figure 28 . Shaft Relative Motion Vibra tion Limits. K, = 0.35 H I G H F R E Q U E N CY G EA R OR R O L L E R B E A R I N G
SOURCES
Of course , any vibration criteria utilized should be compared Five quality grades for measured vibration are defined f(Jr
with the manufacturer's recommendations and user's experi these charts :
ence with the particular type of machinery. Obviously, the man A No faults (typical new equipment) .
ufacturer's vibration limits should be used , if they are more B Acceptable (correction is not necessary) .
stringent and have a clear basis . C Marginal (correction is recommended to save on future
The housing vibration chart (Figure 27) corresponds to vibra maintenance) .
tion, which is measured on the machine casing or bearing hous D Failure probable (watch closely for changes and prepare to
ing utilizing velocity pickups or accelerometers . For more con shut down or change operating conditions to reduce vibration) .
venient analysis of vibration problems , this chart is divided into F Danger of immediate failure .
regions where velocity or acceleration measurements are most Some important points should be emphasized to assure that
appropriate . these vibration charts (Figures 27, 28) are properly applied
It is important to realize that some subjective judgement is when machine vibration measurements are evaluated:
required in conjunction with these charts . To equitably accom
modate different machine designs , installations, and vibration • Vibration limits for machine housing vibration should be
problems , the charts for housing vibration and shaft relative vib- considered separately from relative shaft motion .
114 PROCEEDINGS OF THE NINETEENTH TURBO MACHINERY SYMPOSIU M
• A n effective vibration value must be obtained b y multiply A filtered bearing housing vibration of 1 . 2 mils was measured
ing the measured vibration value by one or more applicable cor during the resonance period . Multiplying this vibration by the
rection factors listed in Table 4. Correction factors that apply to appropriate correction factor ( 1 . 2 X 3 . 5) yields 4 . 2 mils , which
housing vibration include : rotor rigidity, foundation type, power is in the "F" or immediate failure region of Figure 27 for 2700
rating, casing to rotor weight ratio , high frequency vibration , in cpm . After several days of operation and several startups , the
stabilities , and unfiltered vibration . Shaft vibration service fac vibrations increased suddenly and the fan was tripped off. Sub
tors differentiate running speed vibration from rotor in sequent inspection revealed blade cracks near the welds , which
stabilities , power rating, and unfiltered data. indicated that blade modifications were necessary to avoid exci
tation of transient resonance s .
• If vibration frequencies below running speed are apparent, • I n the third example , bearing housing measurement indi
then one must determine if the source is an instability or forced cated 14 . 5 mils filtered vibration on a 900 rpm induced draft fan .
excitation . If it is an instability, then the factor of 3 . 3 must be A s the rotor critical speed and foundation resonance were found
applied to the measured data and entered in the chart at the run by testing to be 1100 cpm and 720 cpm , respectively, then cor
ning speed frequency. rection factors for rigid rotors (K5) and soft foundations (K6) can
be applied . The effective vibration is calculated by multiplying
• The severity criteria based on unfiltered vibration readings these factors by the measured vibration (14 . 6 X 0 . 7 X 0 . 6) which
is only applicable for excitations near running speed frequency. yields 6 . 1 mils . Entering this effective vibration on the housing
In this cas e , an additional correction factor of 0. 85 is applied and severity chart (Figure 26) indicates the severity level is between
the resultant effective values entered on the charts at shaft rota marginal and probable failure , which is undesirable for longterm
tional speed . operation. Correction of the thermal bow problem and rebalanc
ing reduced the vibration to 2 . 0 mils , or an effective vibration
• To convert filtered vibration displacement in mils to veloc of 0 . 8 mils , which is acceptable .
ity in /sec, and acceleration in "g, " the following formulas should
• A lube oil pump motor for an auxiliary turbine had experi
be used:
enced repeated bearing failures prior to commissioning. An in
5 vestigation of the system dynamics found that 13 mils filtered
v peak = 5 . 23 X 10- Dp-p X N casing vibration occurred at the running speed frequency of
3600 cpm and the applicable correction factors , K4 and K5, es
2
A peak = 1 . 41 X w-s Dp-p X N sentially cancel each other and the effective vibration is well into
the immediate danger region so the pump was shutdown .
where Dp-p - peak-to-peak displacement measurement, mils Further investigation yielded that the motor shaft had excessive
V peak - 0 to peak, velocity measurement, in/sec misalignment and the motor casing resonance existed only 1 . 5
A peak - 0 to peak, acceleration measurement , g percent below running speed . Once the alignment problem was
N - frequency, cpm fixed and the motor support was stiffened to move the case reso
The application of these vibration severity criteria are nance, vibrations were reduced to acceptable level and no
clarified by reviewing the results obtained from field studies of further problems were reported .
actual plant equipment . • An investigation of a 3000 hp steam turbine was conducted
• The first example involved a troubleshooting investigating to determine if it was safe to operate the unit . M aximum vibra
and balancing of two boiler feed pump turbines operating at tion readings of 1 . 5 in/sec bearing housing motion and 1 . 6 mils
4000 to 5800 rpm that had experienced bearing failure . Turbine shaft relative motion were found at running speed frequency of
rotor resonances were found at 2200 and 5800 rpm . For one tur 11, 400 cpm . In evaluating the vibration severity, a service factor
bine , the apparent filtered shaft relative vibration was about 3 . 5 of 1 . 0 was applied as no other factors were appropriate . It was
mils near top speed , but further investigation revealed that 1 . 4 decided to shut the turbine down immediately as the bearing
mils o f the signal was due t o shaft runout. For the remaining 2 . 1 housing was in the "F," or danger, region and shaft vibrations
mils o f actual shaft vibration , a correction factor (K1 ) o f 1 would were in the "D , " or probable failure , region. Inspection revealed
apply. Comparison of this effective vibration with Figure 28 indi some seal rub damage and excessive unbalance , which was 40
cates the level was marginal . times of specified value . Further investigation indicated that
On the adjacent turbine , a filtered bearing housing vibration high unbalance was due to the previous shaft repair which in
of3 . 1 mils was measured at maximum running speed . The struc volved sleeving one of the journals causing shift of the rotor mass
ture was considered soft mounted since a turbine pedestal reso center. Rebalancing at high speed (flexible shaft) using multi
nance of 3400 rpm was present below running spee d . Applying plane balancing techniques and adding weights on both sides of
a correction factor (K6) of 0. 6 yields an effective vibration of 1. 8 the sleeve journal brought the vibration down to the no fault re
mils . Comparison of this effective vibration with Figure 27 indi gion, "A ; " housing vibration was reduced to 0 . 025 in/sec and
cates that failure is probable , which was consistent with experi shaft vibration was reduced to . 5 mils . Subsequently no prob
ence on the turbines . lems have been reported for this turbine .
Balancing reduced vibrations to less than 1 . 0 mil o n housing To summarize major points for applying vibration severity
and shaft . Thu s , after applying the appropriate correction fac criteria to assess machinery condition :
tors , final vibrations were acceptable for both turbines .
• Draft fan blade failures are especially dangerous , as the • First hand experience with a variety o f equipment has
housing cannot restrain the massive parts that might be released shown that the simplified vibration criteria used in the past is
from the impeller. In the second example, an investigation of the less than adequate for many actual rotating components .
original design blade vibrations on a 900 rpm fan revealed a
blade resonance of only about six percent above three times
running speed when the fan was cold. Changes in operational • To equitably accommodate different machine design s , in
temperatures every time the unit was started would cause this stallation s , and vibration problems , an effective vibration value
resonance to match with the excitation frequency at three times must be obtained by multiplying the measured vibration value
running speed for brief periods . by one or more applicable correction factors listed in Table 4 .
• There are no absolute limits that will assure successful

longterm operation , nor firm upper limit that will cause failure S EPA R AT I O N
for any specific machin e . As a result, some subjective judgement MARGIN
is always required in application of any vibration limits .
• Any vibration criteria utilized should be compared with the
manufacturer's recommendations and user's experience \\ith
the particular type of machinery.
• An important factor to recognize in applying vibration limits

is growth of vibration amplitudes with time. Machines where a
mechanical defect has deteriorated to an unacceptable level,
rapid growth in vibration and consequent failure can be 1 ST C R ITICAL O P E R AT I N G 2ND CR ITICAL
SPEED SPEED SPEED
expected .
Figure 30. Rotor Critical Speed Effects on Vibration Response.
• To properly assess the severity of existent problem, i t i s
necessary t o establish the source o f machine vibration and
selected appropriate transducer to measure the vibration have occurred as a problem in the field . The blade design should
(machine housing or shaft) most likely to reveal the expected fail be investigated thoroughly prior to actual manufacturing of the
ure mechanism . unit . The fundamental principle in dealing with blade vibrations
is to avoid resommce . The blade natural frequency should not
be \\ithin 10 percent of any identifiable excitation frequencies
SOURCES O F VIBRATION at any point in the normal operating range , including multiples
Establishing the proper source of excessive vibration is prob of blade passage frequency and multiples of running speed. The
ably the most difficult task of machinery analysis . After the aerodynamic phenomena are too complex to be predictable at
source of vibration is determined, it is generally possible to as the present time and vary greatly from machine to machine . As
sess the severity of the existing problem by comparing the vibra a result, any attempt to live with an excitation of the lower four
tion level against a specified vibration standard . As the sources or five blade modes is likely to cause blade loss failure regardless
of the various excitation forces often occur at different frequen of excitation force predictions . If the frequency of the driving
cies, it is very useful to characterize measured vibration signals forces cannot be changed, the blade design should be changed
with respect to frequency. Thus, spectral analysis is a recom to detune it from resonance . The calculation of the natural fre
mended procedure for solving vibration problems and evaluat quencies of a blade are quite involved and should include the
ing the reliability of equipment. effect of centrifugal forces combined with a knowledge of the
Vibration amplitudes obtained from spectral analysis should root fixity factor. Simplified calculation s , however, can give suf
be investigated in relation to machine design , installation, and ficient accuracy for the lower modes to determine if a blade vi
vibration source . Judicious placement of vibration transducers bration can be a problem .
so that a vibration source can be separately identified is neces Subsynchronous Instabilities
sary to gain maximum use of the vibration criteria. Separation
Subsynchronous instabilities are another possible source of
of gear mesh excitation from turbine blade vibrations , for exam
vibration in plant operating equipment. "Instabilities" is the
ple , might be determined by analyzing signals from a few trial
term given to those vibration sources that are self excited by
locations and applying judgement based on knowledge of the
some mechanism where fluid, aerodynamic, or frictional forces
machine's design and exact frequencies of each source . An over
interact \\ith the rotor. Instabilities usually exhibit vibration fre
view of several most common vibration sources follows .
quencies at aboutl/2 running speed (Figure 31) and have a ten
Resonance dency to suddenly increase in amplitude with disastrous results .
Resonance is probably the most common cause of high vibra Rotating stall of draft fan air foils at low flow is another source of
tion and most vibration related failures . Very often user vibra subsynchronous rotor vibration , but generally is not as danger
tion specifications ignore resonances completely and depend ous because the excitations are not unstable , but are simply the
upon the manufacturer for such criteria. If resonant margins are result of unbalanced flow forces . The following item s increase
not specified and verified by testing, the component could be the susceptibility of a rotor to subsynchronous instabilities :
very precisely balanced or-adjusted to meet the acceptable vi • Rotors operating above the first critical speed .
bration limits at the time of purchase . However, longterm use
of such a machine may result in high maintenance requirements • Compressors with high outlet pressure and high molecular
as slight unbalances or other distortions accumulate causing vi weight gas .
brations to increase to unacceptable levels . General guidelines
on critical speed margins for lateral shaft vibration are provided • Lightly loaded bearings .
by API standards for several different types of rotating equip
ment . For example, compressors and pumps are required to • Presence of high pressure pulsations- and acoustical
have lateral critical speed margins of 20 percent above maxi resonances .
mum operating speed or 15 percent below minimum operating
speed and amplification factors should not exceed eight while • S tiff radial bearings relative to the shaft flexural stiffues s .
going through criticals (Figure 30) . If the measured critical
speed falls \\ithin the excluded range , then the manufacturer • Inadequate seal design .
must demonstrate that the vibrations at the critical speed are
\\ithin acceptable vibration limits with considerably more unba • S udden shock loads .
lance than would normally be expected .
-Special attention should be given to blade related problems . • Presence of rotor frictional sources such as light shrink fits
Very little can be done to minimize blade vibrations after they over a long section .
116 PROCEEDINGS OF THE NINETEENTH TURBO MACHINERY SYMPOSIUM
/ S U B S Y N C H R O N O U S FREQUENCY or casing which is in exact synchronization 'A-ith rotor speed (Fig

ure 32) . Unavoidable geometry imperfections due to fabrication
or material variation will result in rotor unbalance such that vir
tually all new high speed machines will require balancing at the
2x S U B S Y N C H R O N O U S F R EQU ENCY factory. Some very sensitive machines will also require rebalanc
ing after installation . Vibration should be expected to gradually
change With time as the effects of erosion, wear, and particle
adhesion act to change the unbalance . Thus , rebalancing will
2 eventually be necessary on most machinery. Rotor parts such as
thrust collars , cooling fans and coupling hubs may be significant
FR EQU ENCY x R U N N I N G SPEED
sources of unbalance , but are often overlooked because of their
Figure 31 . Spectral C haracteristics of Subsy nchro nou s small size .
Instabilities.
Most common types of instabilities include :
• Fluidfilm bearing instability. S elf excited rotor vibration due

to fluid film bearing instability, known as oil whirl and oil whip,
have been an area of concern for a long time . This type of insta
bility can often be eliminated by a change in bearing configura
tion to more stable design . Bearing designs which have the high
est order of stability are tilting pad and squeeze film bearings .
• Floating ring seal instability. Floating ring seals can act in

1 2 3 4
a similar manner as journal bearings and thus potentially con
tribute to self excited vibration . The analog to a circular short FREQU ENCY x RUNNING SPEED
bearing is quite obvious and the corresponding theories allow Figure 32. Spectral Characteristics of Imbalance.
the stiffness and damping properties in the locked up condition
to be estimated . In lockup , floatin:g ring deals are held radially
by friction forces which do not allow them to follow the shaft mo Shaft bow is a type of unbalance caused by gravity sag or shaft
tion . Their destabilizing effect may be reduced by balancing the thermal gradients . This bow will "roll out" or change with time
floating ring seal axial force in such a way that the friction forces after the rotor is starte d . The application of turning gear and roll
are as small as possible . Another possibility is to optimize the out procedures, to minimize shaft bow is well known in the
clearance geometry in order to obtain favorable dynamic stiff industry.
ness properties . Shaft bow due to thermal distortion, which has been observed
on ID fans , is a more difficult problem to identify and solve than
• Labyrinth seal instability. One of the dominant self excita normal shaft bow. The root cause of the thermal distortion is an
tion mechanisms in high pressure compressors has been found apparent variation of thermal coefficient of expansion or Young's
to have its origin in the labyrinth seals . Labyrinth seal instability modulus throughout the shaft or impeller. A bimetallic strip ef
problems depend heavily on the labyrinth seal geometry and fect occurs as the rotor is heated, causing the shaft's center of
machine operating conditions; more specifically: labyrinth mass to be moved away from the center of rotation . Careful cor
radius , strip height, rotational speed, internal pressure s , Mach relation of vibration with shaft thermal growth is necessary to
number, and inlet swirl velocity. Generally, the best way to re identify this source of variable unbalance .
duce labyrinth seal instability is by suppressing the inlet swirl
velocity of the leakage flow entering the seal . Magnetic Unbalance
Magnetic unbalance in motors is an excitation source that is
• Aero dynamic w hirl instability. Aerodynamic whirl instabil often mistaken for mass unbalance . Forces due to magnetic
ity is related to lateral aerodynamic forces acting on the compres asymmetry usually rotate at electrical synchronous speed and
sor or turbine blade s . These forces are generated by varying tip will beat with mass unbalance on induction machines . A test to
clearances when the rotor is displaced from the center position . distinguish magnetic unbalance from mass unbalance is to ob
It has been determined that susceptibility of the rotor to this serve the real time vibration spectra as power to the machine is
type of instability increases with an increase in gas density and cut off. If a significant vibration reduction appears to occur in
rotational speed of the machine . stantaneously, then that portion was due to magnetic unbalance
and the remainder due to mass unbalance .
• Hysteretic instability. Hysteretic (frictional) instability usu Misalignment
ally occurs if rotors operate above the first critical speed and in
ternal forces that lag the bending displacements ar e allowed to Excessiv e misalignment of rotating elements driven through
occur. The source of excitation is normally the frictionally sup flexible couplings is usually indicated by a large second order
pressed movement in the shrink fits of wide discs or impellers vibration component (Figure 33) . Occasionally, large first order
shrunk to the shaft . To reduce this type of instability, the axial vibrations are also observed. High axial vibration is another indi
contact length of the shrink fit should be as short and as tight as cation that misalignment is likely. Shaft operating misalignment
possible without exceeding the yield strength of the material. is affected by relative thermal growth, static forces applied by
piping or condenser attachments , deterioration of support
Unbalance grouting, etc . Vibration due to first order misalignment can be
Unbalance is a common source of machinery vibrations . Un differentiated from unbalance by recording vibration in relation
balance is always characterized by radial vibration of the shaft to speed . Unbalance vibration will increase with speed squared,
TUTORIAL ON \'1BRATION MONITORING OF TURBOMACHINERY 117
while misalignment vibration "Will not change if resonances are

not involved . For machines which cannot be conveniently shut
y
down , it is recommended that a record be made of the vibration
spectrum when the machine is first started up and in good align
ment. This can be used for later comparison to determine if the
alignment of the machine is still satisfactory.
Figure 34. Shaft Orbit for Rub Induced Synchronous Vibration.
1 2 3 4
vibration sensor. In one example , a two bearing industrial gas
FREQUENCY x RUNNING SPEED turbine exhibited increasingly higher vibration measured by dis
placement probes during startup [ 19] . Initially, the gas turbine
Figure 33. Spectral Characteristics of Misalignment.
alarmed regularly and tripped occasionally. Inspection showed
that a significant number of compressor blades had moved
Radial Rubs axially in the third stage , causing rubbing against the adjacent
diaphragm . The fact that simple alarm and trip limits on vibra
Radial rubs between a rotor and close clearance stationary
tion level have detected blade migration suggests that more crit
components can cause damage to seals and blade tip s . Rubbing
ical trend monitoring of vibration vector changes (accounting for
is usually undesirable and sometimes catastrophic. The re
both amplitude and phase) should increase the detectability of
ported characteristics of radial rub induced vibration include
blade migration problems should they occur.
fixed rotor subharmonics , subsynchronous vibration at a natural
frequency, supersynchronous vibration at a natural frequency, Foreign Object D amage
time varying synchronous vibrations , and vibration at multiples
of running speed . Thus , it is difficult to make general statements Foreign object damage occurs when a loose object passes
about radial rubs , and the symptoms of rubbing are influenced through a machine. Small amounts of damage may go unde
by the materials in contact, the impedance to motion of the seal tected . Partial or complete loss of blade or blade distortion is
when contacted, location of natural frequencies , etc . Radial rubs likely to show up with sensitive vibration monitoring as a vector
of stainless steel rotors appear particularly damaging and are change in vibration or sometimes a trip . Again, close monitoring
known to cause rapid permanent rotor deformation . H igh per of vibration vectors should detect severe foreign object damage
formance steam turbine s , on the other hand, are designed with in the event that no other operational symptoms are detected.
very tight labyrinth clearances and are almost expected to rub
mildly during the early stages of commissioning. These tempo REFERENCES
rary rubs can cause localized heating and a temporary bow in the
l. Bently Nevada, "What The 3300 System I s , " Advanced
rotor, with associated high unbalance , particularly when the
Technical Information (Oct 1987) .
rotor is running below its first critical speed. The symptoms of
this particular rub induced vibration are a slow time variation of 2. Bently Nevada, "Computerized M achinery M onitoring
synchronous amplitude and phas e , and often a square like orbit System 64: Bently Nevada's Newest Computerized
(Figure 34) . Monitoring System: Up to 64 Monitor Racks with One
Computer, Orbit, pp . 10-16, (August �_}, __ _
Axial Rubs
3 . GenRad, " Structural Test Products ," Machinery Interac
Axial rubs between stationary and rotating components can tive Display and Analysis Software Gune 1987) .
occur when relative axial motion between the two is sufficient
to eliminate operating clearance . This problem is known to 4 . S cientific Atlanta, Chammp - Computerized Health
occur in large industrial gas turbines where the blades migrates Analysis for Machinery Malfunction Prediction (October
in their slots due to breakage or deterioration of the blade lock 1986) .
ing mechanism (the problem has occurred in both compressor 5 . S cientific Atlanta, "DYMAC M800 - New Technology for
and turbine sections) , and relaxation or creep of the vane or noz M achinery Monitoring/Protection" Guly 1987) .
zle under load at high temperature . The consequences can
range from progres sive gradual "machining" of the contacting 6 . M itchell, J . S . , An Introduction to Machinery Analysis and
parts to catastrophic failure [ 19] . Mon itoring, Tulsa, Oklahoma: PennWell B ooks (1981) .
There has been some limited success in detecting blade mi 7. Gilstrap , M . , "Shaft Versus Housing M easure ments , "
gration as a result of change in vibration at the bearing mounted Orbit, p p . 14- 16 Gune 1983) .
118 PROCEEDINGS OF THE NINETEENTH TURBO MACHINERY SYMPOSIU}.I
8. Lifson, A. and Smalley, A. J . , "Combustion Turbine Vibra 16. IRD M echanalysis , Audio-Visual Customer Training In
tion M otion, " E lectric Power Research Institute , Final Re struction M anual.
port AP-6119 (December 1988) . 17 . Ludwig, G . A . and E rdman, 0 . D . , "Gas Turbine \'ibration
9 . International S tandard I S O 2.372, "Mechanical Vibration of Limits-A Fundamental View," AS J\I E Paper 73-C:fR-48
M achines with Operating Speeds fi·om 10 to 200 rev/s-Basis (1973) .
for SpecifYing E valuation Standards , " 1st E d . , 1974-11-01, 18 . Blake , M . P. , "New Vibration Standards for Maintenance , "
Ref. No . ISO 2372- 1974(E) ( 1974) . Hydrocarbon Processing and Petroleum Refinery, 43, pp .
10. International S tandard I S O 394.5 , "Mechanical Vibration of 111- 114 , (January 1964) .
Large Rotating M achines with Speed Run from 10 to 200 19 . Lifson , A . , Quentin , G . H . , Smalley, A. J . , and Knauf, C .
rev/s-Measurement and Evaluation of Vibration Severity L . , "Assessment of Gas Turbine Vibration Monitoring,"
in Situ, " 1st E d . , 1977-03-1.5 , Ref. N o . I S O 394.5- 1977(E) Journal of Engineering for Gas Turbines and Power, 111,
(1977) . pp . 275-263 (1989) .
11. AGMA Standard , "Specification for M easurement of Lat 20 . Scalzo, A . J . , Allen, J . M . , and Anto s , R. J . , "Analysis and
eral Vibration on High Speed Helical and Herringbone Solution of a Nonsvnchronous Vibration Problem in the
Gear Units , " American Gear Manufacturers Association Last Row Turbine B lade of a Large Industrial Turbine , "
No . 426 . 01 (April 1974) . Journal for G a s Turbine and Power, 108 (October 198 6) .
12 . API-610, "Centrifugal Pumps for General Refinery Serv 21. Simmon s , H . R . , "A Non-Intrusive M ethod for Detecting
ice s , " 6th E d . , American Petroleum Institute (January HP Turbine Blade Resonance , " ASME Paper No. 86-
1982) . JPCG-Pwr-36 ( 1986) .
13 . API-617, "Centrifugal Compressors fiJr General Refinery 22 . Gabriel , F. K . , and Donato, V. , ''Telemetry M easurement
S ervice", 4th E d . (November 1979) . of Combustion Turbine Blade Vibration in a High Tempera
14 . API-673, "Special Purpose Centrifugal Fans f()l' General ture Environment," A S M E Paper No. 86-GT- 207 ( 1986) .
Refinery Services", 1st E d . (January 1982) . 23 . Simmon s , H . R . , and Smalley, A . J . , "Effective Tools for
15 . General Guidelines for Vibration, Dresser-Clark, Incorpo Diagnosing Elusive Turbomachinery Dynamics Problems
rated ( 1968) . in the Field, " A S M E Paper No. 89-GT-71 (1989) .

Vibration Monitoring of Turbomachinery by Alex Lifson: Section Engineer Chandler Evans West Hartford, Connecticut

Uploaded by

Copyright:

Available Formats

Vibration Monitoring of Turbomachinery by Alex Lifson: Section Engineer Chandler Evans West Hartford, Connecticut

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Vibration Monitoring of Turbomachinery by Alex Lifson: Section Engineer Chandler Evans West Hartford, Connecticut

Uploaded by

Copyright:

Available Formats

View metadata, citation and similar papers at core.ac.

uk brought to you by CORE

VIBRATION MONITORING OF TURBOMACHINERY

West Hartford, Connecticut

Southwest Research Institute

San Antonio, Texas

GENERAL VIBRATION CONCEPTS

• machine housing vibration,

• shaft vibration relative to hearing cap , or

• absolute shaft vibration

casing tiedown , foundation material failure , and thermal casing 4

frequency related malfunctions are best determined by a vibra­

\Vhere both V(shaft) and V(housing) are absolute motions of those

Using this definition , we can say that machines with high

new digital protective monitors is also reduced. • Bufferino � I

DATA Programmability • Alarm Driven

vibration monitoring. Additional information can be found in • ANALYSIS • Management of Ccnfllcts

• General Purpose Data Base lnt-ce

A Bode plot generally shows synchronous amplitude (and 180

ous amplitude vs phase , during startup or coastdown, in polar

Trend plots show variation in some measure of vibration as a

Figure 9. Polar Plot Showing Amplitude, Phase, and RPM Dur­

Figure 7. Unfiltered Orbit Display 1Jipical of Misalignment.

........... , ............... , 12.0

Figure 8 . B od!le Plot Showing Amplitude and Phase During o.o

of a portable data acquisition system is that they eliminate the

nance and downtime cost due to early problem identification,

The costs represent recent estimates and are presented as

Accelerometers-Measurements of Housing Vibration

A typical piezoelectric transducer consists of one or several

racy of 0 . 002 in . Inductive interference requires that the turbine

MORE COMPREHENSIVE VIBRATION

• Performance loss due to internal machine clearance rubs

ration should be used with a table of appropriate correction fac­

1l1ble 4. Vibratio n Service Fa ctors.

0.1 I I 1 11 1 I I I Il l I I J lll K, = 0.6 SOFT FO U N DATI O N M A C H I N ES

• There are no absolute limits that will assure successful

• An important factor to recognize in applying vibration limits

/ S U B S Y N C H R O N O U S FREQUENCY or casing which is in exact synchronization 'A-ith rotor speed (Fig­

Most common types of instabilities include :

• Fluidfilm bearing instability. S elf excited rotor vibration due

• Floating ring seal instability. Floating ring seals can act in

while misalignment vibration "Will not change if resonances are

Figure 34. Shaft Orbit for Rub Induced Synchronous Vibration.

You might also like

frequency related malfunctions are best determined by a vibra

Figure 9. Polar Plot Showing Amplitude, Phase, and RPM Dur

ration should be used with a table of appropriate correction fac

/ S U B S Y N C H R O N O U S FREQUENCY or casing which is in exact synchronization 'A-ith rotor speed (Fig