Professional Documents
Culture Documents
BN 1330
BN 1330
and Shock
Measurements
Brel & Kjr
Mechanical Vibration
and
Shock Measurements
by
Professor Je n s Trampe Broch
Dipl Ing. E T.H.
Revised by:
Jolle Cour re ch , Dr. Ing.
J o h n H assall, M . Sc., M .1.0.A.
P h ilip HoIHngbery, M .A ., M.I.E.E.
Torben L ic h t, M.Sc.
Hans M rs k -M e lle r, M .Sc.
R. B. Randall, B. Tech., B.A.
D avid R e d fe rn , H.N.C. Mech.
Hans Jorgen R in d o rf, M.Sc.
J o n a th a n Wort, M .S c ., M I.M ech. E.
April 1984
ISBN 87 87355 3 4
FOREWORD
The book "M echanical V ib ra tio n and Shock M e a su re m e n ts has long been
a m ainstay in Brel & K j r's series of books on th e fu n dam entals of physical
m easurem ent and analysis. A lth o u g h the basic th e o ry is s till va lid , a great
m any developm ents have occurred since the book w as last revised by Profes
sor Jens Trampe Broch in M ay 1 9 7 2 , and it w a s decided in 1 9 7 9 to bring
th e book fu lly up-to-date, p a rtly by a revision o f existing m a te ria l, and partly
by adding new chapters w h e re relevant. It w a s decided to keep as closely as
possible to Professor Trampe B roch's o rig in a l te xt, and he has had the oppor
tu n ity to approve the m o difications. Even so, it is probably as w e ll to make it
clear how m uch of the present book is taken from the o rig in a l, and how
m uch is com pletely new.
Chapter 2 w as revised by Bob Randall, and fo llo w s th e o rig in a l very
closely. Some m aterial has been added on d ig ita l frequency analysis, a field
w h ich has developed rapidly in the last fe w years.
Chapter 3 w as revised by Jolle Courrech and also fo llo w s th e original
quite closely. The discussion of n o n -lin e a r system s has been expanded som e
w h a t, and in addition to th e exact tre a tm e n t o f the vibration of continuous
stru ctures, as given by Professor Tram pe Broch, a brief in tro d u ctio n is given
to tw o approxim ate m ethods w h ic h have increased in im portance in recent
years, viz. fin ite elem ent te ch n iq u e s and sta tis tic a l energy analysis.
The origin al Chapter 4 has been sp lit into tw o , both revised by John Hassall. Chapter 4 covers th a t part of th e o riginal w h ic h considered th e effects of
vibrations and shock on m echanical system s, w h ile Chapter 5 discusses the
quite d iffe re n t topic of the effects of vib ra tio n and shock on m an (Human V i
bration). There have been included some new developm ents in both areas, for
example a discussion of m ean stress effects and acoustic fa tig u e in Chapter
4, and m ore detailed in fo rm a tio n on hand-arm vibration in C hapter 5, but o th
erw ise these sections are based on the o rig in a ls.
Chapter 6 on the selection of in s tru m e n ta tio n is com pletely n e w and was
w ritte n by David Redfern. Not o n ly w e re very fe w of the in s tru m e n ts de
scribed in the original book s till available, but th e re has also been a consider
CONTENTS
IN T R O D U C T IO N ......................................................................................................1 3
1 .1 .
C H A R A C T E R IS T IC S O F V IB R A TIO N A N D S H O C K ................................ 2 0
2.1 .
2 .2 .
2 .3 .
2 .4 .
2 .5 .
R E S P O N S E OF M E C H A N IC A L S Y S T E M S TO
V IB R A T IO N S A N D S H O C K S ............................................................................. 40
3.1 .
3 .2 .
3 .3 .
3 .4 .
3 .5 .
3 .6 .
3 .7 .
3 .8 .
3 .9 .
EFFECTS OF V IB R A T IO N S A ND S H O C K
ON M E C H A N IC A L S Y S TE M S ............................................................................. 7 2
4.1 .
4 .2 .
4 .3 .
5.
EFFECTS OF V IB R A T IO N S A N D S H O C K ON M A N .................................... 8 5
5 .1 .
5 .2 .
5 .3 .
6.
85
91
V IB R A T IO N M E A S U R IN G IN S T R U M E N T A T IO N
A N D T E C H N IQ U E S ............................................................................................... 9 7
6 .1 .
6 .2 .
7.
F R E Q U E N C Y A N A LY S IS OF V IB R A T IO N A N D SHO CK .....................1 6 0
7 .1 .
7 .2 .
7 .3 .
7 .4 .
7 .5 .
8.
V IB R A T IO N M E A S U R E M E N T S FOR M A C H IN E
HEALTH M O N IT O R IN G ...................................................................................... 19 7
8 .1 .
8 .2 .
8 .3 .
8 .4 .
8 .5 .
9.
A C O U S T IC E M IS S IO N
9 .1 .
9 .2 .
9 .3 .
9 .4 .
..................................................................................... 2 1 2
9 .5 .
9 .6 .
9 .7 .
9 .8 .
9 .9 .
10.
V IB R A T IO N A N D SH O C K T E S T IN G ...........................................................2 3 7
1 0 .1 .
1 0 .2 .
1 0 .3 .
11.
B A L A N C IN G OF R O TATIN G M A C H IN E S .................................................2 6 8
1 1 .1 .
1 1 .2 .
1 1 .3 .
11 .4 .
11 .5 .
11 .6 .
1 1.7.
INTRODUCTION ................................................................................... 2 6 8
UNBALANCE OF RIGID ROTORS ................................................... 2 6 8
ROTOR SUPPORT SYSTEM ............................................................ 2 7 0
SETTING THE STANDARDS ............................................................. 2 7 3
BALANCING MACHINES .................................................................... 2 7 6
FIELD BALANCING ............................................................................. 2 7 8
DIFFICULT BALANCING TASKS ...................................................... 2 8 5
1 1 .7 .1 . Fine Balancing .........................................................................2 8 5
1 1 .7 .2 . Flexible Rotors .........................................................................2 8 5
1 1 .7 .3 . Crankshaft B alancing ......................................................... 2 8 7
1 1 .7 .4 . M ultiple-S pan S hafts .........................................................2 8 8
1 1 . 8 . SELECTED BIBLIOGRAPHY ............................................................ 2 8 9
12.
F U N D A M E N T A L S OF SHO CK A N D V IB R A T IO N C O N TR O L ............2 9 2
1 2 .1 .
A P P E N D IC E S ................................................................................................................ 3 2 7
APPENDIX A
On th e S tatistical In te rp reta tio n of the R M S-Value .................... 3 2 8
APPENDIX B
Response Versus E xcitation C haracteristics fo r
Linear S in g le Degree-of-Freedom System s
............................. 3 3 0
APPENDIX C
On th e W ave-Shape D is to rtio n in
N on-Linear M echanical System s ................................................... 3 3 4
APPENDIX D
C onnection Between th e Fourier S pectrum of a
Shock Pulse and the Residual Shock S p e ctru m ........................ 3 3 7
APPENDIX E
E lectronic Integration o f Accelerom eter
O utput S ignals ...................................................................................... 3 3 9
APPENDIX F
On th e Use of Decibels ...................................................................... 3 4 5
APPENDIX G
C onversion Charts, Tables e tc ........................................................... 3 5 0
APPENDIX H
S tandards Related to V ib ra tio n and
Shock M easurem ents ......................................................................... 3 5 5
IN D E X
362
1. INTRODUCTION
M echanical vib ra tio n s and shock are dynamic phenom ena, i.e. th e ir in
tensity varies w ith tim e . Both th e m axim um in te n s ity , however, and the rate
of change in in te n sity w ith tim e , spread over w id e m easurem ent ranges and
often require h ighly specialized equipm ent for th e ir precise d e te rm in atio n .
Ground m otions caused by fa r-o ff earthquakes (or explosions) m ay, fo r in
stance, be barely detectable w h ile vibrations caused by large com bustion en
gines can cause severe m echanical fatigue damage.
A lthough in m ost cases m echanical shocks and vibrations are undesired
byproducts of o th e rw is e useful processes, and g reat efforts are s p e n t to re
duce th e ir effects, some vib ra tio n s are produced on purpose. Typical exam
ples are the vib ra tio n s produced by conveying and screening m achines, me
chanical ham m ers, ultrasonic c le a n in g baths, etc., w h ile desirable shock-effects are built into riveting ham m ers and pile-drivers.
As the same m ethods of d escription and m e asurem ent apply, in general,
w h e th e r the vib ra tio n s or shocks being characterized are w a n te d or un
w anted , no clear distin ctio n has been made th ro u g h o u t th is book. The various
chapters have been laid out w ith a view m ainly to describe m easurem ent
data and techniques necessary to characterize vib ra tio n s and shocks and to
evaluate th e ir effects on a responding m edium . For more com prehensive
treatm ents of the ore tica l aspects th e reader is re fe rre d to standard textbooks
and to literature cited in the bibliography.
Chapter 2 re vie w s briefly the basic characteristics of mechanical vibrations
and shocks and th e various q u a n titie s used to characterize them . S ection 2.1
deals w ith periodic (discrete frequency) signals w h ile section 2.2 extends the
discussion to station a ry random fu n c tio n s w h ich m u st be described in terms
of th e ir statistical param eters. In both cases it is seen how d e scription in
term s of RMS (Root Mean Square) values can be advantageous because of
th e ir relation to the pow er c o n te n t of the vibrations, and how the d is trib u tio n
of power w ith frequ e n cy can be determ ined using th e Fourier tra n s fo rm . Sec13
on mean stress e ffe cts, since m ost structural e le m e n ts have to bear a steady
load in addition to th e dynamic o n e , and another section deals b rie fly w ith
th e case of intense acoustic e x cita tio n w hich has been found to cause failure
of aerospace and oth e r stru ctu re s. Section 4 .2 discusses briefly th e dam ag
ing effects of shocks and tra n s ie n ts , w hich may give sudden b rittle fa ilu re as
opposed to fatigue fa ilu re .
Chapter 5 sum m arise s the m ost im portant effects o f vibration and shock on
m an, ranging fro m K inetosis (m o tio n sickness) in th e fractional H ertz range,
thro u g h Whole B ody Vibration in th e range 1 8 0 Hz, and in clu d in g Handa rm vibration in th e range 8 1 0 0 0 Hz. Some c rite ria are included from the
relevant standards.
Chapter 6 is a detailed guide to th e proper selection of equipm ent fo r vibra
tio n and shock m easurem ent and evaluation. A fte r a b rie f in tro d u ctio n in sec
tio n 6 .1 , sections 6 .2 and 6 .3 provide the necessary inform ation to a llo w the
optim um choice of accelerom eter and pream plifier, n o w w idely accepted as
being the best a ll-ro u n d vibration transducer, even w h e re param eters other
th a n acceleration are to be m easured. Sections 6 .4 , 6 .5 and 6.6 cover the
practical questions o f calibration, choice of force and impedance transducers,
and accelerom eter m ounting m e th o d , respectively.
Section 6.7 discusses the choice of portable battery-operated in stru m e n ts,
including vibration m eters, tu n a b le filte rs , graphic recorders, tape recorders,
transport shock or "b u m p " recorders as w ell as stroboscopes and w aveform
analyzers. The tape recorder w o u ld often be used fo r field recording of sig
nals w hich are to be analyzed in d e ta il on sophisticated m ains-operated equip
m ent as discussed in section 6 . 8 . The possibilities include frequency analysis
using either sw eeping or sequential filte r analyzers and digital re a l-tim e anal
yzers based on th e FFT (Fast F o u rie r Transform) or d ig ita l filte r p rin cip le s. Re
su lts can be w ritte n out g ra phically on an X-Y re co rd e r, stored d ig ita lly on a
dig ita l cassette recorder fo r later processing or processed on a large scale us
ing a desktop ca lcu la to r or com puter.
Chapter 7 gives specific in fo rm a tio n on how to o p tim ise the fre q u e n c y ana
lysis of vibration and shock sig n a ls, th is having a lready been show n to be the
m ost im portant analysis te ch n iq u e . Section 7.1 g ives an in tro d u ctio n to the
mode of operation of typical analyzer systems, both analogue and d ig ita l, and
to the choice of basic analysis param eters such as bandw idth and logarithm ic
vs. linear scales. Section 7.2 deals w ith the serial analysis of sta tio n a ry sig
nals, concentrating on w h a t is probably the most im p o rta n t rem aining applica
tio n area, viz., fie ld analysis using portable battery-operated e q u ipm e n t. Sec
tio n 7 . 3 , on the oth e r hand, discusses the use of real-tim e digital analyzers
fo r the more de tailed analysis of both stationary s ig n a ls and shocks and tra n
sients. Finally, section 7 .4 discusses the analysis o f non-stationary signals by
a m oving tim e -w in d o w approach, th is being useful fo r treating n o n -sta tio n a ri15
16
balancing of fle xible rotors (ru n n in g above the c ritic a l speed) and m u ltip le span shafts, and cra n ksh a ft balancing fo r reciprocating m achines.
The fin a l chapter. Chapter 12, discusses the re m a in in g methods fo r redu
cing th e transm ission of vibration and shocks, in cases w h e re it is not possi
ble to elim in ate th e m at th e source, and covers is o la tio n , dynamic absorption,
and dam ping.
The basic prin ciple of vibration is o la tio n consists in selecting a suitable
spring m ounting fo r the e quipm ent so that the n a tu ra l frequency of the
spring-m ass system is appreciably less than the lo w e s t frequency to be iso
lated, assuring at th e same tim e th a t th e resonance w ill not be excited. W ith
regard to shock is o la tio n , also tre a te d in section 1 2 . 1 , th is is treated on the
basis of the shock response sp e ctru m and it is fo u n d th a t sim ilar co nditions
apply. However, in th is case the choice of proper d a m ping is more im p o rta n t,
as are the non -lin e a r properties of th e m ounts.
Section 12.2.1 discusses the a p p lication of d ynam ic absorbers w h ic h can
be used to detune a resonance w h e re there is one m a in excitation frequency.
In com bination w ith appropriate d a m p in g , a dynam ic absorber can also be de
signed to have a lim ite d response o ver a wide fre q u e n cy range (in clu d in g the
resonances) and s till attenuate at h ig h frequencies.
Section 1 2 .2 .2 discusses the a p plication of da m p in g treatm ents in cases
w h e re it is im possible to avoid e xcitin g resonances because of th e ir num ber
and close spacing. A guide is given as to the o p tim u m w a y to apply dam ping
m aterial so as to obtain the best re su lts from a g ive n am ount of tre a tm e n t,
and fin a lly there is a discussion of th e various m ethods available fo r d e te rm in
ing th e dam ping properties of m a te ria ls.
1 .1 . G EN ER A L R EFER EN C ES
Books:
BISHOP, R E D. and
JOHNSON, D.C.:
CREMER, L. and
HECKL, M.:
K rperschall.
Springer V e rla g .
B e rlin /H e id e lb e rg
/N e w York 1 9 6 7 . Also E nglish version S tru ctu re Borne S ound, 1 973
Den HRTOG, J P :
17
HARRIS, C M. and
CREDE, C.E.:
JACOBSEN, L .S . and
AYRE, R.S.:
MORROW, C.T.:
MORSE, P.M.:
Vibration a n d Sound.
Inc 1 9 4 8
SNOWDON, J.C.:
TIMOSHENKO, S.:
MEIROVITCH, L.:
M c G ra w -H ill Book
W ile y and
M c G ra w -H ill
Book
Journals:
A cu stica
A kusticheskH Z h u rn a t
E xperim ental
M e chanics
J.A .S .A .
J o u rn a l o f S ound
a n d Vibration
J o u rn a l o f the
A co u s tic a l Society o f
Japan
18
K a m p f dem Lrm
Verlag
f r
angewandte
Baden-Baden
Noise Control
E ngineering
W issenschaften
GmbH.
19
2.
2 .1 . P E R IO D IC V IB R A T IO N
Periodic vibration may be looked upon as an o s c illa tin g motion o f a particle,
or body, about a reference po sitio n, the motion repeating itself exa ctly after
certain periods o f tim e . The s im p le s t form of pe rio d ic vibration is th e socalled harm onic m o tio n w hich w h e n plotted as a fu n c tio n of tim e , is repre
sented by a sinu so id a l curve, F ig .2 .1 . Here T is th e period of v ib ra tio n , i.e.
th e tim e elapsed betw een tw o successive, exactly equal conditions o f m otion.
20
If the vib ration has the fo rm of a pure tra n sla tio n a l o s c illa tio n along one
axis (x) only, the insta n ta n e o u s displacem ent of the particle (or body) from
the reference position can be m a th e m a tica lly described by m eans of the equa
tion:
x = Xpeak sin
[2n
(2.2)
w here
at = 2 n f = a ngular fre q u e n cy
Xpeak = M axim um d isp la ce m e n t from th e reference p o sitio n
t = tim e
As the velocity of a m oving p article (or body) is the tim e ra te of change of
the displacem ent, the m otion can also be described in term s o f ve lo city (v):
dx
l/ = =
dt
CD
(2 .3 )
Finally, the acceleration (a) of the motion is the tim e rate o f change of the
velocity:
a =
_ = CO2 Xpeak s in (cot) A p eak s in (cot) = Apeak s in ( c o t + n)
dt
a t2
(2 .4 )
From the above equations it can be seen th a t the form and period of vib ra
tio n rem ain the same w h e th e r it is the displacem ent, the v e lo c ity or the accel
eration th a t is being studied. H ow ever, the ve lo city leads th e displacem ent by
a phase angle of 9 0 ( n / 2 ) and th e acceleration again leads th e velocity by a
phase angle of 9 0 ( n / 2 ) . A s characterizing values for th e m agnitude the
peak values have been used, i.e. Xpeak Vpeak and Apeak . The m agnitude de
scription in te rm s of peak v a lu e s is quite u se fu l as long as p u re harm onic v i
bration is considered because it applies d ire c tly in the equ a tio n s given above.
If, on the oth e r hand, m ore com plex v ib ra tio n s are being stu d ie d other de
scriptive q u a n titie s may be p re fe rre d . One o f th e reasons fo r th is is that the
peak value describes the v ib ra tio n in term s o f a quantity w h ic h depends only
upon an instantaneous v ib ra tio n m agnitude regardless o f th e tim e history
producing it.
A fu rth e r descriptive q u a n tity , w hich does take the tim e h is to ry into ac
count, is the average absolute va lu e , defined as (see also F ig .2 .2):
21
Even tho ugh this q u a n tity takes into account the tim e h isto ry of the v ib ra
tion over one period [T) it has been found to be of lim ite d practical interest. A
much m ore useful d e scrip tive quantity w h ic h also takes th e tim e history in to
account, is the RMS (root m e a n square) va lu e (Fig.2.2):
X rm s = f / ~
I *x 2( t ) d t
(2 .5 )
X rm
~ 2 '\ / 2
veras e
^ y ~ 2 ^ P eak
X RMS ~
Ff
X / \ verag e
Xpeak
'c
or
22
Y
a RMS
A A verage
Y
''P e a k
ARMS
(2 '6 >
The fa cto rs Ff and Fc are called "fo rm -fa c to r" and "c re s t-fa c to r", respec
tively, and give some in d ic a tio n of the w aveshape of the vib ra tio n s being s tu d
ied.
For pure harm onic m otion:
and
Fc = \ 2 = 1 .4 1 4 (= 3 dB)
M ost of the vibrations encountered in d a ily life are not pure harm onic m o
tions even though m any o f them may be characterized as periodic. A typical
non -harm onic periodic m o tio n is show n in F ig .2.3 (piston acceleration of a
com bustion engine). By dete rm in in g the Peak, Average A b so lu te and RM Svalue of th is vibration as w e ll as the fo rm -fa c to r and cre s t-fa c to r a lot of use
fu l in fo rm a tio n is obtained, and it can be c le a rly concluded th a t the m otion is
not harm onic. However, it w ill be pra ctica lly impossible, on th e basis of th is
in fo rm a tio n, to predict all th e various effects th a t the vib ra tio n m ight produce
in connected structural e lem ents. O ther m ethods of d e scrip tio n must th e re
fore be used.
(2 .7 )
23
I
f, (= )
T,
a)
f 2 *= ; H
T2
b )
Frequency, f
271261
2 .2 . S T A T IO N A R Y R A N D O M V IB R A T IO N
Random vibra tio n s are m et rather fre q u e n tly in nature and m ay be charac
terized as vibratory processes in w hich the vib ra tin g particles undergo irregu
lar m otion cycles that never repeat them selves exactly, see F ig .2 .7 . To obtain
a com plete description of th e vibrations, an in fin ite ly long tim e record is thus
th e o re tica lly necessary. This is of course an im possible re q u ire m e n t, and f i
nite tim e records w ould have to be used in practice. Even so, if the tim e re
cord becomes too long it w ill also become a very inconvenient means of de
scription and other methods have therefore been devised and are com m only
used. These m ethods have th e ir origin in s ta tis tic a l m echanics and com m uni
cation th eory and involve concepts such as a m p litu d e p ro b a b ility d istributions
and p rob ability densities, and continuous v ib ra tio n frequency spectra in term s
of mean square spectral d e n s itie s *.
M ean sq ua re sp ectra l density is a lso o fte n te rm e d " P o w e r Spectral D e n s ity " (P .S .D .) because
th e m ean sq ua re is a q u a n tity p ro p o rtio n a l to pow e r.
26
? A t n
= A t, + A t 2 + A t 3 + A t 4
271262
Fig. 2 .8 S ketch illu s tra tin g th e concepts o f p ro b a b ility and p ro b a b ility density
In the study of co n tin u o u s processes such as stationary* random vib ra
tions it is often convenient to use the concept of probability d e n sity instead of
probability. Physically the p ro b a b ility density can be defined as the probability
of finding instantaneous a m p litu d e values w ith in a certain a m p litu d e interval,
A x, divided by the size of th a t interval (thus: density), see F ig .2 . 8 . This
means th a t w h ile pro b a b ilitie s are d im e n sio n le ss q uantities th e probability
density is a q u a n tity having a certain d im e n sio n .
M a them atica lly fo rm u la te d th e probability d e n sity at some specified a m p li
tude level, x, is:
( 2 .8)
'O
Here p(x) designates the probability density w h ile P(x) is th e probability th a t
any instantaneous am p litu d e value exceeds th e level x and P(x + A x) is the
probability of occurrence of instantaneous a m p litu d e values exceeding the le
vel x+ A x. By plotting the value of p(x) for all values of x a p ro b a b ility density
curve is obtained w hich has th e feature th a t an integration of th e curve from
a value x) to a value x 2 im m ediately te lls th e probability of occurrence of in
stantaneous am plitude values w ith in the in te rv a l (x2 X j), independent of
the actual m agnitude of x ? and x2 The p re se n ta tio n of experim ental probabil
ity data in te rm s of p robability density curves bears some advantages because
it allow s fo r a direct com parison of data betw e e n experim ents (and between
experim enters) independent of the w id th of th e am plitude in te rv a l, A x, used in
the experim ent. Finally, th e o re tica l p ro b a b ility data are co m m o n ly presented
S ta tio n a ry ra nd o m v ib ra tio n s a re d e fin e d as ra nd o m v ib ra tio n s w hose s ta tis tic a l c h ara cteristics
do not ch a n g e w ith tim e .
27
Hm -1 \ 2' f ( t ) f ( t + z )d t
T-> co '
(2 .9 )
271263
Amplitude
fit)
fit, +
t)
f( 'l>
tim e
t! + r
269034
F ig .2. 10. Basic concepts involved in d e riv in g the a u to co rre la tio n fu n c tio n
In th e case of an "d e a l stationary random process (w h ite noise) th e a u to
co rre la tio n function w o u ld consist o f an in fin ite ly n a rro w im p u lse -fu n ctio n
around zero ( t = 0), see F ig.2.11 a), as in such a process each instantaneous
am plitu de value should be com pletely independent of a ll other instantaneous
am plitu de values.
H ow ever, in practice th e a u to co rre la tio n functions associated w ith s ta tio n
ary random vibrations clu ste r around 7 = 0, but are never "in fin ite ly n a rro w "
im p u lse -fu n ctio n s, F ig .2 .1 1 b) and c). The reason fo r th is spreading o u t of
the curve around zero is that all p ra ctica l random processes are frequency li
m ited, and the n a rro w e r the fre q u e n cy lim its the m ore spread-out are the
corresponding autocorre la tio n fu n c tio n s (because the rate at w hich a signal
can change from its c u rre n t value is m uch more lim ited).
From the autoco rrela tio n function a n o th e r, very im p o rta n t function in p ra c
tice, can be deduced, w h ic h has a c e rta in resem blance to the Fourier fre
quency spectra described in section 2.1 for periodic v ib ra tio n s. This fu n c tio n
has been term ed th e m ean square s p e c tra l density fu n c tio n (power spectral
den sity functio n) and can be derived fro m the auto co rre la tio n fun ctio n as fo l
lows: A ssum ing th a t th e integral o f ifi(r) from oo to +oo is fin ite (see
F ig.2.1 1) one can w rite :
S (f) = |
y / ( r ) e - / 2nfx dz
(2 . 10)
w h e re f is frequency.
From th e theory of Fourier in te g ra ls it is fu rth e rm o re know n that ip (r) can
also be found from th e above integral by inversion:
( 2. 1
1)
29
fu n ctio n s
fo r an id e a l stationary ra n d o m process
0 to (co n sta n t spectral density)
for a "p ra c tic a l" wide b a n d stationary
for a n a rro w band s ta tio n a ry random
The Fourier integral re la tio n s between iJj( t ) and S(f) are o fte n called the
W ie n e r-K h in ch in relations and play a very im p o rta n t role in th e theory of
random processes.
In physically realizable sta tio n a ry processes one operates w ith positive fre
quencies o n ly* and tp(r) = t//( r) w hereby the in te g ra l for iJj( t) becomes:
W {r) = 2 | ^ S ( / ) cos { 2 n f x ) d f
.0
then
(/(r) =
fo r f> 0
G (f) cos (2 n fz ) d f
(2 .1 2 )
30
y /(o ) = Hm | 2
+ o )d t =
T-+aoT
and
lim I 2 f 2( t ) d t
T )- 2
V '(o ) = | G ( f ) d f
'o
T
th u s
Hm ~ I 2T f 2(t) d t = I G ( f ) d f
r^ c c T J -L
Jo
(2 .1 3 )
|o G (f ) d f =
i',+ B
G ( 0 d f = lim -
/ f)
!
| \ fB ( t ) dt
Here fB (t) is th e above-m entioned part of the com plete signal, f(t), w hich
has frequency com ponents w ith in B. If now B is m ade so small th a t G (f) can
be considered constant w ith in th is frequency range then
\ ' +BG ( f ) d f = G ( f) B
G ( f) =
Hm
B ^O
hm
2 f
oo D ' '-L
(t) d t
(2 .1 4 )
31
2 .3 . T R A N S IE N T P H E N O M E N A A N D SHOCKS
T ransient phenom ena and m echanical shocks are, like random v ib ra tio n s
encountered re latively often in d a ily life . They may o rig in a te from such w id e ly
d iffe re n t releases of energy as ro ugh handling of equipm ent, explosions and
supersonic m otion. However, com m on fo r this type of energy release is its
sh o rt duration and sudden occurrence.
A sim ple shock m ay be defined as a transm ission o f kinetic energy to a sys
tem w h ich takes place in a re la tive ly s h o rt tim e co m pared w ith the n a tu ra l p e
rio d o f o scillation o f the system , w h ile transient phenom ena (also te rm e d
com plex shocks) m ay last for se ve ra l periods o f vib ra tio n o f the system .
33
Shocks and tra n s ie n t vib ra tio n s may be described in term s of fo rce , acceler
atio n, velocity or displacem ent and for a co m p le te description it is necessary
to obtain an exact tim e history record of the q u a n tity in question.
In many cases th e ultim ate goal is not th e w a ve fo rm itself, but rather a
means to e stim a te the effect th a t the corresponding shock or tra n s ie n t vibra
tio n w ould have on a certain m echanical system . A more useful m ethod of de
scription m ig ht th e n again be fou n d in the fo rm of Fourier analysis. If the
tim e function fo r a shock is f(t) th e n its Fourier tra n s fo rm is given by:
F ( f) = I
f( t) e ~ '2n,td t
(2.15)
GO
The analogy betw e e n this expression and th e m ean square spectral density
fu n ctio n of s ta tio n a ry random vibrations (E quation (2.10)) is re a d ily seen.
There is, h ow ever, a very d is tin c t difference in th a t th e mean square spectral
density fu n ctio n fo r stationary random vib ra tio n s is the Fourier tra n sfo rm of
an already tim e-averaged, even fu n ctio n , th e a utocorrelation fu n c tio n , w ith
th e dim ensions o f am plitude squared. In the above Fourier in te g ra l fo r tra n
sient or shock fu n c tio n s the fu n c tio n f(t) its e lf m u st be tim e -lim ite d and has
th e dim ensions o f am plitude o n ly. Because it in general is not an even fu n c
tio n , its Fourier tra n sfo rm w ill be complex, but it is found that th e square of
th e am plitude o f th e Fourier tra n s fo rm at each fre q u e n cy gives a m easure of
th e energy d is trib u tio n of the tra n s ie n t.
It may be u sefu l to see h ow th is difference in dim ensions in flu e n ce s the
u n its in a p a rtic u la r case. A s s u m in g that sig na l am plitude is expressed in
volts (V), then th e a utocorrelation function fo r a stationary random signal
w o u ld have u n its o f volts squared (V2 ) or p o w e r. The Fourier tra n s fo rm of
th is has the u n its V 2 s, or V 2 / Hz, i.e. pow er per un it frequency or pow er
sp ectral density. A shock or tra n s ie n t fu n ctio n , how ever, has u n its of Volts
and its Fourier tra n sfo rm (am plitude) units of Volt-seconds (Vs). The am pli
tu d e squared th u s has units V 2 s 2 or V 2 s /H z , i.e . energy per u n it frequency
or energy sp e ctra l density. A tra n s ie n t of course has fin ite energy w h e n inte
grated over all tim e , w h ile a s ta tio n a ry random sig n a l w ould have in fin ite en
ergy, though fin ite pow er.
M ost analyzers assume a signal is continuous and give a result in term s of
pow er per analysis bandw idth. The conversion o f th is to the co rre ct units is
discussed in C hapter 7 and in m ore detail in th e B & K book "F re q u e n cy A na
ly s is ".
In Fig.2 .1 3 va rio u s shock tim e fu n ctio n s and th e am plitudes of th e ir Fou
rie r spectra are g ive n . It is seen fro m the figure th a t in general a shock pulse
contains energy spread over all frequencies from zero to in fin ity, and th a t the
spectra are co n tin u o u s w ith no discre te frequency com ponents.
34
In the case o f tra n sie n t phenom ena the s itu a tio n is som ew hat different.
Such phenom ena, in the sense used in th is book, may consist e ith e r of a
single period ''sh o c k -w a v e '', or o f an o scillating tra n sie n t. The F o u rie r spec
tru m fu nction of a typical o s c illa tin g transient is sh o w n in F ig .2 .1 4 and it is
seen that the m agnitude of th e spectrum co m p o n e n ts in th is special case
tends tow ards zero as the fre q u e n cy f goes to zero. Also, a m a xim u m m agni
tude of the sp e ctru m is reached around f0 w h ic h corresponds ro u g h ly to the
frequency of o scilla tio n of the tra n s ie n t. This m a xim u m is broader th e quicker
th e tran sient phenom enon ceases.
tion
If the tra n s ie n t" does not cease at all, i.e. w h e n the "tr a n s ie n t" is no
longer a tra n sie n t but a periodic phenom enon (in th is case a h a rm o n ic vibra
tio n), the freque ncy spectrum degenerates into a discrete spectral lin e (in fin
itely narrow m a xim u m at f0).
2 .4 . N O N -S T A T IO N A R Y R A N D O M V IB R A T IO N
T heoretically all kinds of random vibrations e n co u n te re d in practice are nonsta tion ary because th e ir statistical properties va ry w ith tim e. H ow ever, from
an engineering p o in t of view th is va ria tio n in s ta tis tic a l properties m ay be so
slo w , or of such a character, th a t m any of the phenom ena studied can be con
sidered stationary in a practical sense.
N on-stat/onary ra n d o m vibrations m ay therefore, in practice, be d e fin e d as
ran dom vibrations w hose s ta tis tic a l properties vary w ith time w ith in tim e in
tervals considered e sse n tia l fo r th e ir proper d e scrip tio n . To analyze and de
scribe such vib ra tio n data it is th u s necessary to ta ke th e ir variation in sta tisti
cal properties w ith tim e into a cco u n t. A typical exam ple of seriously non-sta36
tio n a ry random vib ra tio n s is the vib ra tio n s induced in space vehicles d u rin g
launch and re-entry.
To th eoretica lly analyze n o n -sta tio n a ry random v ib ra tio n s properly it is n e
cessary to introduce th e concept of ensem ble averaging. A n ensemble a ve r
age is an average taken over a large nu m b e r (an ensem ble) of repeated e x p e r
im ents, see Fig.2 .1 5 . A s can be seen fro m the fig u re an ensemble average
can be taken at any p a rtic u la r instant o f tim e t j , t2 , t3 e tc ., and when th e av
erage values are plotted against tim e a more or less com plete description of
the vib ration is obtained. There are, on th e other hand, several reasons w h y
th is m ethod of description is not very useful in practice. Firstly, it re q u ire s
th a t th e n on-statio na ry process can be repeated a ve ry large n u m b e r of
tim es. In the case of space vehicle la u n ch and re -e n try fo r instance th is is
not possible due to th e cost of such experim ents. S econdly, the a m o u n t of
data necessary for a th o ro u g h description is so large th a t th e ir proper m e a s
urem ent w ill pose serious problems.
It is therefore n o rm a lly necessary to seek other m ethods o f description, and
in general some sort o f tim e averaging is used. There are, how ever, ce rta in li
m ita tio n s imposed upon th is kind of tim e averaging in th a t the response and
averaging tim e of the m easurem ent e quipm ent em ployed should pre fe ra b ly
2 .5 . SELECTED B IB LIO G R A P H Y
ANSI:
38
DIN:
DIN 1 3 1 1 . (February 1 9 70) S ch w in g u n g sle h re K in em atische B e g riffe . B e u th-V ertrieb GmbH, B e rlin .
IEEE:
KHARKEVICH, A .A .:
MORROW , C.T.:
PIERSOL, A.G.:
RICE, S.O.:
RUZICKA, J.E.:
THRALL, G P. and
BENDAT, J.S.:
39
3.
3 .1 . R E S P O N S E OF LIN EA R M E C H A N IC A L S Y S T E M S
TO V IB R A T IO N S . R E S O N A N C E
The m otion of a m echanical system subjected to external forces is com
m only term ed th e response o f th e system to th e pa rticu la r forces in question.
S im ila rly, th e external forces a cting upon th e system are te rm e d the exciting
forces, or sim p ly the e xcita tio n . These te rm s are general and have to be speci
fied closer w h e n the behaviour of a pa rticu la r system is being investigated.
To aid such specifications it is norm ally necessary to co n s tru c t a som ew hat
sim plified m echanical model and, on the basis of the m odel, to form ulate the
equations of m otion fo r the system . This m odel can then be used as a basis
fo r a fu rth e r analysis.
One of the sim plest m odels of a vibrating system is show n in Fig.3 .1 .a),
and consists of a mass, a sp rin g and a dam per. If the system behaves linearly
40
(and tim e -in va ria n t) th e equation of fre e m otion of th e m ass is (no e xternal
force applied):
(3 .1 )
(3 .3 )
41
6 {x )d x = 1
e>0 J - e
* ( 0 = | f
h {T )f(t-T )d T
(3.4)
oo
w h e re
j = f - z
X (f) = (
x ( t ) e ~ /2nf'd t = I
e~l2 n ,td t\
f(r)h(t-r)dr
J - 00
00
* 00
= i"
-
dt f
00
h { t - T)e~J2nhf ( T ) d x
J GO
J - 00
h(,)e~i2 n f i d I 00 f { i ) e ~ i 2l t,Td i
= I
J 00
J - 00
Thus:
X(f) = H(f)F(f)
(3 .5 )
The Fourier tra n sfo rm of the response is the product of the Fourier tra n s
form of th e excitation and the frequency response fu n c tio n .
Note th a t in th is case the so lu tio n is built up fro m a superposition of
steady state responses to simple h a rm o n ic excitations.
The com plex fre q u e n cy response fu n c tio n H(f) of th e system s h o w n in
Fig 3 1 is found sim ply by solving the equation of m o tio n for an a rb itra ry Fou
rie r com ponent, F0 e i2 n ,t:
m d2 l + c + k x = F0 e,2nft
d t2
dt
(3.6)
X( f ) = H { f ) F 0 e '2n,t
Here
( ] = ---------
\fj
(3 .7 )
(3-8)
Q f0
43
f0
J -
2n
Q = ^ Vkm is a m easure of th e system 's dam ping. Q is norm ally c a lle d the qu
a lity factor of th e system and th e larger the va lu e of Q the s m a lle r is the
dam ping. For a com pletely undam ped system Q = co w hile fo r a critically
damped system Q = 1/ 2. A n approxim ate m easure of Q is o b ta in e d in prac
tice by m easuring the w id th of th e response cu rve , \H(f)\, at th e ha lf power
points, see F ig.3 .3 .
[ | H ( f ) \ = V 1/ 2 \ H ( f 0) |2 = 0 ,7 0 7 \H ( fo)^\
\ H { f )\2 = ! 2 \ H ( f 0) \ 2
If th is w id th is A f then
(3 .1 0 )
1X
Frequency
Fig. 3 .5 . Exam ple o f a m u lti degree-of-freedom system (three degrees-of-freedom) a n d its frequency response fu n c tio n
3 .2 . S O M E R E S P O N S E P R O PER TIES OF N O N -L IN E A R S Y S T E M S
In the previous section som e im portant response cha ra cte ristics of linear
system s have been discussed, in p a rticu la r th e ir so-called frequency
response fu n c tio n s . These fu n c tio n s can be derived m a th e m a tica lly from the
linea r d iffe re n tia l equation of m otion for the system . In the case of a single
degree of freedom system th is equation was given as
46
(3 .1 1 )
w here
m
Because this d iffe re n tia l equation is no longer lin e a r the p rin cip le o f su p e r
p o sitio n cannot be ap plied and th e derivations o u tlin e d in section 3.1 are
th ere fore no longer valid. In p rin c ip le each p a rtic u la r non-linear vib ra tio n
problem has to be solved on its o w n . However, ce rta in general properties
w h ic h are of considerable practical in te re st may be discussed w ith o u t actu
ally solving the equatio n .
Consider fo r instance the case w h e re only th e "S tiffn e s s -te rm is non
linear. The m otion of th e mass m is th e n governed by th e equation.
m ~ ^ + c ^ - + F{ x) = f { t )
dt2
dt
(3 .1 2 )
(3 .1 3 )
47
th e n
relation
can be obtained
by integrating
Equation
v + F( x ) = 0
dx
+
^
F{x)dx = O
Ja
U(x) =
then
dx
dt
F( ) d
(3 .1 4 )
= v = \/2[U{a) ~U(x)]
i'd
t (a) = 4
(3 .1 5 )
Jo V 2 [ U ( a ) - U ( x ) ]
W h e n lig ht dam ping and periodic e x cita tio n are added to the system , the
steady-state response is generally p e rio d ic, w ith the sam e period as th e e x c i
ta tio n unless there is subharm onic resonance, as discussed later. The w ave
shape of th e response o scillation is in general diffe re n t fro m that of the e x c i
ta tio n and also d iffe re n t from the free vib ra tio n wave shapes. M oreover, the
response waveshape changes w ith the level of the e xcita tio n .
For fixed excitation am plitude and lig h t dam ping, th e response curves have
the fo rm s show n in F ig s.3 .8 and 3 .9 . They are like the corresponding curves
fo r lin e a r system s but th e "backbones of th e resonant peaks are the n o n -lin
ear fre e vibration a m p litu d e -fre q u e n cy re la tio n s shown in F ig .3.7.
A,
A i
1
11
L.
-------- - f
A
/v
//iVV
i
- f
263382
nents harm onically related to the fre q u e n cy of the d rivin g force. This can be
confirm ed m athe m a tica lly, for in sta n ce , by a pproxim ating the so lution to the
n o n -lin e a r d iffe re n tia l equation by m eans of a series expansion, and e xp e ri
m e n ta lly by analyzing th e response w ave shape by m eans of a fre q u e n cy a n
alyzer.
Under certain circum stances (very lo w dam ping) a very special phe n o
m enon occurs in n o n -lin e a r resonant systems of th e type described above.
This is the phenom enon of subharm onics. A su b h a rm on ic is a response v ib ra
tio n occurring at 1 / 2 , 1 /3 , 1 /4 , 1 / 5 etc. of th e frequency of th e d riving
force. A n intuitive explanation fo r th e occurrence o f subharm onics may be
given in th a t the d riv in g force supplies energy to one of the harm onics o f the
n o n -lin e a r system and w hen energy is supplied it w ill start to oscilla te . The
higher harm onic th e n pulls all the o th e r harm onics w ith it, as the sp e cifica lly
excited harm onic is an integral part o f th e w hole m otion.
W h ile the occurrence of subharm onics in practice is relatively rare, " o rd i
n a ry " harm onics (som etim es called superharm onics) are present to a greater
or lesser extent in all non-linear system s. Even if th e ir am plitude values are
ra th e r sm all they m ay play an im p o rta n t role w h e n th e vibration of com plex
(m ulti-degre e-of-fre e d o m ) m echanical system s is considered. A n exam ple
readily illustrates th is statem ent:
C onsider for in sta n ce the case w h e re a n o n -lin e a r spring e le m e n t in a
m u lti-d e g re e -o f-fre e d o m system produces a th ird ha rm o n ic of the order of
1%. If th e frequency o f th is h a rm onic by chance coincides w ith the resonant
freq uency of anoth e r resonance in th e system w h ic h happens to have a re
sonance a m plifica tio n factor Q = 1 0 0 th is specific resonance w ill respond
w ith th e same am p litu d e as the a ctu a lly excited fre q u e n cy even though its fre
quency did not exist in th e w ave-shape of the driving force!
A n o th e r im portant case*, that w ill not be explained in detail is w h e n the
n o n -lin e a rity is velocity-dependent o n ly, i.e. the e q u a tio n of m otion fo r the
system can be w ritte n :
(3 .1 6 )
52
3 .3 .
R O T A T IO N A L A N D T O R S IO N A L V IB R A T IO N S
I ~
+ c ' ~ + l ' ,(P =
d t2
dt
(3.1 7)
+k' <p = M = F r
(3 .1 8 )
53
a)
b)
271313
F ig.3 .1 2 . M odel o f a n u n sym m e trica lly loaded fo u n d a tio n , and the co rre
sponding e q u a tio n s o f m otion
54
3 .4 . R E S P O N S E OF M E C H A N IC A L S Y S T E M S TO S T A T IO N A R Y
R A N D O M V IB R A T IO N S
In section 3.1 it w as show n that th e response of any lin e a r system to a p re
scribed excitation can be determ ined fro m a know ledge o f the system 's im
pulse response fu n c tio n or its com plex frequency response function. If n ow
the e xcitation consists o f a Gaussian random process characterized by m eans
of its auto correlation fu n c tio n (or, w h a t m ay be more com m on in practice, by
m eans of its mean square spectral d e n s ity function) w h a t w ould the re la tio n
ship betw een the excita tio n and the response then be?
S ta rtin g w ith the a u to-correlation fu n c tio n re p resentation of the response
th is can be form ulated in term s of the syste m 's im pulse response fu n ctio n :
( ( t ) x ( t + r) = i
f ( t - Ti)h(T,)dT, (
J an
f ( t + t - r 2) h ( r 2) d r 2
J 00
f{t- t ,)f(t + T -
T2) h ( T i ) h ( T 2 ) d i d r 2
Ij
x( t)x( t + t ) dt
'~2
Urn
f ( t - T i ) f ( t + r - r 2) h ( T : ) h ( r 2)dT
dt
rJ-i
th e n
Vx ( t ) =
y/f ( t + t , - T2) h ( r - l ) h ( T 2) d x : d r g
(3 .1 9 )
f ( t + t - r 2) - f ( t - r, + t - t 2 + t ,)
(3 .2 0 )
55
or wh at is the same:
and inserting th e form ula given above for i/j x( t ) in to this expression:
as
Sf ( f ) =
J - oc
r
J -
thus:
m i x
T, -
T2 ) e i 2 K n T * x' - ^ d x
00
Sx (f) = H * (f ) H (f) Sf (f ) = | / y ( / ) | 2 ^ ( 0
(3 21)
3 .5 . S H O C K R ESPO N SE A N D SHO CK S PE C TR A
We have defined earlier a shock as a tra n sm issio n of k in e tic energy to a
system, the tra nsitio n ta kin g place in a re la tiv e ly short tim e compared w ith
th e natural period of o scilla tio n of the system . A rectangular shock pulse may
therefore co n stitu te one or tw o shocks depending upon th e na tu ra l period o f
oscillation o f the system in flu e nce d by it and the duration o f the pulse. This
is illustrated in Fig.3.13.1) and II), and it sh o u ld be noted th a t the m axim um
system response may in th e case of Fig.3.1 3.1) reach a va lu e w h ic h is tw ic e
the m agnitude of the shock pulse.
b)
___________________________________ 271275
response being calculated in the fre q u e n cy dom ain, the inverse Fourier tra n s
fo rm a tio n operation produces the response tim e history o f th e system.
The second spectral decom position, w h ic h has proved to be of considerable
value w ith respect to th e com parison of shock m otions, to th e design of e q u ip
m ent to w ith s ta n d shocks, and to the fo rm u la tio n of laboratory tests as a
means to sim ulate e n viro n m e n ta l co n d itio n s , is the shock response s p e c
trum , or b rie fly the shock spectrum (Note: This is not to be confused w ith th e
Fourier spectrum of the fo rcin g shock p ulse!). The shock spectrum is obtained
by le ttin g the shock pulse in question be applied to a se rie s of linear, u n
damped single degree-of-freedom system s and plotting fo r exam ple the m a x i
m um response of the system as a fu n c tio n o f the system 's n a tu ra l frequency.
V ario us types of shock spectra are used depending upon th e intended a p p li
cation of the in fo rm a tio n obtained. These m ay be the in itia l shock spectrum
w h ic h is obtained from the m axim um response w h ile th e shock pulse is s till
acting, or the residual shock spectrum w h ic h is obtained fro m the m axim um
response a fter the pulse has occurred.
O ther d e fin itio n s may be the overall or m axim ax spectrum w hich is p lotted
on the basis of the m axim um response w ith o u t regard to tim e , and the o v e r
all negative m axim um shock spectrum w h ic h is obtained by considering th e
m axim um response of the single degree-of-freedom system in the negative d i
rection.
In practical m easurem ent systems the re q u ire m e n t of zero damping in th e
responding single degree-of-freedom system m ay be d iffic u lt to achieve, but a
selectable dam ping gives th e advantage th a t the effect o f th e shock on v a r
ious stru ctu re s can be judged sim ply by considering th e in flu e n ce of da m p
ing. H ow ever, fo r re la tive ly sm all am ounts o f damping th e shock spectra w ill
not be esse ntially d iffe re n t from the spectra obtained w ith no damping, since
the response for the firs t fe w cycles w ill be v irtu a lly id e n tic a l. The response
of an undam ped single degree-of-freedom system to a shock pulse can be c a l
culated re lative ly easily fo r sim ple shock w a ve form s, using fo r example La
place tra n sfo rm m ethods. F igs.3 .1 4 , 3 .1 5 and 3 .1 6 illu s tra te shock spectra
obtained fo r a rectangular, saw tooth and h a lfsin e shock pulse of duration T.
The m axim ax shock spectra are found s im p ly by taking th e hig h est of the tw o
spectrum values at any frequency.
Recently, th e shock spectrum concept has been extended to m ultiple d e
gree o f freedom system s, n o n -lin e a r system s etc. See the bibliography at th e
end of th e chapter.
It may be w o rth w h ile in connection w ith th e discussion o f shock spectra to
58
(3.22)
3 .6 . V IB R A T IO N S IN S TR U C T U R E S . M E C H A N IC A L W AVES
The m echanical system s considered in the preceding text have been of the
so-called idealized lum ped param eter type, i.e. masses have been assum ed to
be rigid bodies w h e re all points w ith in the body m ove in phase, and elastic
e le m ents have been assumed to have no mass. In practice all masses have a
ce rta in elasticity and all spring e le m e n ts have m asses. For instance a beam
or a plate is a c o n tin u o u s com bination of masses and springs.
A s the num ber of degrees-of-freedom of a m echanical system w a s defined
as the num ber of ela stic m ovem ents o f masses (resonances, see se ctio n 3 . 1 )
it fo llo w s that s tru ctu re s like beams and plates have an in fin ite n u m b e r o f de
grees-of-freedom . The in fin ite num ber o f "re s o n a n c e s " resulting from th e in fi
n ite num ber of degrees-of-freedom are in the case o f structures n o rm a lly
te rm e d "n a tu ra l m o d e s", or sim ply "m o d e s ".
W h ile in lumped param eter system s all points w ith in a mass are supposed
to move in phase, th is is no longer tru e for c o n tin u o u s structures. A sim ple
exam ple illu strates th is clearly: In F ig .3 .1 7 the v ib ra tio n of a beam clam ped
at one end and acted upon by an o scillating force in the other is show n.
W h e n the frequency o f th e o scillating force coincides w ith one of th e beam 's
60
61
269019
. Beam
YA
/
'A
a)
Beam
P
vr
"
\JT
b)
271280
g S ~
d t2
+~ ( E I
dx2 \
w here
p = M ass d e n sity of the beam m aterial
62
)=0
dx2J
( 3.23)
A . /E l
= ----- r
2-n VPSI
E = Y o u n g 's m odulus
I = Area m o m e n t o f inertia o f beam cross
section
I = Length o f beam
p = Mass d e n s ity o f beam material
S = Area o f cross-section
A = C o e ffic ie n t from table below
271281
+2
+ + /2g (/~ l!l^ ff = o
3x 4
dx2 dy2
dyA
Eh2
d t2
(3 24)
d x2
1st Mode
2nd M ode
(3 .2 5 )
E d t2
3rd Mode
4th Mode
6th Mode
ton/J D/pha4
Nodal lines
D/pha4
////////
///////,
////<
///////,
Nodal lines
V /////Z
j J D/pha4
Nodal lines
777777/
'777777/
- 27rfn
D = E h3/ 1 2 (1 - u 2 )
' 7777^7?
/
% > /////
271282
Fig. 3 .2 2 . S ketch sho w in g th e derivation o f th e equation o f m o tio n for com p re ssio n a l (lo n g itu d in a l) vibrations in a beam
This equation has the same form as the so-called w ave -e q u a tio n w hich go v
erns various kinds of wave phenom ena in th e o re tica l physics. Compressional
vibrations are therefo re also often referred to as m e ch a n ica l waves w ith a
w ave velocity (v) of:
-V 2
T - TE- :
'
i f 1 Q'
13 261
The m ain reason fo r the in clu sio n of th is b rie f section on stru ctu ra l v ib ra
tions in the book has been to illustrate th a t the actual vib ra tio n s m easured
on a com plica ted co n stru ctio n may be w id e ly d iffe re n t fro m p o in t to p o in t
even i f the m easurin g p o in ts considered are situ a te d a re la tiv e ly short d is
tance apart from each other. Also, the d ire c tio n in space o f the vibrations
may vary and fo r thorough investigations it is therefore necessary to m easure
the vib rations both as a fu n c tio n of fre q u e n c y and as a fu n c tio n of space
coordinates at each m easuring point.
3 .7 . S H O C K A N D V IB R A T IO N A N A L Y S IS U S IN G FINITE
E LE M E N T T E C H N IQ U E S
A more com prehensive th e o re tica l tre a tm e n t of stru ctu ra l vibration is o u t
side the scope of th is book and can be fo u n d in many e xce lle n t textbooks on
the subject. W e w ill how ever discuss th e fin ite elem ent m ethod for shock
and vibration analysis w h ic h is becoming m ore and more used due to the ad
vancem ent in sophisticated digital com puters. Several fin ite element p ro
gram s are c u rre n tly available from various sources. W h ile w e have until n o w
only discussed th e exact so lu tio n of the d yn a m ic problem s, th e fin ite e lem ent
method is an approxim ate so lu tio n.
65
There are tw o fin ite e le m e n t techniques available: the m a trix displacem ent
o r s tiffn e ss m e th o d w h e re th e displacem ents are chosen as unknow ns, and
th e m a trix force or fle x ib ility m ethod w h e re forces are unknow ns. In both
cases, the continu u m is divided into a fin ite num ber of ele m e n ts connected
only at the node points. Each discrete e le m e n t is idealised. The d isp la ce
m ents of these points are assumed and th e complete s o lu tio n is obtained
com bining displacem ents so th a t it sa tis fie s force e q u ilib riu m and d isp la ce
m ent co m p a tib ility at th e jo in ts of th e e lem ents. The m a trix equations are
th e n solved n um e rica lly by a com puter. The m atrix displacem ent m ethod is
c u rre n tly the most popular and is the one w e w ill discuss in th is section.
S tatic analysis
The equation is of the fo rm ,
[F ] = [K ] [ U ]
(3 .2 7 )
k -k ~
-k
~u:
k_ P 2-
(Wi - u 2)
F 2 = *1 (u 2 ~ u ,) + k 2 ( u 2 - u 3)
= - /r, w, + (k , + k 2) u 2 - k 2u 3
F2
66
^2 ( U3 ~ u t )
uT
f
k,
=
-k ,
- /r,
/c-]
CO
O'
k2
- k2
k2
[,K ]
-*1
k.
k^
Thus
[K ] =
k2
- k2
k2
N>
Jr-
-k ,
[K ] =
k^
-k 2
* i.
k2
NJ
(2)
-k ;
=
N
J
(1 )
[K ]
1_______ 1
H ow ever, the stiffne ss m atrix can also be assembled fro m the stiffn e ss of
the individ ual elem ents
O'
k ^
k^^ k2
k2
kn
k?
D ynam ic Analysis
The dynam ic equation of motion can be w ritte n in m a trix form as,
[M ] [X ] + [C ] [ * ] + [K ] [X ] = [ F ( t ) l
(3 .2 8 )
w here
[M ] = mass m atrix
[C] = dam ping matrix
[K\ = stiffn e ss m atrix
[ f ( 0 ] = external force vector
W e see th a t tw o ad dition a l term s, the m ass and the dam p in g , are needed.
The mass m atrix can be found in d iffe re n t ways
67
m/ 2
m
. 0
m /2 .
or at three nodes
m /3
m l3
m/3__
_0
68
The fin ite elem ent m ethod is in constant e vo lu tion and new er and better
elem ents are c o n tin u a lly being developed. Its im p o rta n t place is due both to
its ease of application and its u n ive rsa lity in solving both stru ctu ra l and nonstru ctu ra l problem s such as heat tra n s fe r, flu id flo w etc.
3 .8 . S TA T IS T IC A L E N E R G Y A N A L Y S IS
It is possible at relatively high frequencies to get reasonably accurate
predictions of the m ean response by averaging over space as w ell as over fre
quency bands w h ic h include m any modes, instead o f considering each p a rti
cu la r mode at one tim e . This m ethod, known as s ta tis tic a l energy analysis
(SEA), is explained in more detail by Lyon in the reference given at th e end
of th e chapter. It circu m ve n ts problem s like boundary conditions (w h ic h are
not know n anyway) and exact shape of the individual elem ents. It is based on
the average over th e stru ctu ra l area, a panel, for exam ple, and the average
over a frequency band, such as 1 / 3 octave. It assum es a know ledge of the
modal density n(f) i.e. the num ber of modes per u n it frequency, defined as
( 0 = ^
(3 .2 9 )
3 .9 . SELECTED B IB LIO G R A P H Y
BENDAT, J.S . and
PiERSOL, A .G .:
BOURGINE, A.:
Sur une A pproche S ta tis tiq u e de la D yn a m iq u e Vibratoire des S tructures. O ffice National d 'E tudes et de
Recherches Aerospatiales 1 9 7 3
BROCH, J.T.:
CRANDALL, S .H . and
M ARK W .D.:
Acad
K AR M AN , T. and
BIOT, M .A.:
KITTELSEN.K.E.:
LYON, R.H.:
MINORSKY, N.:
In tro d u c tio n to N o n -L in e a r
Bros., A n n A rb o r, M ich. 1 9 4 7
MORROW , C.T.:
MORROW , C.T.:
MORSE, P.M.:
V ibration a n d Sound.
Inc. 1 9 4 8
70
o f Shock.
M echanics.
Bruel &
Edwards
PILKEY, W . and
PILKEY, B.:
ROBSON, J.D .:
SALTER, J.P.:
Steady S ta te Vibration,
England 1 9 6 9
SESHADRI, T.V.:
SESHADRI, T V.:
VIGNESS, I.:
YOUNG, D.:
ZIENKIEWICZ, U.C.:
Kenneth M ason
Book Co.,
71
4.
4 .1 . D A M A G IN G EFFECTS OF V IB R A T IO N S . M E C H A N IC A L FA TIG U E
Even though m echanical failure due to m aterial fa tig u e is by far th e most
com m only know n d e te rio ratin g e ffe ct o f vibrations, a vibrating m echanical
co n stru ctio n may fa il in practice fo r o th e r reasons as w e ll. Failure may, fo r in
stance, be caused by th e occurrence o f one, or a fe w , excessive vib ra tio n am
plitudes (brittle m a te ria ls, c o n ta c t-fa ilu re s in relays and sw itches, co llisio n s
betw een tw o vib ra tin g system s etc.), or by the fa c t th a t a certain vib ra tio n
am p litu d e value is exceeded for too g re a t a time.
72
w here
x = crack length
N = n u m b e r of stress reversals
c, m, n = co n sta n ts dependent upon the m aterial properties
(a reasonable assum ption seems in m a n y cases to be
m = 2 , n - 1)
er = relative strain
73
Log N
271315
Fig. 4 .2 . Typical histogram obtained fro m fa tig u e e xperim ents (after B loom er
a n d Roy lance)
the fa tig u e life of notched a lu m in iu m specim ens. The re s u lts show n w ere ob
tained from tests at a s in g le vibration stress level.
By m aking sim ila r tests at a num ber of vib ra tio n stress levels, a curve, co m
m only term ed the S-N cu rve (W hler-K urve) can be obta in e d , w hich show s
the re la tio n sh ip betw een th e average n u m b e r o f stress re ve rsa ls to fa ilu re
a n d the vibration stress level.
The actual S -N curve fo r a m aterial does not only depend upon the v ib ra
tio n a l effects, but is also affected by fa cto rs such as te m p e ra tu re , atm o s
pheric cond itions (corrosion effects), p re -tre a tm e n t of th e m aterial, etc.
Fig.4 .3 illu stra te s a set o f S -N curves va lid fo r 4 3 4 0 steel under norm al a t
m ospheric conditions. The curves show n are based on p u re harm onic v ib ra
tion loading only. In practice, how ever, a m echanical part, or m aterial, is very
rarely, if ever, subjected to pure harm onic vibrations of c o n sta n t m axim um
am plitude during its com plete " life " .
In order to take varying am plitudes in to account in th e o re tic a l estim ations
of th e average fatigue life , a " ru le " of lin e a r accum ulation of damage has
been suggested (Palmgren, M inor):
(4.3)
it is som etim es possible to establish a closed m athem atical form ula for D.
Two co nditions w hich have to be fu lfille d w h e n use is to be m ade of the fo r
m ulae fo r D and the S -N curve are, how ever, th a t each stre ss reversal has
an approxim ately sinusoidal w ave-shape and th a t the mean stress is zero.
These co n d ition s are fu lfille d , fo r instance, by th e vibrational stresses o ccu r
ring in a lin e a r single degree-of-freedom system excited by random vib ra
tions, see F ig.4 .4 .
F urtherm ore, in such system s the sta tistica l d istrib u tio n of m axim um v ib ra
tion am plitudes (peaks) can be m athem atically described by m eans of the socalled R a yleig h -d istrib u tio n , F ig.4.5:
X 2
p ( x ) d x = ^ e 2 2 dx
o 2-
( 4 .4 )
w here p(x)dx is the p ro b a b ility of occurrence of peaks w ith in the in fin ite ly
75
Fig. 4 .4 . Illu stra tio n o f th e s tre ss-ve rsu s-tim e trace p ro d u c e d in a sin g le degree-of-freedom system excited b y random vib ra tio n s
Fig. 4 .5 . Typical peak p ro b a b ility density curve for n a rro w b a n d random vib ra
tio n s (Rayleigh d istrib u tio n )
sm all am plitu de " w in d o w " dx (Figs.4 .4 and 4.5). As th e total num ber of
peaks occurring w ith in dx is n(x) = f0 . T . p(x)dx, and th e p a rtia l fatigue dam
age caused by these stress reversals aro u n d the vibration level x is
D x = n ^ l = fo . T P W d x
N (x )
0
N (x )
the accum ulated damage over all vib ra tio n peak levels d u rin g the period of
tim e , T, is:
76
x . n { x ) __
N (x )
|' p ( x ) d x
0 Jo
N (x )
,------(4.5)
f0 ( V 2 o) b r ( i + |)
F ig .4 .6 . Typical stress-ve rsu s-tim e trace at a " c ritic a l p o in t" on a com plicated
stru ctu re
77
take the sig n ific a n t peaks in to consideration, agree more closely w ith prac
tic e . U nfo rtunate ly, much of th is is proprietary in fo rm a tio n and n o t generally
available in th e open lite ra tu re at th e tim e of w ritin g .
C ycle s, N
800791
4 .1 .2 . Acoustic Fatigue
A rather d iffe re n t type of fa tig u e fa ilu re has becom e of p a rticu la r interest in
(Sv), tons in
Amplitude
Fluctuating
F iq .4 .9 . Typical v a ria b le /m e a n
(A STM)
80
stre ss
curves
fo r
unno tch e d
800790
specim ens
Further im portant areas w here acoustic fatigue is a problem are a ir-m o vin g
equipm ent, pow er s ta tio n generating equipm ent, and nu cle ar power plant.
The problem of random load fa tig u e has been stu d ie d quite extensively
over the past 1 5 to 2 0 years, and interested readers are referred to th e se
lected bibliography cited at the end of th e Chapter, in pa rticu la r to th e Brel
& Kj r Technical R eview s No. 1 1 9 6 8 and No.4 1 9 6 8 (included in "S e
lected R eprints").
4 .2 . D A M A G IN G EFFECTS OF S H O C K A N D T R A N S IE N T S
A s discussed e a rlie r in sections 2 .3 and 3 .5 , a shock is characterised by a
sudden, not alw ays expected, occurrence, and a s h o rt d uration in re la tio n to
th e natural frequency of th e system on w hich the shock is acting. A tra n
sien t, however, may last fo r a tim e w h ic h corresponds to several cycles at
th e natural frequency o f th e system. W hatever p a ram eter is used to describe
the rapid transfer of energy w hich c o n stitu te s a shock i.e. force, acceleration,
velocity, or displacem ent, it is alw ays necessary to o b ta in a tim e h is to ry of
th e shock in the u n its chosen. This in fo rm a tio n is re q u ire d , not as a re s u lt in
itself, but as a m eans of estim ating th e effect that th e shock has on a p a rtic u
lar system . The e n g in e e r is more inte re ste d in the response of his system to
th e application of a shock, e.g. from rough handling, explosions, im pact, c o lli
sion, etc. than he is in th e shock its e lf. The length, rise tim e, and specific
shape of the shock p ulse, all have an influence on th e ultim a te response of
th e stru ctu re in q u e stio n . It is for th is reason that th e concept of the shock re
sponse spectrum has been w idely used to compare shock m otions, to fo rm u
late laboratory tests, and to design eq u ipm e n t to w ith s ta n d shock inputs. The
m otion of the system on w h ich the shock acts is dependent upon th e fre
quency of both the shock excitation and th e natural fre q u e n c y of the system .
M echanical shocks are of particular im portance in th e design and operation
of m achinery and s tru c tu re s , because instantaneous in p u t levels are fre
q u e n tly an order of m agnitude greater than steady sta te levels. Shock and
tra n s ie n t analysis are n o rm a lly concerned w ith the design of the system to
survive the e n viro n m e n t, or w ith the atte n u a tio n of th e in p u t to the system
by means of packaging or isolators to
reduce the dam aging effect of the
shock input. Damage caused by shock
loading w ill u su a lly fall into fa ilu re
categories b) and c) in F ig .4 .1 . Category b) may be considered a form
of " fa
tig u e " fa ilu re for w h ic h o n ly shocks are counted and fo r w h ic h the life can be
defined sim ply as the nu m b e r of shocks to failure. C ategory c) im plies in sta n t
catastrophic fa ilu re as soon as a 0 is exceeded, i.e. the design shock occurs.
Depending on the a p p lica tio n , tw o d iffe re n t m eanings can be attached to the
w o rd survival:
82
1.
2.
th e system su ffe rs no dete rio ratio n of perform ance eith e r during or after
th e shock or series of shocks.
4 .3 . SELECTED B IB LIO G R A PH Y
ASTM
STP 3 3 8 1 9 6 3
BLOOMER, N T. and
ROYLANCE, T.F.:
BROCH, J.T.:
FORSYTH, P.J.E.:
HARRIS, C .M ., and
CREDE, C.E. (eds.):
Blackie and
Book
A pplied
83
M ADAYAG, A .E . Ed.:
McLEAN D.:
M e c h a n ic a l Properties o f M etals. J o h n
Sons, 1 9 6 2
MINER, M .A .:
C u m u la tive Damage in
M ech. V ol. 1 2 1 9 4 5
NATO AGARD:
NATO AGARD:
PALMGREN, A.:
'Die Lebensdauer
1 9 2 4 , 3 3 9 41
SANDOR, B l.:
von
Fatigue.
W iley and
J o u rn a l
K ugellagern',
VDI
of Appl.
V o l. 68 ,
WEIBULL, W.:
YEH, L.:
Journal:
84
The
Naval Re
5.
5 .1 . W H O L E -B O D Y V IB R A T IO N
The h u m a n body is both physically and biologically a "s y s te m " of an e x
trem e ly com plex nature. W hen looked upon as a m echanical system it c o n
tains a num ber of lin e a r as w ell as n o n -lin e a r "e le m e n ts ", and the m e ch a n i
cal properties are quite d iffe re n t from person to person.
B iologically the situ a tio n is by no m eans sim pler, e specially when psycho
logical e ffects are included. In considering the response o f man to v ib ra tio n s
and shocks it is necessary, however, to take into a cco u n t both m echanical
and psychological effects.
Because experim ents w ith human b eings are d iffic u lt, tim e -co n su m in g and
in extrem e cases une sth e tica l, much o f th e knowledge gained to date has
been ob tained from experim ents on a n im a ls. It is, of course, not always po ssi
ble to " s c a le " results obtained from a n im a l experim ents to reactions expected
from m an, but nevertheless such e xp e rim e n ts often re s u lt in valuable in fo r
m ation.
As th e purpose of th is Chapter is m ore to review some o f th e present k n o w
ledge th a n to discuss p a rticu la r e xp e rim e n ts in detail, th e follow ing pages
contain a brief description of some m a jo r facts w hich are no w known about
m an's response to vib ra tio n s and shocks.
C onsidering firs t the hum an body as a m echanical "s y s te m " it may, at lo w
frequ encies and low vib ra tio n levels, be roughly approxim ated by a lin e a r
lum ped param eter system of the type s h o w n in Fig.5 .1 . O ne of the m ost im
portant "p a rts " of th is system w ith respect to vibration and shock effect seem s
to be th e part marked "thorax-abdom en system . This is due to a d is tin c t re
sonance effect occurring in the 3 6 Hz range as indicated on Figs.5 .2 and
5.3 and w h ic h makes e ffic ie n t vibration isolation of a s ittin g or standing p e r
son very d iffic u lt. A fu rth e r resonance e ffe c t is found in th e 2 0 to 3 0 Hz re
gion and is caused by th e head-neck-shoulder system. F ig .5 .3 .
85
10
20
30
40
Frequency, H z
271285
Fig. 5 .2 . T ransm issibility o f vertical v ib ra tio n from table to various parts o f the
body o f a sta n d in g hum an s u b je c t as a fu n c tio n o f frequency. (A fte r
D ieckm ann; data for tra n s m is s io n to belt, a fte r Ftadke)
From a shock and vib ra tio n point o f v ie w the low fre q u e n cy range m ay be
considered most im p o rta n t. Some ve ry interesting m easurem ents have here
been made by von Bekesy and concern th e attenuation o f vibration along the
hum an body. In F ig .5 .4 the re su lts obtained at 5 0 Hz are reproduced and
show th a t the a ttenu a tio n from fo o t to head is of th e order of 30 dB. S im i
larly, th e attenuation fro m hand to head is roughly 4 0 dB.
A p a rt from the m echanical responses mentioned above, both physiological
and psychological effects are observed. A lth o u g h these e ffe cts are ra th e r c o m
plex and d ifficu lt to measure, it seem s th a t physiological results obtained
fro m anim al experim ents also apply to m an to a ce rta in extent. These e xp e ri
m ents have been m ostly concerned w ith changes in food assim ilation, m u s c u
lar a ctivity, reproductive activity etc. as w e ll as actual in te rn a l injury.
Psychological effects such as p e rce p tio n , discom fort, and pain, have re
cently been studied in some detail. M o s t of the studies have been carried out
on ve hicle drivers and aircraft p ilo ts , w hose ability to perform com plex tasks
under adverse environm ental c o n d itio n s , including vib ra tio n , is p a rtic u la rly
87
Frequency Hz
270159
m /s 2
Fig. 5 .5 . Vertical vib ra tio n exposure crite ria curves d e fin in g equal fa tig u e -d e
creased p ro fic ie n c y boundaries
th e task but the curves indicate th e general range fo r onset of such in te rfe r
ence and the tim e dependency observed.
A n upper b o u n d to exposure considered
w e ll as perform ance) is taken to be tw ic e as
decreased p ro ficie n cy boundary s h o w n in
fo rt bo undary" is assum ed to be a b o u t one
levels.
m/s 2
0,5
0,8
1,25
2,0
3,15
5,0
8,0
12,5
20
31,5
50
80
790538
Fig. 5 .6 . Lateral vibration exposure crite ria curves d e fin in g equal fa tig u e -d e c
reased p ro ficie n c y boundaries
lar m otion (i.e. in ya w , roll and pitch) whose ce n tre s of rotation lie at a dis
tance from the poin t of application to the body, can be adequately a pproxi
m ated by purely tra n sla to ry m otion. However, th e re is much c u rre n t w ork
aim ed at the eventual d efinition of crite ria curves s im ila r to those of F ig s.5.5
and 5 .6 fo r tra n sla to ry vibration.
Finally, F ig.5 .7 indicates the to le ra n ce of hum an subjects to s in g le shock
acceleration pulses of the type produced in the flo o r near drop forges or s im i
lar equipm ent (results from a single study).
5 .2 . H A N D -A R M V IB R A T IO N
H and-arm vib ra tio n is the second large problem area where tra n sm issio n
into the hum an body is concerned. It is, h o w e ver, rather d iffe re n t from
w ho le-body vibration in the type o f problem s to w h ic h it gives rise. W hereas
vib ra tio n tra n sm itte d into the sta n d in g or seated body norm ally gives rise to
problem s of a general nature e.g. m otion sickness, discom fort, reduced-w ork91
Maximum peak-to-peak
displacement of pulse, cm
92
0, 050
0,040
0,030
0,020
0,015
>
>
0,010
0,008
0,006
0,005
0,004
0,003
5 .3 . SELECTED B IB LIO G R A PH Y
AGATE, J.N . and
DRUETT, H .A.:
AXELSSON, SVEN A K E .A nalysis o f V ibrations in Power Saws. Studia Forestalia Suecica. N r.5 9 1 9 6 8
ALLEN, G.R.:
COERMANN, R. et al
COERMANN, R.:
94
DIECKMANN, D.:
DUPUIS, H.,
HARTUNG, E. and
HAM MER, W .:
HARRIS, C M . and
CREDE, C.E. (Eds.):
I S O. 2 6 3 1 -1 9 7 8 :
M IW A, T.:
RATHBONE, T.C.:
REIHER, H. und
MEISTER, F.J.:
95
TELFORD, E D.,
McCANN, M B. and
MacCORM ACK, D.H.:
TEMPEST, W . (Ed.):
VDI 2 0 5 6 :
VDI 2 0 5 7 :
C o n fe r
A spects
D undee
volum e
96
6.
VIBRATION
MEASURING INSTRUMENTATION
AND TECHNIQUES
6 .1 . G E N E R A L M E A S U R E M E N T C O N S ID E R A T IO N S
As indicated p reviously in this book th e re are three q u a n titie s w h ic h are of
in te re st in vibration stu d ie s, the vib ra to ry displacem ent, velocity and a cce le ra
tion (peak or RMS values). These th re e quantities are related sim ply, as de
scribed in Chapter 2. If the phase re la tio n sh ip s b e tw e e n the three p a ra m e
ters are neglected as is always the case w hen making tim e-average m e a su re
m ents, then the ve lo city at a given fre q u e n cy can be obtained by d ivid in g the
acceleration by a fa cto r proportional to th e frequency, and the d isplacem ent
can be obtained by divid in g the a cceleration level by a factor p roportional to
the frequency squared. This operation is perform ed in electronic m easuring
in stru m e n ts by an inte g ra tio n process. The re la tio n s h ip between th e three
param eters as a fu n c tio n of fre q u e n cy is shown in F ig .6.1. Note th a t the
axes are logarithm ic.
The firs t vib ra tio n pickups producing an e le c tric a l output w e re rather bulky
velocity sensitive devices. D u rin g th e last fe w ye a rs there has been a marked
move tow ards th e use of a cceleration sensitive transducers, c a lle d acceler
om eters Reasons for this tra n s fe r of preference are that accelerom eters are
generally m uch sm aller ph ysica lly than ve lo city pick-ups and th a t their fre
quency and d yna m ic ranges are s ig n ifica n tly w id e r, even after in te g ra tio n to
velocity. A w id e r dynam ic and frequency range is a prime re q u ire m e n t of the
modern vibration pickup, p a rtic u la rly to cater fo r th e growing in te re s t in high
frequency vib ra tio n as a ca rrie r o f inform ation on th e running c o n d itio n of m a
chinery and th e corresponding w id e range of vib ra tio n levels to be detected.
A n additional fa cto r w hich u n d e rlin e s the b e n e fits of accelerom eters is the
fact that an acceleration signal can be easily and validly in tegrated e lectroni
cally to obtain velocity and displacem ent w h e re a s electronic d iffe re n tia tio n
used w ith velocity and d isplacem ent transducers is a more co m p le x and dubi
ous affair
In theory it is irrelevant w h ic h o f the three param eters, acce le ra tio n , veloc
ity or displacem ent are chosen to measure v ib ra tio n . If one p lo tte d a n a rro w
band frequency analysis of a vib ra tio n signal in te rm s of the th re e parame
ters, they w o u ld all show the sam e frequency com ponents but w o u ld have d if
fe re n t average slopes as seen in F ig.6.1. It can be seen th a t displacem ent
m easurem ents give low fre q u e n cy com ponents m ost w eight and conversely
acceleration m easurem ents w e ig h t the high frequency com ponents. This
leads to a practical consideration th a t can in flu e n c e the choice o f param eter.
It is advantageous to select th e param eter w h ic h gives the fla tte s t frequency
spectrum in order to best u tilis e th e dynamic range of the m e a su rin g in stru
m entation.
The nature o f mechanical system s is such th a t appreciable displacem ents
o nly occur at lo w frequencies, th e re fo re d isplacem ent m easurem ents are of li
m ited value in th e general study o f m echanical vib ra tio n s. D isplacem ent is o f
ten used as an ind ica tor of u nbalance in rotating m achine parts because rela
tive ly large displacem ents u s u a lly occur at th e shaft rotation frequency,
w h ic h is the frequ e n cy of prim ary interest fo r ba la ncin g purposes.
RMS velocity m easurem ents are w id e ly used fo r vibration 's e v e rity meas
urem ents. This is due to the fa ct th a t vibratory v e lo c ity is simply re la te d to vib
ratory energy and is therefore a m easure of th e destructive e ffe c t of vibra
tio n . A given velocity level also signifies c o n sta n t stress for geom etrically
s im ila r co nstru ction s vibrating in th e same mode.
Because acceleration m easurem ents are w e ig h te d towards h igh frequency
vibration com ponents, th is p a ram eter is preferred w h e re the fre q u e n c y range
of interest covers high frequencies.
98
V ib ra tio n pickups of sm all size and w e ig h t are desirable because the pickup
should load the s tru ctu ra l member on w h ic h it is m ounted as little as p o ssi
ble and because it is often required to m easure vibration at a point on a s tru c
ture rathe r than over an area.
The frequency range of interest in vib ra tio n m easurem ents has been in c re a
sing steadily over the past tw o or th re e decades. Today m any vibration m e a s
urem ents are carried o u t up to 10 kHz, and often higher. The increased in te r
est in higher frequencies has been prom pted by the developm ent of h ig h
speed m achinery and th e recognition th a t high fre q u e n cy vibrations ca rry va
luable inform a tion about the condition of ro llin g e le m e n t (ball, roller, needle)
bearings, gear teeth, turb o m a ch in e ry blading etc.
The vibration associated w ith flu id flo w , je t noise, ca vita tio n etc. is e sse n
tia lly random in nature and must often be measured e ith e r alone or to g e th e r
w ith periodic vibration com ponents. This again calls fo r more com plicated
m easurem ent te chn iqu e s than was com m on in earlier days.
To be able to predict the effects of vib ra tio n on m echanical structures and
m an, it is not norm a lly su fficie n t to m easure the overall vibration level over
the freq uency range of interest. Frequency analysis is necessary to reveal th e
ind ividua l frequency com ponents m aking up the w id e band signal. For th is
purpose a filte r is contained in or attached to the vib ra tio n m easuring in s tru
m ent, th u s making a frequency analyzer. The filte r a llo w s only frequency c o m
ponents to be m easured w hich are contained in a sp e cific frequency band.
The pass band of the filte r is moved sequentially over th e w hole fre q u e n cy
range of interest so th a t a separate vib ra tio n level reading can be obtained at
each frequency.
The filte r section can consist of a nu m b e r of individual contiguous fixed fr e
quency filte rs w h ich are scanned se q u e n tia lly or a tu n a b le filte r w hich can be
tuned contin uou sly over the frequency range. A th ird a lte rn a tive , w hich is b e
com ing rapidly w idespread due to advances in digital sig n a l handling te c h n o l
ogy, is the use of re a l-tim e analyzers w h ic h present a continuously updated
com plete frequency spectrum on a display screen. A n o th e r fundam ental d if
fere nce betw een th e various filte r and analyzer types is in the filte r b a n d
w id th , n arro w or w ide and w h e th e r it is a fixed percentage of the tun e d fr e
quency or is a co nsta n t number of Hz absolute b a n d w id th independent of
tu n e d frequency.
It is very often necessary in practical vibration w o rk to use analyzers w ith
an exceedingly n a rro w bandw idth in order to separate closely spaced s in u
soidal com ponents or because the s tru c tu re s excited by vibration often c o n
tain m echanical resonances w ith large Q -values (lig htly dam ped resonances).
W h ic h type of in stru m e n ta tio n should be used in a p a rticu la r m easurem ent
99
situation m ust be decided upon by the u ltim a te use of the data obtained and
the m easuring equipm ent a vailable. This and th e follow ing chapters w ill dem
onstrate the fa c ilitie s provided by a w ide ra nge of in stru m e n ts and discuss
th e ir application to practical problem s.
A ccelero m eter
Type
Weight
(gram)
Charge
Mounted
S e n sitivity Resonance
(pC/m s- 2 )* Frequency
(kHz)
~ 4,5
4367
J |
13
~ 2
32
4368
||J
30
~ 4,5
27
4369
14
~ 2
32
11
1 2%
35
54
10 2%
18
4371
. 1
4370
D elta Shear
C o n stru c tio n having
g ood all-round
characteristics and
p a rtic u la rly low
se n sitiv ity to
tem perature
tran sients and
base strains
4375
2
excl.
cable
~ 0,3
60
M in ia tu re size, low
w eigh t D elta Shear type.
High resonance frequency
4374
0,7
excl.
cable
-0 ,1
75
Su bm ia tu re size, low
w e igh t shear type. V e ry
high resonant frequency
8309
3
excl.
cable
~ 0,0 0 4
180
4321
A,
55
1 2%
40
8305
~ 0,12
40
V ib ra tio n measurements
in three m utually
perpendicular directio n s
30
Reference standard
fo r com parison c alib
ratio n of accelerometers
V e ry High Uni-Gain
se n sitiv ity . B uilt-in
Pream p and L P filter.
Requires 2 8 V 2 m A DC
p o w e r supply
500
1000
8308
100
1 2%
30
8310
100
excl.
cable
1 20%
30
G eneral vibration
measurements. High
sensitivity fo r low-level
measurements
S h o c k measurements up
to 1 m illio n ms- 2 High
frequency vibr.m easurem ents
1 kH z
LP filte r
b uilt in
M in ia tu re size. Integral
fix in g stud.Integral
cable.
8306
A p p licatio n
Areas
27
28
4366
Im portant
Characteristics
A Shear Types
791117
Fig. 6 .3 . M a in ch aracte ristics and a p p lica tio n areas fo r B & K accelerom eters
in general, it can be said th a t the shear co n figuration gives the best a ll-ro u n d
results fo r general purpose accelerom eters and the com pression design is
used fo r accelerom eters w h ich are aim ed at particular applications.
101
The table in F ig. 6 .3 in d ica tes the a p p lica tio n and main characteristics of
the B & K accelerom eter range. A t firs t glance there may seem to be a c o n fu
singly large range of accelerom eters available. But it w ill be seen, a fte r
closer inspection, that th e y can be divided in to tw o main groups. A group of
general purpose types, w ith various s e n sitiv itie s and a choice of top or side
connectors, w h ic h w ill s a tis fy most needs, and a range of accelerom eters
w h ic h have th e ir cha ra cte ristics slanted to w a rd s a particular application.
W hen selecting an accelerom eter for a p a rtic u la r application the acceler
om eter's param eters and th e enviro n m e n ta l conditions it is to be used under
need to be considered as fo llo w s :
Frequency Range: The fre q u e n c y response o f an accelerom eter has a charac
te ristic shape as shown in F ig . 6 .4. M ea su re m e n ts are n o rm a lly confined to
using the lin e a r portion of th e response cu rve w hich at th e high frequency
end is lim ite d by the acce le ro m e te r's n a tu ra l resonance. A s a rule of th u m b
the upper freque ncy lim it fo r m easurem ents can be set to o n e -th ird of the ac
celerom eter's resonance fre q u e n cy so th a t vib ra tio n com ponents measured at
th is lim it w ill be in error by no more than + 1 2% (~ 1 dB). S m a ll, lo w mass ac
celerom eters can have a re so n a n t frequency as high as 1 8 0 kHz but for the
m ore sensitive general purpose accelerom eters resonant frequencies of
3 0 kHz (giving an upper fre q u e n c y lim it of 1 0 kHz) are typical.
It should be noted how ever th a t an accelerom eter's useful frequency range
102
is sig n ifica n tly higher, i.e. to 1 / 2 or 2 / 3 of its resonant fre q u e n cy, w here for
exam ple 3 dB line arity is acceptable. This m ay be the case w h e re vibration
m easurem ents are being used to m onitor th e in te rn a l condition o f machines
because repeatability is there m ore im portant th a n linearity.
In practice th e low er m ea su rin g frequency lim it is determ ined by tw o fac
tors. The firs t is the lo w -fre q u e n cy c u t-o ff of th e associated p re a m p lifie r, but
th is is not n orm ally a problem as th e lim it is u s u a lly w ell below 1 Hz. The sec
ond is the e ffe ct of am bient te m p e ra tu re flu c tu a tio n s (tem perature transients)
to w h ich the accelerom eter is sensitive. W ith m odern shear typ e accelerom e
ters this e ffe ct is typically 2 0 dB low er than fo r corresponding compression
types w hich th u s allow s m e asurem ent dow n to w e ll below 1 Hz fo r normal
environm ents.
Sensitivity, Mass and D ynam ic Range: Ideally, the higher th e transducer
sensitivity th e better, but a com prom ise has to be made because high se n sitiv
ity norm ally e ntails a large piezoelectric assem bly and co n se q u e n tly a rela
tive ly large, heavy unit w ith lo w resonant frequency. In no rm a l c ircu m
stances the sensitivity is not too critical a fa c to r as modern p re a m p lifie rs are
designed to accept these lo w -le v e l signals.
A ccelerom eter mass becom es im portant w h e n m easuring on lig h t test ob
jects. The accelerom eter sh o u ld load the s tru c tu ra l member as little as possi
ble: additional mass can s ig n ific a n tly change th e vibration levels and fre q u e n
cies present at the m easuring point and in va lid a te the m easured results. A n
approxim ate indication of th e change in s tru c tu ra l response due to loading
can be foun d using the fo llo w in g equations:
(6 .
1)
w here:
am = acceleration m easured w ith accelerom eter mounted
as = acceleration w ith o u t accelerom eter
fm = resonance frequency m easured w ith accelerom eter m ounted
fs = resonance frequency w ith o u t accelerom eter
ma = accelerom eter mass
ms = effective mass of th a t " p a rt of the s tru c tu re to w h ic h th e ac
celerom eter is mounted.
As a general rule, the accelerom eter m ass should be no g re a te r than onete n th of th e effective (dynam ic) mass of th e part of the s tru c tu re to w hich th e
accelerom eter is m ounted.
103
Fig. 6 .5 . M e a su rin g range nom ogram fo r B & K accelerom eters under the s p e
c ific conditions sta te d
104
connecting cables and a m p lifie r circu itry. This lim it is n o rm a lly below
0,01 m s 2 w ith general purpose in stru m e n ts m easuring over a w ide band.
S ig n ifica n tly lo w e r levels may be measured w h e n using a filte r fo r frequency
analysis.
The upper dynam ic lim it is determ ined by th e accelerom eter's structural
strength. G eneral purpose accelerom eters are lin e a r up to 5 0 to 1 0 0 kms - 2
(5 0 0 0 to 1 0 0 0 0 # ) w hich is w e ll into the range of m echanical shocks. For
very high shock m easurem ents th e shock accelerom eter (Type 8 3 0 9 ) shown
in F ig . 6 .3 can be used up to 1 0 6 m s - 2 (1 0 0 0 0 0 # ) .
A typical exam ple of how th e various factors lim it the o p erating range of
B & K accelerom eter types under specific co n d ition s is show n in the nom o
gram , Fig. 6 .5.
Transverse Response: The tra n sve rse se n s itiv ity of an accelerom eter is its
sensitivity to accelerations in a plane perpendicular to the m ain accelerom e
te r axis as show n in Fig. 6 . 6 . It is norm ally expressed in percent of the main
axis sensitivity and should be as low as possible. The transverse se nsitivity
varies according to w hich d ire ctio n is considered, the dire ctio n of m inim um
sensitivity is marked on m ost B & K accelerom eter types w ith a red paint
spot. M easurem ent of the m axim um transverse se n sitivity is part of the indiA x is o f
m axim um
sensitivity
A x is o f m ax im u m
Transverse
sensitivity
A x is of m in im u m
Transverse se n sitivity
(ideally zero
sensitivity at the
calibration frequency)
780732
1 06
___________________________________________________________________________________________________________________________________________750890/3
Fig. 6.1 1. The use o f a m ica washer and h e a t sink w ill e nable accelerom eter
to be used on su rfa ce s at te m p e ra tu re s ra th e r h ig h e r than the a c
celerom eter's d e sig n m axim um
Tem perature Transients: Piezoelectric accelerom eters also exhibit a v a ry
ing output w he n subjected to small te m p e ra tu re flu c tu a tio n s called te m p e ra
ture tra n sie n ts in the m easuring e n viro n m e n t. This is n o rm a lly only a p ro b
lem w here very low level or low fre q u e n cy vibrations are being m easured.
M odern shear type accelerom eters have a very low s e n s itiv ity to tem perature
transients.
Cable Noise: Since piezoelectric accelerom eters have a high output im p e d
ance problem s can som e tim e s arise w ith noise signals induced into the c o n
necting cable to the p re a m p lifie r. These spurious sig n a ls can result fro m
ground loops, trib o e le ctric noise, or e le ctrom agnetic noise.
G round loop currents can flo w in th e shield of accelerom eter cables be
cause of slig h t differe n ce s in the e le ctrica l potential of grounding p o in ts
w h e n the accelerom eter and the m easuring equipm ent are grounded se p a r
ately. The loop is broken by electrically is o la tin g the accelerom eter base fro m
the m ounting surface by means of an isolating stud (m ax. te m perature
2 5 0C ) and mica w asher.
T riboelectric noise can be generated by th e accelerom eter cable due to lo
cal capacity and charge changes betw een the conductor and shield as th e
cable vibrates. This problem is avoided by using a proper in te rn a lly graphited
accelerom eter cable and fix in g it to avoid cable m ovem ents as m uch as p ossi
ble.
Electrom agnetic noise can be a problem w hen the accelerom eter cable lies
in the vic in ity of runn in g electrical m achinery. Double shielded cable helps to
reduce th is problem but in severe cases a balanced accelerom eter and d iffe r
109
ential prea m plifie r should be used. The la tte r is standard practice w ith p e rm a
nent vibration m onitoring e quipm ent on in d u s tria l m achinery.
Base Strains: W hen an accelerom eter is m ounted on a surface w hich is u n
dergoing dynam ic d e fo rm a tio n s a spurious o u tp u t w ill be generated as a re
sult of strain being tra n s m itte d to the sensing elem ent. A ccelerom eters have
thick s tiff bases to m in im ise th is effect. N ote th a t Delta S h e a r ty p e s have a
p a rticu la rly lo w se n sitivity to base strains because the piezoelectric elem ent
is m ounted on a centre post rather than d ire c tly to the accelerom eter base,
and this does not so d ire c tly re su lt in shear deform ation of th e elem ents.
Nuclear Radiation: M o st general purpose accelerom eter types may be
used under gamma ra d ia tio n rates of 10 k R a d /h up to an accum ulated dose
of 2 M Rad w ith o u t s ig n ific a n t changes in characteristics. Types intended fo r
perm anent insta lla tio n on m achines and equipm ent may also be used in
heavy neutron radiation at rates of 1 M R a d /h up to accu m u la te d doses of
1 0 0 M Rad. Special m in e ra l insulated " h a rd lin e ' cables are necessary in th is
case.
M agnetic Fields: The m a g n e tic sen sitivity o f piezoelectric accelerom eters is
very low , norm ally less th a n 1 0 ms ~ 2 / T w ith a 50 Hz a lte rn a tin g field and
least favourable orie n ta tio n of th e accelerom eter.
Hum idity: M ost accelerom eters are sealed, either by epoxy bonding or
w elding and are therefore im pervious to m o isture . However, cable connectors
only o ffe r superficial p ro te ctio n . For sh o rt term exposure th e transducer
socket and cable plug can be dipped in s ilic o n e grease before the connector
is assem bled. For exposure to high h u m id ity or direct im m e rsio n the w hole
assembled connector should be encapsulated in an acid-free RTV silicone ru b
ber com pound. In w et e n viro n m e n ts standard Teflon insulated accelerom eter
cables should always be used.
Corrosive Substances: M o s t accelerom eters are encased in stainless steel,
tita n iu m or beryllium and th e re fo re have a h ig h resistance to m ost of the c o r
rosive substances encountered.
Acoustic Noise: The noise levels present in m achinery are not norm ally
su fficie n t to cause any s ig n ific a n t error in vib ra tio n m easurem ents. It may be
expected th a t the a coustically induced vib ra tio n in the s tru c tu re on w hich the
accelerom eter is m ounted w ill be far g reater than the a irb o rn e excitation of
the accelerom eter.
This concludes the list o f enviro n m e n ta l fa cto rs w hich need to be consid
ered w hen selecting and using piezoelectric accelerom eters. To illu stra te the
sort of fig u re s to be expected, th e fu ll sp e cifica tion s of a typ ica l general p u r
pose accelerom eter, B & K Type 4 3 7 0 , are ta b u la te d in F ig.6.1 2.
11 0
Weight:*
54 grams
Charge Se nsitivity:*
10 pC/m s- 2 2%
(~ 100 pC/g)
T y p ic a l A c o u stic Sensitivity:
(E q uiv. A c c'n . at 154 dB S P L ,
2 - 100 Hz)
0,001 ms~2 (0,0001 g)
M ax Transverse S e n sitivity :* * *
<4%
Piezoelectric M aterial:
PZ 23
T y p ic a l Tem perature Response
M in im u m Leakage Resistance:
(at 20C)
20 Gn
M ax im u m A m b ie n t Tem perature:
25 0 C (500 F)
a O b je c t:______________
B r e l Si K j r
***
****
r Urn. F r e q ______ Hz
W riting Spaed:
m m /sec.
Paper Spaed:
m m /w c 740761/1
791132
111
6 .3 . SELECTION OF A C C E L E R O M E T E R P R E A M P LIFIE R S
D irect loading of a piezoelectric accelerom eter's o u tp u t, even by re la tive ly
high impedance loads can greatly re d u ce the accelerom eter's s e n s itiv ity as
w e ll as lim iting its fre q u e n cy response. To elim inate th is effect th e a cce le r
om ete r output signal is fed through a pream plifier w h ic h has a very h ig h in
put im pedance and a lo w output im pedance, suitable fo r connection to th e re l
ative ly lo w input im pedance of m e a su rin g and analyzing in s tru m e n ta tio n . In
add ition to the fu n c tio n of im pedance conversion m ost pream plifiers include
va riable a m plification and many o th e r fa cilitie s for c o n d ition in g th e signal.
P ream plifiers are p o w ered either d ire c tly from AC m a in s or from b u ilt-in bat
te rie s or external DC supplies. A s u m m a ry of p re a m p lifie r types is s h o w n in
F ig .6.1 5. The type of p re a m p lifie r used depends on w h e th e r the a cce le ro m e
te r is considered as a voltage or charge source. Both voltage and charge sen
s itiv itie s of the accelerom eter are stated in the specifications.
The difference can be understood by considering th e equivalent c irc u it d ia
gram o f an accelerom eter in Fig.6 .1 3 . W ith in the fre q u e n c y range in qu e s
tio n , th e accelerom eter can be considered as a g enerator w hich is p a ra lle l coupled w ith an in te rn a l capacitance C. A s the charge on the capacitor is de
fin e d as
q = e . C
w here:
q = th e electrical charge
e = th e voltage across th e capacitor
C = th e internal capacitance
it fo llo w s that an a cceleration w h ic h produces a certain voltage, e, also p ro
duces a charge q on th e capacitor C.
V o ltag e pream plifiers, w hich w o rk on th e varying o u tp u t voltage fro m the
accelerom eter, have been w idely used, but have fa lle n fro m favour in recent
years since charge p re a m p lifie rs, w h ic h see the accelerom eter as a charge
source, have become available. The m a in reason for th is is that w ith voltage
791121
u
w here:
e0
qa
A
e
Ca
Cc
Cj
Sq
a
=
=
=
=
=
=
=
=
-
Q a'A
= e, A
Ca + Cc + Cj - C f (A - 1)
Sg a_________
Ca + Cc + C, - Cf (A - 1)
___________Cg
Ca + Cc + C, - Cf (A - 1 )
C, A
Cf A
( 6 . 2)
2635
*
Pream plifier
Type
3 digi
con d i
noise
am pli
sen sitivity
ion ng, low
char ge
ier
Measurement
Modes
A cceleration
V e lo c ity
Displacem ent
Acceleration
S e n sitivity
0,1 m V to
10 V /p C
( - 2 0 to
+ 80 dB)
Frequency
Range
2626
2651
2634
2650
t e i
3 d igi sen sitivity
c o n d il ion ng, low
noise char ge
a m p li ier.
1
Ud
3-Uni-C a n sensitiv ity s(tt ings.
Charge a r np lifier.
V e ry lo w frequency mea su rement
capabil tv
A c ce le ra tio n
V e lo c ity
0,1 m V to 1 V /pC 0 , 1 - 1 - 1 0 m V /p C
( - 2 0 to + 60 dB) ( - 2 0 to + 2 0 dB)
0,00 3 H z to
2 00 k H z
A cceleration
Low
0,2; 1; 2; 10 Hz
0,3; 3; 10; 30 Hz
L im its
High
1; 3; 10; 30;
> 100 k H z
2 00 k H z
Power Su pply
Internal batteries
o r Ext. DC
A C M ains
E x t. D C
Other
Features
Overload in d i
cator. Test
oscillator.
Battery con
d itio n indicator.
Input signal
ground flo ating
or grounded
vity
Lo w
and
fier
A cceleration
0,1 m V /p C to 100
0,9 to 10 m V /p C
m V /p C . 100 m V /V
internally a d ju st
to 100 V / V
able (0 to 2 0 dB) ( - 2 0 to + 40 dB)
1 Hz to 2 00 k H z
0,3 Hz to 200 k H z
0,3 & 3 Hz
and 2 kH z
Selectable
O verload in d i
cato r. D irect and
tra n sform e r cou
p led outputs.
Sm all, robust,
adjustable gain,
charge am p lifie r.
Excellent e le c
trom agnetic
radiation im
munity
1; 3; 10; 30;
> 2 0 0 kH z
Ext. D C
Normal o r d if
ferential inp u t.
Can be fix e d to
machine frames.
A C Mains
O verload indicator.
Test oscillator.
P articularly used
fo r com parison
c alibration of
accelerometers
791118
6 .4 . C A L IB R A T IO N A N D S Y S T E M P E R F O R M A N C E C H E C K S
Each in s tru m e n t produced by Bruel & K j r has been in d iv id u a lly checked
fo r agreem ent w ith published specifications before leaving th e factory. In th e
case of accelerom eters th e q u a lity control procedure includes a detailed c a li
bration of th e sensitivity and response of each individual accelerom eter. A ty
pical ca libration chart is s h o w n in Fig.6.1 6 .
The tw o le ft-h a n d colum ns give individually measured values fo r ch a ra cte ris
tics such as sensitivity, capacitance, tra n sve rse sensitivity etc. together w ith
a tem perature response cu rve and specified representative values for the ac
celerom eter's sensitivity to various e n viro n m e n ta l effects. The frequency re
sponse curve on the rig h t is individually plotted for each accelerom eter.
W here accelerom eters are stored and operated w ith in th e ir e nvironm ental
lim its, i.e. not subjected to excessive shocks, tem peratures, radiation doses
115
Acceleration
ms- 2
V e lo city
mm s~1
Displacement
Idm
Peak
10
20
40
Peak-toPeak
20
40
80
RMS
7,07
14,1
28,3
800127
U nknow n Signal
" i
M easuring A m p lifie r
2610
Level Recorder
2309 o r 2307
Unknow n
Accelerom eter
Charge
A m p lifie r
2634
l_ .
:
331? *
===
ODD<
)
M easuring A m p lifie r
2610
Calibration E x c ite r
4290
Compressor
Fig. 6 .2 0 . In stru m e n ta tio n se t-u p for p lo ttin g the frequency response o f a ccel
erom eters
many accelerom eters. C a libration levels are rather low , th a t is, in the order
of 1 ms 2 .
Some vibration testing specifications dem and that tra n sd u ce rs used on v a lu
able test specim ens are calibrated at a level equal to or h ig h e r than the test
level. Since the levels th e se transducers w ill experience are often unknow n,
a
high
a rb itra ry c o n firm a tio n
level
is selected,
com m only
100#
(~ 1 0 0 0 m s ~ 2 ). V ib ra tio n exciters capable of exciting an accelerom eter up to
this level are far m ore p o w e rfu l than th e tw o "h a n d y " e xciters previously
show n. The 3 8 0 / 4 4 0 N (8 5 /1 0 0 Ibf) force B & K exciter system includes an
118
Fig. 6.21 .(a) A transducer c a lib ra tio n exciter capable o l e x c itin g accelerom e
te rs a t 1 00 0 m / s 2 (B & K Type 4 8 0 1 T + 4 8 1 5 )
(b) Close-up view o f the Exciter H e a d Type 4 8 1 5 s h o w in g an accel
e ro m e te r to be c a lib ra te d m o u n te d on the Reference Standard A c
ce le ro m e ter Type 8 3 0 5
Compression
Sensitivity
Com parator
2970
C ontrol &
Balance
Standard------ *
Accelerom eter Signal
Control Signal
Conditioned
Standard Signal
2626
-Unknown
Accelerom eter
Signal
Reference Standard
A ccelerom eter 8305
(B u ilt in to 4815)
-C o n d itio n e d
Unknow n Signal
Ca lib ratio n
Head 4815
2926
2707 Power A m p lifie r !
C o nd itio nin g
A m p lifie r
D rive Signal
C onditioning
A m p lifier
V ib ra tio n
E x c ite r B o dy 4805
23 07 Level Recorder
740313
F ig .6 .2 2 . A n in s tru m e n ta tio n arrangem ent fo r the rapid and accurate back-toback (com parison) calibration o f accelerom eters
exciter head w h ic h is specially optimized to fu lfil the dem anding re q u ire
ments of tra n sd u ce r ca lib ra tio n up to th e se high levels. This is shown in
119
6 .5 . FORCE A N D IM P E D A N C E T R A N S D U C E R S
The forces producing and resulting fro m mechanical v ib ra tio n s are of g re a t
in te re st in m echanical dynam ics. The fo rce acting on a m echanical s tru c tu re
divided by th e vib ra tio n velocity it produces defines th e m echanical im p e d
ance of the structu re .
Like the accelerom eter, the force tra n s d u c e r also uses a piezoelectric e le
m e nt w h ic h , w hen com pressed, produces an electrical o u tp u t proportional to
th e force transm itte d th ro u g h it. For d yn a m ic force s ig n a ls the same sig n a l
co n d ition in g and m easuring in s tru m e n ta tio n as for piezoelectric a ccelerom e
te rs can be used. For lo w frequency and sem i-static forces, p re a m p lifie rs
w ith very long tim e constants are necessary.
The force transducer is m ounted in th e force tra n sm issio n path so th a t it is
subjected to th e forces to be m easured. It can measure both tensile and com -
Mass
Piezoelectric
Element
Acceleration
Output
Force Output
Piezoelectric
Element
Accelerometer
Force Transducer
8200
4370
Signal
Power A m p lifie r
2712
V ib ra tio n E x c ite r
4808
nN
Response Signal
Co nd itio n in g
A m p lifie r
2635
780269/1
6 .6 . P R A C TIC A L C O N S ID E R A T IO N S IN M O U N T IN G A C C E L E R O M E T E R S
H aving selected an accelerom eter, bearing in m ind th e param eters d is
cussed in section 6 .2 , a su itable m o u n tin g position m u st be chosen. The a c
celerom eter is m ounted w ith its main s e n s itiv ity axis a lig n e d w ith the desired
m easuring direction. A s previously m en tio n e d the accelerom eter w ill respond
to vib ra tio n in directions o th e r than its m a in axis. S e n s itiv ity decreases as th e
angle betw een the m ain axis and the d ire c tio n of vib ra tio n increases u n til at
a plane norm al to the m a in axis the transverse se n sitivity is a m inim um o f up
to 3 to 4% of the main axis sensitivity. The direction of m in im u m transverse
s e n sitivity is indicated by a red spot painted on many accelerom eters; in th is
directio n the transverse s e n s itiv ity is v irtu a lly zero.
The reason for m easuring vibration w ill norm ally d icta te th e accelerom eter
m o unting position. It sh o u ld be chosen so as to obtain a s h o rt rigid m e ch a n i
cal path from the vib ra tio n source avoiding gaskets etc., fo r example w ith ro
tating m achinery, bearing housings are ideal. Valuable in fo rm a tio n can be ob
tained fro m m easurem ents both in the axial direction and e ith e r the h o rizo n
tal or vertical radial d ire c tio n , w hichever is expected to have the lowest s tiff
ness.
The response
phenom enon, so
sure sig n ific a n tly
jacent m easuring
nant fre quen cy, w hich can severely lim it th e useful fre q u e n cy range of th e a c
celerom eter. The ideal m o u n tin g is by a th re a d e d stud o n to a fla t, smooth s u r
face as show n in Fig.6 .2 6 (a). A th in la ye r of grease a pplied to the m o u n tin g
surface before tighte nin g dow n the accelerom eter w ill usu a lly improve th e
m ounting stiffness. The tapped hole in th e machine p a rt should be s u ffic
iently deep so that the s tu d is not forced in to the base of th e accelerom eter.
The d ra w in g shows a typ ica l response c u rv e of a general purpose accelerom e
ter m ounted w ith a fixed stud on a fla t surface. The re so n a n t frequency a t
tained is alm ost as high as the 32 kHz m ounted resonant frequency a tta in e d
under ca lib ra tio n w here th e mounting s u rfa c e is dead fla t and smooth.
A com m only used a lte rn a tive m ounting m ethod is the use of a thin layer of
bees-w ax fo r sticking th e accelerom eter in to place. As can be seen from the
response curve. Fig.6 .2 6 (b) the resonant frequency is o n ly slightly reduced.
Because bees-w ax becom es soft at h ig h e r tem peratures, th e method is re s
tricted to about 40C . W ith clean surfaces, bees-wax fix in g is usable up to ac
celeration levels of about 1 0 0 m /s 2
A mica w ashe r and isolated stud are used as show n in F ig .6 .2 6 (c) w h e re
the body of the accelerom eter should be ele ctrica lly isolated from the m e a s u r
ing object. This is n o rm a lly to prevent g ro u n d loops as discussed e lse w h e re .
A th in slice should be peeled from the th ic k mica w asher supplied. This fix in g
m ethod also gives good re su lts, the resonance frequency of the test a c c e le r
om eter only being reduced slightly.
W here perm anent m easuring points are to be established on a m a ch in e
and it is not w ished to d rill and tap fix in g holes, cem enting studs can be used
as sh ow n in Fig.6 .2 6 (d). They are a tta ch e d to the m ea su rin g point by m eans
of a hard glue; epoxy and cyanoacrylate types are recom m ended as so ft
glues can considerably reduce the usable frequency range of the a ccelerom e
ter. Epoxy w ill be the longest lasting w h ile cyanoacrylate is quickest and e a s i
est to apply.
D ouble sided adhesive tape is a quick and easy m edium fo r m ounting a c c e l
erom eters. Thin tape used on flat, sm o o th surfaces gives fa irly good re su lts
but it can be seen fro m th e response c u rv e in Fig.6 .2 6 (e) that a s ig n ific a n t
reduction in frequency range can be expected, especially w ith thick tape.
A p erm an ent m agnet, as shown in F ig .6 .2 6 (f), is a simple a tta c h m e n t
m ethod w h e re the m easuring point is a fla t fe rro -m a g n etic surface. It also
e le ctrica lly isolates the accelerom eter. T h is method reduced the resonant fr e
quency of the test accelerom eter to a b o u t 7 kHz and consequently ca n n o t be
used fo r m easurem ents m uch above 2 kHz. The holding force of the m a g n e t
is s u ffic ie n t fo r vibratio n levels up to 1 0 0 0 to 2 0 0 0 m / s 2 depending on th e
123
124
Double sided
Adhesive Disc
200
500
1k
2k
5k
10 k 20 k 50 k 100 k
Frequency Hz________
M agnet
Max. Temp.
150C(300F)
10k 20 k
Frequency Hz
100
50 k 100k
200
500
1k
2k
5k
10 k 20 k
50 k 100 k
Frequency Hz_____ 791120
Fig. 6 .2 6 . M ethods o f m o u n tin g B & K pie zo ele ctric accelerom eters and e xa m
ples o f typical fre q u e n cy responses obtained
size of th e accelerom eter. Note th a t the use of the m agnet does not ensure
absolute repeatability of positio nin g .
A hand-held probe as show n in F ig .6 .2 6 (g) w ith th e accelerom eter
m ounted on top is very convenient fo r q u ick-lo o k survey w o rk , but can give
125
10
20
50
770805
126
The cable sho uld leave the specimen at the po int of lowest v ib ra tio n
____________________ __________________ _____________________ ___________________________ 800458
Fig 6.31. Accelerometer cables should be fastened down to avoid cable whip
and thus, interference from triboelectric noise
Measuring A m p lifie r
or Frequency A n a ly z e r
A cce lero m e te r
Pre a m p lifier
Accelerom eter
Accelerom eter
Isolation
V7777mn
/T7v7777/
"7
Mains
Ground
b) C orrect grounding of con ve ntio na l charge or voltage p re a m p lifie r having non-isolated input.
Measuring A m p lifie r
or Frequency A n a ly z e r
Accelerom eter
77777.;77777
A cce lero m e te r
Floating
p reampMfier
" I s o la tio n 'L ____
2634
Mains
Ground
c) Co rrect grounding of charge p re a m p lifie rs having flo a tin g in p u t isolated from ground.
130
A c ce le ra tio n
d B/O ctave
1 kH z
Frequency
Frequency
Displacement
- g
ro r*
O <u
Frequency
273046
, QO
V e lo city
Small gearbox/m otor
3% B/W_____________
0 mm/s
20-1
mm/s
o _ 0 ,1 m m /s t i t
2
Hi
10
Q P Q 1 4 3 Pot Ranga:_
50
Ractifiar.: R M S
100
200
50 0
j
lOOO
Hi
20 0 0
W ritin g Spaad:
50 00
10000
m m /w c
20000
Papar Spaad:
50 00 0
100000
0.1 m m /M C .
78 0 236
lysis
An addition al feature of th is filte r is th a t it can be tuned to any ratio co m b i
nation of the tu nin g (tachom eter) signal fre q u e n cy betw een 1 / 9 9 and 9 9 / 1 .
This enables order analysis to be perform ed, th a t is, the vib ra tio n level a ttri
butable to th e various harm onics and su b -h a rm o n ics of a m achine's fu n d a
m ental ro ta tio n frequency are measured as a fu n c tio n of ro ta tio n speed.
The tracking filte r also provides syn ch ro n isa tio n signals fo r the portable le134
135
Precision S o u n d L e ve l M e te r Type
2 2 0 9 fitte d w ith in te g ra to r adaptor
ZR 0 0 2 0 , 1 /3 O ctave F ilte r Set
Type 161 6 a n d an accelerom eter
Sound
Level
Meter
Type
Frequency
Acc.
Range
w ith Integrator V e l.
Z R 0020
Disp.
( 0,5 dB)
w o
ea
s
T
;
ft
2209
3 Hz to 30 kH z
31,5 Hz to 12,5 k H z
25 Hz to 5 k H z
25 H z to 5 kH z
31,5 Hz to 2 k H z
50 Hz to 2 k H z
50 Hz to 2 kH z
2210
''F a s t''
"S lo w "
2218
50 Hz to 2 k H z
RM S
Impulse
Peak
A ll w ith o r w ith
o u t m ax. Hold
R M S,
Impulse
Peak H old
Impulse H old
RMS
Im pulse
Peak Hold
Provision fo r
Connecting
External
Filters
V ib ra tio n
Meter
Scales
fSW-. *
2203
Detector
Modes
O utputs
A
fe w
10 Hz to 2 0 k H z
RMS
Averaging
Tim es
Yes
A C Lin
A C Lin
D C Lin
Conversion
from d B
Necessary
Yes
A C Lin, DC Log
Conversion fro m d B
Necessary
791115
137
For convenience in fie ld use the S L M 's are a va ila b le in carrying cases con
ta in in g the accessories necessary fo r sound and v ib ra tio n m easurem ents.
DR
FM
70 dB
60 dB
2,5 H z*
50 kH z
acceptable
poor
DC
10 kH z
acceptable
'P la y b a c k
speed 1 0 x
excellent
good
good
record in g speed
Fig. 6 .4 8 . The Portable tape recorder being u s e d to sim u lta n e o u sly record v i
b ra tio n in the v e rtic a l and tw o h o riz o n ta l planes on the p illa r o f a
ra ilw a y bridge fo r subsequent a n a lysis in the laboratory. The separ
ate carrying case (KA 2 0 0 0 ) co n ta in s three P re a m p lifie rs Type
2 6 3 5 and a Portable C alibrator Type 4 2 9 1 . a ll o f w h ic h are battery
operated
Fig. 6 .4 9 . Back in the lab o ra to ry taped sig n a ls are examined, h e re using the
N a rro w Band A n a lyze r Type 2 0 3 1 a n d X -Y Recorder Type 2 3 0 8
140
choice of high and low-pass filte rs can be selected so th a t unw anted signals,
noise, etc. can be prevented from in flu e ncin g the m easurem ents.
For the application show n in F ig .6 .4 8 three p re a m p lifie rs w e re used to
feed three channels of the recorder, the fourth ch a n n e l was used fo r verbal
com m ents via th e m icrophone / loudspeaker inclu d e d w ith the recorder. The
portable ca lib rator also show n in Fig 6 .4 8 provides a reference vibration
source for checking out the w h o le m easuring c h a in and allows a 1 0 m s- 2
(~1 g) reference vibratio n level to be recorded on th e tape. The p re a m p lifie rs
and calibrator are m ounted in a co nvenient carrying case.
141
IB R H H IIH IW
, 9
ts t
-jjj
143
Fig. 6 .5 4 . B a tte ry operated in s tru m e n ta tio n s e t-u p for recording the tim e h is
to ry o f periodic w aveform s
Averaged orb it plots can be d ra w n by feeding th e outputs from X and Y axis
transducers via the w aveform re trie ve r to an X-Y recorder.
6 .7 .6 . M onitoring for M echanical Bumps and Shocks
M echanical shock is a severe e nvironm ental hazard during th e transporta
tion of m any item s of m achinery, equipm ent and more or less fragile pro
ducts. A specially designed in s tru m e n t for c o n tin u o u sly m o n ito rin g and re
cording the level of excessive shocks is sh o w n in Fig.6 .5 5 . It is a self-con
tained, fu lly electronic apparatus th a t m easures th e maximum ve lo c ity and ac
celeration to g e th e r w ith the tim e of occurrence of shocks w h ic h exceed a
preset th re sh o ld level and records them on a reel of paper tape.
Shocks are picked up by a piezoelectric accelerom eter w h ic h can be
m ounted inside the sturdy recorder case or re m o te ly on a c ritic a l part of the
transported item . It m onitors sim u lta n e o u sly in three m u tu a lly perpendicular
directions so th a t shocks o ccu rrin g in any random direction are resolved into
three vectors. These are com bined in the in s tru m e n t to represent the m agni
tude of the applied shock.
B u ilt-in rechargeable batteries power the in s tru m e n t for approxim ately 18
days, for longer journeys an e xternal battery pack can be connected. By con
necting a very high se n sitivity accelerom eter to the recorder its sensitivity
can be increased by a factor of 1,000 making it suitable for re cording ground
trem ors and building shocks due to, for exam ple, rock blasting or heavy tra f
fic. The photograph in Fig.6 .5 6 show s the bum p recorder in use.
144
6 .8 .
LA B O R A TO R Y
T IO N
O R IE N T E D
M A IN S -O P E R A T E D
IN S T R U M E N T A
The instru m e nts described in the previous section are powered fro m in te r
nal batteries and can th e re fo re be ru n independently o f AC m ains pow er.
M ost of them can, how e ver, be run fro m the mains in laboratory s itu a tio n s
via one of the battery chargers or pow er supplies available fo r th is purpose.
A lth o u g h battery-pow ered in s tru m e n ts are already sh o w in g a high degree
of sophistication w e s till m ust look to m ains powered in stru m e n ta tio n fo r the
most po w erfu l m easuring and analyzing systems. R esearch and developm ent
w ork in the laboratory and the analysis of tape recorded data is s till th e do
m ain of m ains operated in stru m e n ta tio n .
The firs t stages of th e vibration m easuring c h a in , th e transducer and
prea m p lifie r discussed in 6.1 and 6 .2 apply to both b a tte ry and m ains-operated m easuring and analyzing in stru m e n ta tio n . From th e re on there is a w ide
choice of m ains-operated filte rin g , detection and re co rd in g in stru m e n ta tio n
available.
The m easuring a m p lifie r is a central in stru m e n t he re , it is used alone as a
precision signal detector and level in d ica tor and to g e th e r w ith a filte r se t, to
perform frequency analysis. A m easuring am plifier and filte r-s e t are often
com bined into a signal in s tru m e n t called a frequency analyzer or sp e ctro m e
ter.
6 .8 .1 . Measuring A m plifiers
The choice of m easuring a m p lifie r w ill m ainly depend on w h e th e r it should
be able to be additiona lly operated fro m a DC supply, w h e th e r peak in d ica tio n
6 .8 .2 . Separate Filters
Add-on filte rs fo r m easuring am plifiers fa c ilita te frequency analysis; both
constant ba ndw idth and co n s ta n t percentage ba n d w id th types are available
as show n in Fig 6 .6 0 .
The Heterodyne Slave Filter Type 2 0 2 0 is tu n e d in the fre q u e n c y range
1 0 Hz to 20 kHz to the d iffe re n c e frequency of tw o high fre q u e n c y signals.
Tuning signals are obtained fro m one of the B & K Generators Types 1 023
148
B B
* |
mm
#
m...
- :-j ^
Fig. 6 .6 0 . H eterodyne Slave F ilte r Type 2 0 2 0 (constant b a n d w id th ) and ThirdOctave Bandpass F ilte r Type 1 6 1 8 (constant percentage bandw idth)
and 1 0 2 7 , th e Heterodyne A nalyzer Type 2 0 1 0 or the Tracking Frequency
M u ltip lie r Type 1 9 0 1 . The filte r is w id e ly used fo r analysis on a linear fre
quency scale w h ic h is best su ite d to constant b a ndw idth filte rs , and is p a rtic u
larly valuable fo r analyzing dynam ic system response in co n d itio n s of high
noise and d isto rtio n and fo r tracking analysis w h e n tuned fro m the tracking
frequency m u ltip lie r.
For analysis in standard octave and th ird -o c ta v e co ntiguous bands, filte r
sets such as th a t shown in F ig .6 .6 0 are used. Third-octave and octave band
w idth analysis is of lim ited value in the general study of m echanical vibration
w here n a rro w e r bandw idth resolution is n o rm a lly required. T hird-octave a n a
lysis can be of value h o w e ver w here it is required to co rre la te vibration
spectra w ith noise spectra and also w h e re sim ple spectrum com parison is
needed fo r m achine health m onitoring and q u a lity control purposes.
M ain data fo r add-on filte rs is shown to g e th e r w ith the data fo r dedicated
analyzers in F ig .6 .6 2 .
6 .8 .3 . Frequency Analyzers
Most convenie nt for p e rfo rm in g frequency analysis of vib ra tio n signals in
the laboratory are the range of analyzers sh o w n in Fig.6 .6 1 . T w o types of a n
alyzers are represented, serial analyzers and re al-tim e analyzers. The serial
analyzers m ay be synchronised w ith , and read out to, level or X-Y recorders.
R eal-tim e frequency analyzers provide analysis in all fre q u e n cy bands s im u l
taneously giving a v irtu a lly instantaneous graphical display of frequency
spectra on a large b u ilt-in screen. These spectra can, of course, also be read
out to a graphic recorder.
Looking at the analyzers in tu rn , the firs t analyzer show n in F ig.6.61 a l
lows a co n tin u o u s analysis to be made by sw eeping th ro u g h th e desired fre
quency range. A constant percentage filte r bandw idth as n a rro w as 1% al149
H eterodyne Analyzer
Type 2 0 1 0
N arrow Band
S pectrum A nalyzer
Type 2031
150
1617
Frequency A nalyzers
1618
2120
2010
Continuous
sweep.
Constant
bandwidth
Classi
fication
C o n tin u o u s
sweep.
Constant
b andw idth
C o n tin u o u s
sweep.
Constant
percentage
b a nd w id th
Bandw idth
3,1 6 Hz
10 Hz
31 ,6 Hz
100 Hz
1/1 O cta ve
1/3 O cta ve
1%
3%
10%
1/3 O c t.
(-23% )
3,16 Hz
10 Hz
31,6 Hz
100 Hz
316 Hz
1000 Hz
0,25% o f
selected
freq. range
ie. from
0,025 Hz
to 50 H z
Centre
Frequency
Range
10 H z to
2 0 kH z
2 Hz
to 20 k H z
2 Hz to
200 kH z
0 - 1 0 Hz
to
0 -2 0 kH z
in 11 ranges
Power
S u pp ly
2 Hz to
160 k H z
A C M ains
2 H z to
2 0 kH z
A C M a in s
and
Ext. D C
2033
2031
Constant- bandwidth
Real tim e
2131
O ctave and
third-octave
Real-tim e
A s 2031
plus x 10
1/1 O ctave
zoom
(0,025% BW) 1/3 O ctave
(ie. 0,0025
H z to 5 Hz)
A s 2031
plus x1 0
zoom
1,6 H z to
20 k H z
A C M ains
791116
15 2
M F
=1
Accelerom eter
4368
Magnetic
9
/Tj H
i!
^
M achine
Charge
A m p lifier 2 6 3 5
T ra ck in g Frequency
M u ltip lie r 1901
>
A ccelerom eter
g 5 j *
Digital Cassette Recorder
7400
h1
*
-
. .M
lies
-v
.* aoBsIBB*
D igital
Frequency
An a ly ze r
2031
2033
2131
X - Y Recorder
23 08
Level R e co rd e r
2307
157
6 .9 . A S U M M A R Y
A careful study o f the preceding sections of th is chapter should enable the
vibration engineer to select and u tiliz e the m easuring equipm ent necessary to
perform thoroug h and m eaningful vibration m easurem ents in m ost o f th e s itu
ations occurring in practice. F u rth e r details of th e frequency a n a lysis of con
tin u o u s and tra n s ie n t signals are given in the next chapter (C hapter 7). A fu r
th e r study of th a t chapter may th u s be required before m easurem ents are
made. On th e o th e r hand, w h e th e r the v ib ra tio n s consist of steady (sta tio n
ary) vibrations o r shocks it may be useful at th is stage to o u tlin e a sum m ary
procedure. This su m m a ry is m eant as an aid in rem em bering the m ost im p o rt
a n t factors in th e setting up and use of a v ib ra tio n m easurem ent system,
rath er than as a detailed "tu rn in g -o f-th e -k n o b s type of procedure.
1 . D eterm ine c a re fu lly where to p la ce the vib ra tio n transducer, a n d its po ssi
ble m ass-loa d in g effects (Section 6.2).
2. Estim ate w h a t types and levels o f vibrations th a t are likely to be present
at the tra n sd u c e r m o u n tin g p o in t (periodic vib ra tio n s , random vibrations,
shocks).
3. Select the m o s t suitable vib ra tio n transducer (accelerom eter) co n sidering
item s 1 a n d 2 above as w e ll as e n v iro n m e n ta l factors (te m p e ra tu re , h u
m idity etc.). See section 6 .2 .
4 D eterm ine w h a t type o f m e a su re m e n t w o u ld be m ost appropriate fo r the
problem a t hand. (Overall m e asurem ent of acceleration, ve lo c ity or dis
placem ent, w a ve fo rm recording, m agnetic tape recording, fre q u e n c y ana
lysis).
5. Select the m o st suitable e le c tro n ic equipm ent, considering fre q u e n c y and
phase ch a ra cte ristics, dynam ic range, and convenience of o p e ra tio n . See
sections 6 .1 , 6 .7 and 6.8.
6. Check a nd ca lib ra te the o v e ra ll system in c lu d in g accelerom eter a n d con
n e ctin g cables, see section 6 .4 .
7. M ake a sketch o f the in s tru m e n ta tio n system w ith all type n u m b e rs and
serial num bers included.
8. S elect the appropriate accelerom eter m o u n tin g m ethod, considering vibra158
6 .1 0 . B IB LIO G R A P H Y
B & K PUBLICATIONS:
(M iscellaneous)
V ibration
159
7.
7 .1 . IN T R O D U C T IO N
The traditional w a y of perform ing a frequency a n a ly s is is to pass th e signal
th ro u g h a system co n sistin g of filte r, detector and recorder (or display) as il
lu stra te d in Fig.7 .1 . The tim e taken to obtain a re s u lt for each fre q u e n cy
(bandw idth) w ill be de te rm in ed by th e delays involved in each of the th re e ele
m e nts, and these w ill th u s be discussed in the fo llo w in g . The d iscussion ap
plies to both analogue filte rs, and also to the m o re recent d ig ita l filte rs,
w h ic h behave in a ve ry sim ila r m a n n e r to their analogue counterparts.
160
or
bnR ^ 1
(7.2)
w h ere b is the relative b andw idth (= B/10 ), nR is th e num ber of periods of fre
quency f0 in tim e TR (= fQ Tr ) and f0 is the centre frequency of th e filte r. This
form is more applicable to constant percentage b a n d w id th filte rs.
Thus the response tim e of a 10 Hz b andw idth filte r is approx. 1 0 0 ms
w h ile the response tim e of a 1% b a ndw idth filte r is approx. 100 periods.
F ig.7.2 also illu stra te s th a t th e "e ffe c tiv e le n g th of the im p u lse (TE ) is
also approxim ately 1/ B , w h ile to integrate all of th e energy c o n ta in e d in the
filte r im pulse response it is necessary to integrate o ver at least 3 Tp .
The detector is used to m easure the mean p o w e r in the filte r o u tp u t, and
consists of a squaring section to obtain the in sta n ta n e o u s pow er, fo llo w e d by
an averaging section to sm ooth out flu c tu a tio n s and fin d the m ean square
value over a specified averaging tim e TA .
There is no e ffe ctive delay in th e squaring s e ctio n , but errors can occur if a
tru e squaring is not achieved. In some in s tru m e n ts th e parabola correspond
ing to squaring is approxim ated by piecewise lin e a r sections, and th e result
w ill only be valid fo r crest fa cto rs up to a specified value, e.g. 5. So-called
LMS (log mean square) detectors achieve true sq u a rin g by lo g a rith m ic conver-
161
sion follow ed by a m p lific a tio n by a fa c to r of 2, and are not likely to have crest
fa cto r lim itations w ith in th e ir d yn a m ic range. In b oth digital filte r and FFT in
stru m e n ts, squaring is perform ed n um erically and is thus v irtu a lly perfect
w ith in the dynam ic range lim ita tio n s .
S ince there is a continuous o u tp u t from the filte r, the ideal averaging
w o u ld be a running lin e a r average over the previous TA seconds. T h is is very
d iffic u lt to achieve, how ever, and recourse is n o rm a lly made to 2 a lte rn a
tives.
(a)
(b)
17 13 4 0 /1
162
7 .1 .2 . Digital Filters
A (recursive) digital filte r is a calcu la tio n device w h ic h receives a sequence
of digital values at its in p u t, operates on each sam ple in a d e fin e d manner,
and outputs a sample for each input. The d ig ita l operations ca rrie d out, viz.
addition, m u ltip lica tio n and delays, mean th a t th e relationship betw een in
put and o u tp u t samples can be described by a d iffe re n c e equation, w hose pro
perties can be made very s im ila r to a d iffe re n tia l equation w h ic h m ight de
scribe an analogue filte r. F ig .7 .4 show s a typ ica l 2-pole section used in a
1 /3 -o c ta v e digital filte r analyzer (3 of these are cascaded to give 6 -p o le filtra
tion).
There are tw o ways of changing the pro p e rties of a given d ig ita l filte r c ir
cu it such as th a t shown in F ig .7.4:
(a)
163
(b)
Thus, the three 1/3 -o c ta v e ch a ra cte ristics w ith in each octave are generatted by changing co e fficie n ts, w h ile th e various octaves are covered by re p e tit
ively halving the sam pling frequency. Every tim e th e sam pling fre q u e n c y is
halved, it means th a t only half the num ber of sam ples m ust be processed in
a given tim e, and it w ill be seen th a t th e total nu m b e r of samples fo r all oc
taves low er than th e highest is ( 1 / 2 + 1 / 4 + 1 / 8 + ....) w hich in th e lim it
is the same as th e num ber in the h ig h est octave. T h is means th a t by being
able to calculate tw ic e as fast as is necessary for th e upper octave a lo n e , it is
possible to cover any num ber of lo w e r octaves in re a l-tim e , and th is is the
o th e r reason w h y d ig ita l filte rs are so w e ll adapted to real-tim e c o n sta n t per
centage bandw idth analysis over a w id e frequency range.
7 .1 .3 . FFT Analysis
The FFT algorithm is an extrem ely e ffic ie n t way of calculating the so-called
D iscrete Fourier T ransform (DFT) w h ic h is a discrete, fin ite approxim ation to
th e Fourier T ransform given in C hapter 2. The a ctu a l equation fo r th e fo r
w a rd transform is:
N_j
G (k ) = - L 2 g (n ) e
N n=0
164
2nkn
"
(7 .3 )
g (n ) =
2 G (k)e
k=0
(7 .4 )
165
166
Picket fence effect, re s u ltin g from the discre te sam pling of th e spectrum
in the frequ ency dom ain. It is as though th e spectrum is vie w e d through
the slits in a picket fence, and thus for exam ple peak values are not ne
cessarily seen. The possible erro r resulting fro m th is depends on the over
lapping of adjacent filte r characteristics as s h o w n in F ig.7 .6 , and is not a
unique fe a tu re of FFT analysis. It occurs w henever discrete filte rs are
used such as in typical 1/3 -o c ta v e analysis. It is eased by having a grea
ter overlap of adjacent filte rs , and w ith H a n n in g w e ig h tin g fo r example, it
is a m axim um of 1,4 dB (com pared w ith 3 ,9 dB for fla t w e ig h tin g ). The e r
ror can be com pensated fo r w h e re it is k n o w n that there is only one fre
quency com ponent w h ic h fa lls between tw o spectral lines (e.g. w ith a cal
ibration signal). One special case of the picket fence e ffect is w here the
spectrum sam ples fall at th e zeros betw een th e sidelobes m entioned in
(b), and th e tim e w in d o w e ffe ct becomes invisible (Fig.7.7). This corre
sponds w ith an exact in te g e r num ber of periods in the record length T,
but seldom occurs in practice because th e frequencies in question must
a)
Flat
w in d o w :
i k A m p litu d e
Power
mm. fWVW\
0
b)
H an n in g
0 dB to
- 3 , 9 dB
rT
w in do w :
k Pow er
^ A m p litu d e
0 dB to
- 1 , 4 dB
n
800065
167
168
T h e S h a p e F a c t o r o f a f ilt e r is t h e r a t io o f t h e w id t h o f t h e c h a r a c t e r is t ic a t 6 0 d B to t h a t a t
3 d B ( its 3 d B b a n d w id th ).
Lin .
No. of Spectre: 16
>. of Spectre: _ 2 _
F re q ue ncy (Hz)
170
tory. A logarithm ic fre q u e n c y scale m ay, however, o cca sio na lly be chosen in
co n ju nctio n w ith c o n s ta n t bandwidth (though over a lim ite d frequency range)
in order to dem onstrate a relationship w h ic h is linear on log-log scales (e.g.
conversions between acceleration, v e lo c ity and d isplacem ent, see F ig.6 .1 ).
The am plitude scale o f a frequency analysis should a lm o s t always be lo g a r
ith m ic fo r a num ber o f reasons. In fa c t it is only w h e n th e param eter being
m easured is directly o f in te re st (e.g. s tra in ) that there is any point in having a
linear scale. N orm ally, th e vibrations m easured are an in d ire ct expression of
in te rn a l forces, and th e n the lo g a rith m ic scaling makes th e results less s e n s i
tive to th e influence o f th e more or less random tra n s m is s io n path by w h ic h
the signal has reached the m easurem ent point. F ig .7 .9 illustrates th is for
tw o m easurem ent p o in ts on the same gearbox (both be in g equally re p re se nta
tive of the internal condition). The lo g a rith m ic rep re se nta tio n s of th e tw o
spectra are quite s im ila r, w h ile the lin e a r representations are not only d iffe r
ent, but hide a num ber o f com ponents w h ic h could be im p o rta n t.
7 .2 . S ER IA L A N A L Y S IS OF S T A T IO N A R Y SIG NALS
A lth o u g h serial analyzers have been superseded by re a l-tim e analyzers for
problem s requiring a de tailed analysis, it is still the m o st common te ch n iq u e
used in portable b a tte ry operated e q u ip m e n t (see S e ctio n 6.7) and th is sec
tion w ill thus concentra te on that a p p lica tio n . The g raph of Fig.7 . 10 can be
used m ore generally, how ever, and fo r m ore detail re fe re n ce can be m ade to
the B & K book, "F re q u e n cy A n a ly s is " fo r the selection of analysis p a ra m e
ters.
The norm al procedure to be fo llo w e d is to choose a su ita b le averaging tim e
based on the signal type and analysis bandwidth. From th is an e q u ivale n t
w ritin g speed may be chosen, and fin a lly a paper speed based on th e b a n d
w id th or w ritin g speed. F ig .7 .1 0 c o n ta in s the required in fo rm a tio n.
S W EEP S P E E D F O R SW EPT F R E Q U E N C Y A N A L Y S I S
Frequency f (Hz)
272X4
n ba n d w id th and freq uen cy range, d e te rm in e F ilt e r Dw ell Tim e, Averaging T im e an d Sweep Speed as fo llo w s :
1.
R andom Signal:
fro m the a p p ro p ria te lin e fo r B T A = 10 (h o rizo n ta l lin e s fo r constant ban dw idth, s lo p in g lines fo r
1 y d B R M S e rro r .
F o r a higher B T A p ro d u c t a n d consequent re
d.
F o r A C recording, T A is determ ined b y recorder w ritin g speed W and this can be read d ire c t ly fro m the right-hand scale. S w eep speed can
then be calcu lated fro m S =
2.
D eterm inistic S ign a l (p e rio d ic o r quasi-periodic): R ea d T 0 based on filte r response tim e fro m appropriate lin e fo r B T D = 4 (or 2 where
bandwidth < 1%).
Read also m in im um averagin g tim e T A based on 1/4 d B r ip p le fr o m lin e f T A = 3 (in depen den t o f bandw idth).
BW
The
governing sweep speed is the lesser o f this and th e va lu e as c a lcu la ted fo r D C recording.
In all cases w h ere sweep speed is governed b y filt e r response tim e, increase T A t o th e highest value w h ich does n o t a ffe c t sweep speed,
(i.e. T q /4 fo r D C recording)
Note (1) Averaging tim e s w ill n o rm a lly be con sta n t ov er a t least a h alf decade, whereas filt e r response tim es change a u to m a t ic a lly w ith frequency.
N o te (2) S w ill be in H z /s fo r B in Hz, but w ill be d ire c t ly in m m /s fo r B expressed as e q u iv a le n t recorder paper length in m m .
(3) A b o ve ap p lies t o 100 m m paper w id th . F o r 5 0 m m paper substitute 2W fo r W.
172
2722t
(7 .5 )
Table 7.1 gives values of e (in dB) fo r various values o f th e product BTA
bta
10
20
30
40
50
80
1 20
e (dB)
1,5
1 ,0
0 ,8
0 ,7
0 ,6
0 ,5
0 ,4
Table 7.1.
F ig.7.1 0 can be used to select values o f TA for various bandw idths and c e n
tre frequencies for a BTA product of 1 0 . This value of TA can then be m o d i
fied p roportionally fo r o th e r BTA products.
W here AC recording is used, as w ill norm ally be th e case w ith a portable
analysis set because th e dynam ic range of the m eter DC output is u su a lly
< 50 dB, the averaging tim e is selected in d ire ctly by s e le ctio n of the pen w r it
ing speed. This equivalence is indicated in Fig.7 .1 0 w h e re the values for
50 mm paper w idth can be applied to th e battery operated portable Level Re
corders Types 2 3 0 6 and 2 3 0 9 .
Filter Response T im e This w ill o n ly be the lim ita tio n for d e te rm in is tic
signals w here a sm a ll averaging tim e is required and w h e re the recorder
w ritin g speed does not dom inate. The tim e taken to sweep one b a n d
w id th (Td ) is m ade greater than 4 filte r response tim e s (i.e. BTD S* 4 ) in
order that the fre q u e n c y error (delay) of the recorded spectrum w ill be
less than on e-q uarte r bandw idth.
(b)
Averaging Time This w ill only be th e lim ita tio n fo r DC recording and
is th u s not norm a lly th e lim itin g fa c to r w ith portable analyzers.
(c)
Recorder W riting Speed This w ill always be th e lim ita tio n for AC re
cording of analyses of random s ig n a ls , and may a lso dom inate over filte r
response tim e for d e te rm in istic (discrete frequency) signals.
173
For analysis w ith a typical portable set having bandw idths 3% and 23%, a
logarithm ic frequ e n cy axis w ith 5 0 mm per decade, and 5 0 m m paper w id th
representing 5 0 dB, the options are m uch m ore lim ited. A ssu m in g , as fo r
m ost m achine vibration sig n a ls, that the lo w e s t frequency com ponents (e.g.
from shaft ro tatio n ) are stable and d e te rm in is tic . Table 7 .2 has been drawn
up to a llow th e selection of both pen w ritin g speed and recorder paper speed
as a fun ction of th e low est fre q u e n cy to be analyzed.
2 -5
5 -1 5
1 5 -5 0
5 0 -1 5 0
> 150
23%
B andw idth
\A/(mm/s)
16
16
40
100
100
P (m m /s)
0,3
10
10
3%
Bandw idth
W (m m /s)
16
16
40
100
100
P (m m /s)
0,03
0,1
0,3
3
800716
Table 7.2.
174
5 0 -1 5 0
1 5 0 - 500
H /(m m /s |
16
40
100
100
P (m m /s )
10
10
W /(mm/s)
16
40
100
P (m m /s )
0,3
500 - 1,5 k
> 1,5 k
7 .3 .
R E A L-TIM E A N A L Y S IS
NALS
OF S T A T IO N A R Y A N D T R A N S IE N T S IG
In f r e q u e n c y a n a ly s is , r e a l- t im e o p e ra tio n m e a n s t h a t th e e n t ir e in p u t s ig n a l is a n a ly z e d , in a ll
fr e q u e n c y b a n d s (of in t e r e s t ) a ll of th e tim e , e v e n t h o u g h t h e r e m a y b e a t im e d e la y b e t w e e n in
p u t a n d o u tp u t.
175
Fig. 7. 11. In te r-h a rm o n ic com ponents re s u ltin g fro m in s u ffic ie n tly tig h t a s
sem bly o f jo u rn a l b e a rin g com ponents
does have th e advantage of co vering a w id e fre q u e n cy range w ith a relatively
sm all num ber of filte rs , and is th u s often th e m ost efficient and economical
tech niq ue to use in m achine m onitoring and quality control applications.
O ther applications include th e analysis of shocks and tra n s ie n ts , and w here
th e vibration m easurem ents are to be com pared w ith sound spectra, typically
presented in 1/3 -o c ta v e s .
176
H a lf t h e s a m p lin g fr e q u e n c y .
quency values are ca lcula ted , a lthough because each fre q u e n c y com ponent is
com plex, th is represents the same a m o u n t of data (and occupies th e same
m em ory space) as th e o rig in a l 1 0 2 4 real-valued tim e sam ples. Not all o f the
5 1 2 values can be used; to e lim in a te th e problem of a lia s in g (Section 7 .1 .3 )
a low -pass filte r is applied w ith a c u t-o ff frequency less th a n the N yquist fre
quency to a llo w for its fin ite slope. For the B & K analyzers it is ty p ic a l to
place th e filte r c u t-o ff so th a t the firs t 4 0 0 lines are v a lid , and are displayed,
w h ile th e last 11 2 lin e s are affected by th e filte r and are not operated on fu r
th e r (w hen the com plex spectrum is converted to a p o w e r spectrum etc.).
Thus, the frequency resolution ( Af ) is always 1 / 4 0 0 o f the selected fu llscale frequency ( f ,s ), and the autom atically-selected sam pling fre q u e n cy is
2 ,5 6 tim es the fu ll-sca le frequency ff s . W hen using fla t w e ig h tin g , th e filte r
characteristic is a s in x /x function w ith bandwidth equal to the re so lu tio n A f.
As m entioned in S ection 7 .1 .3 , the spectrum obtained is that of a p e rio d ic
signal of period T (= N A t ) and the lin e spacing A f is th u s 1 / T .
B = Af
Thus,
(7 .6 )
From Eqn. (7.6) it can be seen th a t each spectrum ca lcu la ted by an FFT a n a l
yzer has a BT product of 1. Even w h e re for example H a n n in g w e ig h tin g is
used, and the ba ndw idth is increased by 50%
i.e.
15
(7 .7 )
1,5 A f = - y
Af
(7 .8 )
4 0 0
f,f.S.
ffs
(Hz)
T
20 k
10k
5k
2k
1k
500
2 0 ms 4 0 ms 8 0 ms 20 0 ms 4 0 0 ms 8 0 0 ms
200
100
50
2s
4 s
8s
20
10
20 s 40 s
300
Frequency (Hz)
500
800076
300
Frequency (Hz)
800077
179
180
181
182
Range of zoomed
spectrum (b)
Gearbox
Before Repair
a)
Baseband Spectrum
Frequency (Hz)
800074
(9 0 0 -1 1 1 00 Hz
b) Zoom ed
Spectrum
Gearbox
Before Repai
900
9 2 0 ....... 94 0
960
980
1000
1020
10 40
1060
1080
1100
Frequency (Hz)
Manuring Objct:-------- ------------------------------------------------------------------------------ -------------------------------------------------------------- '800073
183
pinion speed) is also now apparent. On the o th e r hand, because o f speed flu c
tuations, the m a jo r peaks have spread over a num ber of lin e s , and thus a
higher zoom fa c to r w ould not give any benefit.
Fig. 7. 16 BTa prod uct vs. ave ra g in g tim e fo r the D ig ita l F ilte r A n a ly z e r Type
2131
guous spectrum values into th e analyzer fo r display on the screen. F ig.7 .1 7
com pares a 1 /1 2-octave sp e ctru m obtained in th is way w ith th e equivalent
1/3 -o c ta v e spectrum . Note th a t even though th e method is not "re a l-tim e '',
because not all th e spectrum is calculated at one tim e , the e n tire tim e signal
is processed fo r each pass, and th erefore any s h o rt-te rm n o n -s ta tio n a rity w ill
not have any adverse effect on th e results.
The appropriate scale of F ig .7 .1 6 can also be used in th is case fo r selec
tio n of a suitable averaging tim e .
1 85
FFT
D ig ital Filter
187
(for the sam e fu ll-sca le fre q u e n cy) must be ta ke n into account w h e n con
verting to ESD, and in c id e n ta lly results in a 1 0 dB loss of d yn a m ic range.
(2)
188
800071
(7.9)
(7.10)
191
7 .4 . A N A L Y S IS OF N O N -S T A T IO N A R Y S IG N A LS
The type of analysis considered in th is section is th e use of a m oving tim e
w in d o w to see h o w the s h o rt-te rm frequency spectrum varies w ith tim e
(F ig .7 .2 4 ). It is assum ed that the w in d o w length can be chosen so th a t th e in
divid ual w indow ed sections are q u a si-sta tio n a ry, w ith o u t being so s h o rt that
th e attainable re so lu tio n is too coarse.
T w o main application areas are envisaged:
192
(a)
(b)
M any considera tion s are com m on to both cases, since case (a) can be con
sidered a special case of (b) w ith an average over one cycle o n ly. Conse
quently, it is p rim a rily case (b) w h ic h w ill be discussed in d e ta il, though any
special considerations applicable to case (a) w ill be discussed w h e re appropri-
193
Gauss Impulse
M u ltip lie r
5623
Digital Frequency
A n a lyze r
2131
Level Recorder
2307
_____________________________________________________________________________________________________________________________________________ 790429
7 .4 .2 . E xam ple
In order to illu stra te th e general procedure, the exam ple w ill be taken of v i
bration signals (velocity) measured on th e cylinder head of a 4 -c y lin d e r, 4stroke diesel engine ru n n in g at 1 5 0 0 r.p .m . (7 5 0 cpm or 8 0 ms cycle tim e).
The m easurem ent and recording system w e re linear up to 10 kHz, but th e s ig
nals contained in fo rm a tio n up to 2 0 kHz. The s ig n a ls w ere recorded at
1 5 i.p.s. on an FM recorder, in p a rallel w ith a once-per-cycle tacho signal ob
tained w ith a p h o to -e le ctric pickup fro m the ca m sh a ft. The analyses w e re
made w ith the FFT analyzer Type 2 0 3 1 .
V ie w in g the tim e signal on the screen of the analyzer it could be d e te r
m ined th a t individual im pulses had a length of 2 m s, and for th is reason it
w as desirable to have a w in d o w le n g th as short as 8 ms (i.e. h a lf-a m p litu d e
194
length 4 ms w ith H anning w eighting). T his could not be achieved d ire c tly , as
the shorte st w in d o w length available (in 2 0 kHz range) is 2 0 ms (Table 7 .4 ),
but by slow ing dow n th e tape recorder 10:1 on playback it was possible to
achieve the desired param eters in th e 5 kHz range ( T = 8 0 ms). This m eant
th a t th e effective fu ll-s c a le frequency w a s now 5 0 kHz, but only the firs t 1 6 0
lines ( < 2 0 kHz) w ere u tilise d in d isp la yin g the results. The total cycle tim e of
8 0 m s could be covered by varying the "R e co rd s-A fte r-T rig g e r" s e ttin g
(Fig.7 .1 4 ) between 0 ,0 and 9,8 m em ory lengths w ith increm ents of 0 ,2 (i.e.
e ffe ctive ly 1,6 ms steps or 40% of th e h a lf-a m p litu d e length). An average
over 6 4 cycles w as fo u n d to be necessary, partly because of a slight " jit t e r "
of th e tacho pulses w h ic h could have been e lim in a te d by a more elaborate
set-up.
F ig .7 26 shows a 3-dim ensional representation of the results as a fre q u ency-tim e landscape. The m easurem ents were p erform ed autom atically us
ing a desktop ca lcu la tor to control th e analyzer, store th e results, and later
195
plot them on a digital plotter. The results as p lotted have a lin e a r frequency
scale, but th e calculator could be used to c o n ve rt them to c o n sta n t percen
tage ba ndw idth on a logarithm ic frequency scale.
7 .5 . B IB LIO G R A P H Y
BENDAT, J.S . and
PIERSOL, A .G .:
R andom Data: A n a ly s is
dures. W ile y, N.Y., 1971
RANDALL, R.B :
ROTH, O :
THRANE, N.:
THRANE, N.:
THRANE, N.:
UPTON, R.:
and M e a su re m e n t Proce
ZAVERI, K :
196
8.
8 .1 . BASIC C O N S ID E R A T IO N S
A ll m achines vibrate. In the process of ch a n n e llin g energy in to th e job to
be perform ed forces are generated w hich w ill excite the in dividual parts of
th e m achine d ire ctly or via the stru ctu re . Some of the parts in th e trans
m ission path are accessible from th e outside so th a t w e can e asily measure
the vibration re su ltin g from the excitation forces.
As long as th e process is co n sta n t or only varying w ith in certain lim its the
vibration m easured w ill be p ra ctica lly constant. Furtherm ore, it is fo u n d for
m ost m achines th a t the vib ra tio n frequency sp e ctru m has a ch a ra cte ristic
shape w hen th e m achine is in good condition. W h e n faults begin to develop
in the m achine some of the d ynam ic processes in th e machine are changed
and some of th e forces acting on parts - or the m echanical pro p e rties of the
parts them selves - are changed, thereby in flu e n c in g the vibration spectrum .
This is the basis for using vib ra tio n m easurem ents and analysis in machine
health m on itoring.
U nfo rtunate ly it is possible w ith a fe w w ro n g choices of param eters and
procedures to end up w ith system s unable to give reliable results. It is hoped
the fo llo w in g com m ents w ill clear up these m atters.
8 .2 . THE R E L A T IO N S H IP FORCE - V IB R A T IO N
As already m entioned, m achine vibration is u s u a lly measured at p oints ac
cessible on th e outside of the m achine. How does th is relate to w h a t is going
on inside?
In Fig.8.1 a rotor is exerting a varying force on th e bearings of a m achine.
As rotor forces are m ostly tra n s m itte d through bearings, th is is a p re fe rre d lo
cation for m easurem ents. These forces result in both a relative vib ra tio n of
the rotor to th e bearing housing and an absolute vibration of the w h o le bear-
197
Force
Level
^ V ib ./ F o rc e
Ay^ v/
k V ib. Level
A A
Frequency
Force
F re q ue ncy
M o b ility
Freq uency
Vibration
7 9 10 2 7
m easurem ent of the re la tiv e vibration b e tw e e n journal and bearing: the socalled S haft V ibration. Due to the lim ita tio n s in dynam ic range of relative d is
placem ent transducers (ty p ic a lly 100:1) o n e seldom o b ta in s m uch in fo rm a
tion above 3 x the run nin g speed, but fo r phenom ena m a n ife s tin g them selves
w ith in th a t range relative displacem ent m easurem ents are fin d in g a w id e
spread use (see B ibliography VDI 2 0 5 9 , API 6 7 0 ). See F ig.8 .3 .
The absolute vibration as measured on th e bearing ho u sing is often 2 -4
tim es sm aller due to the lo w e r m obility o f th e bearing h o u sin g , but because
of the m uch larger dyn a m ic range of th e accelerom eters ty p ic a lly used here
(1 0 0 ,0 0 0 ,0 0 0 :1 ) and th e p ossibility of s e le ctin g any of th e param eters: abso
lute acceleration, velocity, o r displacem ent, one is able to fo llo w the vibration
spectrum and any changes in it up to very h igh frequencies.
It is som etim es argued th a t because o f in sta b ility, com pliance and re so n
ances in oil film s it is im possible to use bearing vibration m easurem ents on
m any jo u rn a l bearings. If th e se problems w e re im portant th e oil film m o b ility
as seen by th e journal w o u ld not be sta b le and the relative m easurem ents
w ould s u ffe r even m ore, but in fact experience shows th a t both types of
m easurem ent can be m ade w ith good re p e a ta b ility.
Note th a t any vibratory fo rce s not a c tu a lly being absorbed in exciting the ro
to r m ust be transm itted to th e bearing h o u sin g . Should bearing loading be in
creased, th e oil film gets th in n e r (and s tiffe r) so that the re la tive vibration dec
reases. The decreased ro to r motion m e a n s th a t more fo rce is tra n sm itte d to
the bearing housing, w h o s e vibration th u s increases (as o ve ra ll m obility, d o m
inated by bearing supp o rts, remains p ra c tic a lly constant). A s higher loading
199
F g. 8 .4 . R olling e le m e n t bearing
w o uld te n d to accelerate bearing fa ilu re , th e bearing h o using vibration h ere
gives th e rig h t tendency.
In a ro llin g elem ent bearing the oil film is extrem ely th in as there is no re la
tive ve locity between th e surfaces at th e points of contact. The rolling e le
m ents are s tiff so the in n e r m obility and o u te r m obility are v irtu a lly the sam e.
Relative vibration betw een shaft and h o u sing is usually negligible even fo r
large vib ra tio n forces, so here the bearing housing vib ra tio n is quite d o m i
nant. A ccelerom eters w ill clearly give th e largest dyn a m ic range and fr e
quency range (Fig.8.3):
8 .3 . F R E Q U E N C Y R A N G E - D Y N A M IC R A N G E - P A R A M E T E R S
One of th e most im p o rta n t requirem ents of vibration in stru m e n ta tio n fo r
m aintenance is that it is able to register th e e n tire vibration spectrum in a s u f
ficie n tly w id e frequency range that all im p o rta n t com ponents are included.
This includes frequencies associated w ith unbalance, m isa lig n m e n t, ball-passing, gear-m eshing , blade-passing, blade resonances, bearing elem ent radial
resonances; often a fre q u e n cy range re q u ire m e n t of 1 0 - 1 0 ,0 0 0 Hz or m ore.
It is often fo und th a t th e h ig h er fre q u e n cie s contain in fo rm a tio n on fa u lts
developing w e ll before th e y influence the a ctual a bility o f th e machine to do
its job, w he re as the lo w e r frequencies s h o w the fa u lts w h e n they have o c
curred. To be able to predict breakdown, th e higher fre q u e n cie s therefore be
come very im portant.
The vib ra tio n level of m ost machines u su a lly varies w ith frequency. The
range in w h ic h the in stru m e n ta tio n m ust be able to present th e values w ith
an acceptable accuracy (the dynam ic range) m ust include all values w ith in
the freque ncy range (Fig.8 .5 ).
200
___________________________________________________________________________________________________________________________________________ 791030
791031
d raw n through the bottom of each spectrum it can be seen that th e relative
h eights of the peaks are the sam e. If th e vibration fo rce at any fre q u e n c y dou
bles th e n the m agnitude of the v ib ra tio n component doubles, in both param e
te rs, so both are useable for m a intenance purposes. However, P aram eter I
happens to give a ho rizo n ta lly a ligned spectrum w h e re b y the dyn a m ic range
requirem ents become th e sm allest. T his is therefore called the "B e s t Param e
te r " fo r such m easurem ents.
It can be fu rth e r noted that no sin g le com ponent in the spectrum o f Par
am eter I needs to increase by m ore than Range I to give a m easurable
change in the overall vibration level. For Parameter II the change o f some
com ponents must be equal to th e m uch larger Range II. Therefore if one a t
tem pts to use overall vibration levels for indication of m achine co n d ition ,
rath er than frequency analysis, it is imperative th a t one uses th e B est Par
am eter for such m easurem ents.
For m any m achines th e Best P aram eter is velocity and this is one reason
w h y m any standards (e.g. VDI 2 0 5 6 ) specify th is param eter. S tandards such
as VDI 2 0 5 9 relying on relative d isplacem ent m e asurem ent put m ain em p h a
sis on unbalance and m isa lig n m e n t and are forced to disregard large parts of
th e spectrum .
In F ig.8 .7 are sh o w n results fro m a machine w ith a fa u lt developing in one
com ponent. F ollow ing th e m achine w ith 5 periodic m easurem ents it can be
seen th a t the overall level m easured suddenly ju m p s up at the last m e a su re
m ent. Does this indicate th a t the m a ch in e is in the m iddle of a b reakdow n or
can it be operated fo r a nother period before the vib ra tio n level reaches th e li
mit? This cannot be determ ined fro m the overall levels. If we instead fo llo w
th e developm ent of th e affected frequency com ponent over the 5 periodic
m easurem ents w e can extrapolate w ith reasonable accuracy and fin d out
w h e n the vibration level w ill reach th e lim it, thereby predicting th e tim e be-
a)
Lin.
b)
Log.
10 0
a
10x
11
10
10x
}'
10x
;
10x
' i
10x
30
it
10
1
10x
io x ! f |
1 A ll lower values
T
.n n m
791033
8 .4 .
USE OF V IB R A T IO N M E A S U R E M E N T S FOR M A IN T E N A N C E
2.
3.
4.
8 .4 .1 . M aintenance System s
A. R un-to-B reak M a in te n a n c e
In in dustries ru n n in g m any inexpensive machines and having all im p o rta n t
processes duplicated, m achines are u s u a lly run u n til th e y break dow n. Loss
of production is in s ig n ific a n t and the spare m achines can take over in th e usuallly b rie f repair period. There is little advantage in k n o w in g when m achines
w ill break dow n, so vib ra tio n m easurem ents are ju s t used to check th e q u a l
ity of the repair.
In some industries (e.g. chemical plants) the product o fte n has such a high
value th a t there is a tendency to som etim es let even large unduplicated m a
chines run to break. In th is case it is valuable to k n o w w h a t is going w ro n g
and w h e n the fin a l breakdow n can be expected. This in fo rm a tio n can be ob
tained by analyzing spectrum trends fro m regular m easurem ents. H ow ever,
the consequential dam age resulting fro m such breakdow ns often greatly in
creases both the cost of repair and th e production loss du rin g the extended
sh u t-d o w n period. U sing th e condition m onitoring te ch n iq u e s described in
Section 8.4 .1 .C these extra costs can be considerably reduced, w h ile the p e ri
ods betw een shutd o w n s can still be kept to the m axim um .
B. Time B ased P reventive M aintenance
W here im portant m achines are not fu lly duplicated, or w h e re safety of p e r
sonnel is involved, m aintenance w ork is often perform ed at fixed tim e in te r
vals such as every 3 0 0 0 operating hours. The intervals are often determ ined
204
8 .4 .2 . M aintenance S taff
W hen vib ra tio n m easurem ents are introduced into a m aintenance system it
is most com m on to use personnel already in th e system. This is possible be
cause the skills required are easily learned and developed fro m experience,
and good in stru ctio n m aterial is available.
Two types of personnel are usually em ployed. One is the ope ra to r w h o w ill
perform th e actual m easurem ents fo llo w in g a fixed m easurem ent procedure.
In a fu lly developed scheme he has only to record the vibration signals w ith
preset in stru m e n ts and tu rn over the data in th e form of taped signals or
graphs of spectra to the m ain te n a n ce engineer at the end of th e day for later
analysis or evaluation. The m aintenance e n g ine e r is responsible fo r preparing
205
the m easurem ent procedure and often p e rfo rm s analysis and evaluation h im
self, p a rticu la rly during th e developm ent period.
The m easurem ent procedure should in clu d e w hich p o in ts on the m achines
should be measured in w h a t sequence, th e m anner in w h ic h the in stru m e n ts
should be used and th e desired ru n n in g conditions of th e machines. The
points on th e m achines should be prepared for easy a tta c h m e n t of the v ib ra
tion tra nsd ucer and m arked w ith a re fe re n ce number. The engineer should in
stru ct th e operator in th e purpose and th e execution of th e programme. It is
advisable to gain experience by m easuring many tim es on a small num ber of
m achines. This w ill also reveal the re p e a ta b ility of the procedure.
The engineer should also gather te c h n ic a l details to help him later d ia g
nose any fa u lts detected. These are d e ta ils such as ro ta tio n a l speeds, bearing
geom etry, num bers of ro llin g elem ents, g ear teeth, tu rb in e blades etc. This in
fo rm a tio n can be set up in a diagnostic reference sheet fo r each m achine. He
w ill also com pile available inform ation on acceptable v ib ra tio n lim its (relative
or absolute) from m a n u fa ctu re rs and o th e r organisations.
G e n e ra l Purpose
V ib ra tio n Meter
2511
A cce lero m e te r
F r l?
..
' *
8 0 0 434
General Purpose
V ib ra tio n Meter
2511
Accelerom eter
h r
j_
p ..
Accelerom eter
Charge A m p lifie r
2635
Accelerom eter
Tape Recorders
7003
EsE t
u >Ov W
Level Recorder
23 06
Accelerom eter
a ..
>
l<H f 'NI
P ortable V ib ra tio n A n a ly z e r
3513
%
Level Recorder
2306
A ccelerom eters
4369
Tape Recorder
70 03
Charge A m p lifie rs
26 35
C alculator
X - Y Recorder
23 08
8 .4 .4 . Permanent M onitoring
P erm anent m o n ito rin g is a system w h e re b y a set of in s tru m e n ts is c o n tin u
o u sly checking m achine condition at a lim ite d num ber of m easuring points.
208
M o n ito r
D ata Cable
Q
5674
*
Ju n ctio n B ox
rH
----
2505
791037
209
J u n c tio n Box
A ccelerom eter
i t~ i
~
Accelerom eter
5674
M o n ito r + 3 M ultip le xe rs
J u n c tio n Box
Accelerom eter
Narrow Band Spectrum
A n a ly z e r 2031
X Y Recorder
2308
Accelerom eter
5674
Ju n c tio n Box
C a lc u la to r
D isc M em ory
791039
8 .4 .5 . C ost Effectiveness
Equipm ent for vibration m easurem ent sh o u ld only be bought if it can be ex-
210
8 .5 . B IB LIO G R A PH Y
API 670:
D O W NHAM , E. &
W OO DS, R.:
I.S .V .R .:
Sou
VDI 2 0 5 6 :
VDI 2 0 5 9 :
211
9.
ACOUSTIC EMISSION
9 .1 . IN T R O D U C T IO N
M in u te im perfections in loaded s tru c tu re s and m a te ria ls , w hich co u ld lead
to u ltim a te fa ilure , can be revealed by several n o n -d e stru ctive te stin g (NDT)
m ethods in com m on use. Typical exam ples are m icroscopic and X -ra y inspec
tio n , strain m easurem ents and fla w detection by dye penetrants, eddy cu r
rents and ultrasonic transm ission or re fle ctio n .
These methods are common in one respect; th e y reveal im perfections
w ith o u t d iffe re n tia tio n between irre le v a n t "passive irre g u la ritie s w h ic h have
no influence on the a b ility of the s tru c tu re to perform its intended ta sk, and
" a c tiv e " developing fa u lts w h ich , if le ft, could se rio u sly threaten the in te g rity
of th e structure. R esults obtained fro m these tests re q u ire a high level of
judg em en t in d e te rm in in g th e ir s ig n ifica n ce . F urth e rm o re , for critic a l s tru c
tu re s, tests have to be made re g u la rly and extensively to ensure th a t new
fa u lts , fo r example caused by co rro sio n , overloading and w ear, have not oc
curred.
A coustic Emission (AE) is a re la tiv e ly new NDT tech n iq u e w h ic h differs
from th e above-m entioned techniques in th a t it is able to detect w hen a fla w
or a crack occurs and w here it occurs i.e. it detects " a c tiv e flaw s. Once the
existence of a fla w has been established by AE its n a tu re can be investigated
by the other methods. For critical s tru ctu re s, AE m e a su rin g in stru m e n ta tio n
can listen continuo u sly to w arn of fa u lts im m ediately th e y occur.
9 .2 . D E F IN IT IO N
AE (som etim es called Stress W ave E m ission SWE) can be defined as:
The elastic w ave generated by th e release of e n e rg y in te rn a lly stored in a
structure.
A lth o u g h this d e fin itio n includes m any kinds of w aves, (e.g. earthquakes and
21 2
9 .3 . AE S O U R C E S
AE "S o u rc e s " w hich can be described as processes e m ittin g elastic waves,
can be basically classified in 4 different groups:
1.
2.
3.
4.
1-10
5 - 1000
20 - 1 0 0 0
The energy released by a single dislo ca tio n m ovem ent (displacem ent of a
particula r type of line im perfection th ro u g h the crystal lattice) is norm ally too
sm all to be detected by AE equipm ent. H ow ever, many d islocations often co m
bine to fo rm an avalanche of m ovem ents giving rise to a co n tin u o u s A E-signal w h ic h can be detected. A typical phase tra n sfo rm a tio n source is m arten-
213
site fo rm a tio n in carbon steel; a single b u rs t signal can be detected for every
grain transform ed.
Crack fo rm ation occurs at surface notches or at p o in ts inside a m aterial
w h e re local stresses exceed the fra c tu re stress. Crack fo rm a tio n results in
the creation of new su rfa ce s, strain energy is released and th is is partly tra n s
form ed into AE signals. The AE signals generated by crack form ation are of
the burst type and are o fte n em itted at a ve ry high rate.
Friction also occurs in cracks, and th e sudden sliding m echanism releases
burst type signals. Crack form ation and fric tio n burst sig n a ls are very useful
fo r detecting and localizing cracks.
9 .4 . P R O P A G A T IO N
A E-sources behave in a m anner s im ila r to a radio a ntenna w ith a specific
radiation pattern for th e d iffe re n t wave types (shear-w aves and com pressionwaves); how ever the p o sition and properties of AE sources are only know n in
special rare cases.
If the source em its a spherical wave packet, it w ill o n ly be propagated as
such in an in fin ite iso tro p ic, hom ogeneous, ideally e lastic m edium . In real
s tru ctu re s th e propagation w ill be affected by surfaces w h ic h create re fle c
tions and form ation of surface waves (Rayleigh or Lamb). Inhom ogeneities
(e.g. w elds) may also create reflections and d isto rt w avefronts.
A n im p o rta n t example is waves propagating in a plate, especially on large
stru ctu re s like pressure vessels w here source location te ch n iq u e s are used.
These w aves are subject to dispersion as sh o w n in F ig.9 .2 . an and sn denote
the an tisym m e tric and s ym m e tric waves o f th e n th order.
It can be seen that the dispersion re la tio n sh ip s are com plicated, making it
d iffic u lt to predict the exact group velocity involved. F u rth e rm o re , dispersion
changes the w aveform w h ic h introduces am b ig u itie s w h e n tim e differences
have to be m easured fo r localization.
The above m entioned fa c to rs make it extre m e ly d iffic u lt to study the source
m echanism s except under conditions especially designed fo r the purpose.
H owever, research is s till being carried o u t in th is field, and th is w ill fu rth e r
enhance th e usefulness of th e AE technique.
It sh o u ld be noted, th a t in m ost o f the p re s e n t applications it is not neces
sary to have a detailed kn o w le d g e o f the source m echanism s a n d the propaga
tion. These applications in clu d e , for exam ple, the d e te rm in atio n of the typical
214
770 555
9 .5 . AE T R A N S D U C E R P R IN C IP L E S A N D C A L IB R A T IO N
W hen the em itted stress waves reach the tra n sd u ce r position via a propaga
tio n path as show n in Fig.9.3 (a to e) the s tre ss-stra in condition is converted
in to an electrical signal w hich can be treated by e le c tro n ic means.
Piezoelectric (PZ) transducers are by far the m o st w idely used. They are
m ostly undam ped, having very hig h se nsitivities at resonance. A sim p le pie
zoelectric disc w ith electrodes m o u n te d on its faces, perpendicular to th e pola
rization direction is often used fo r AE and ultrasonic transducers. For a plane
in fin ite disc adjo ining a stru ctu re and coupling (dam ping) m a te ria l, th e re
sponse to an incom ing plane w a ve w ith normal incidence can be calculated.
It m ust be stressed th a t this is a typ ica l situation fo r ultrasonic applications,
but n o t for AE signals w hich are in most cases m a in ly surface w a ve s, (Ray
leigh or Lamb).
U ltrasonic transducer calibration m ethods are th e re fo re not su ita b le fo r AE
purposes. This is illustrated in F ig .9 .4 . A PZ-disc, 1 ,9 mm thick and 9 mm in
diam ete r is m ounted directly on a steel structure. The thickness m ode can be
calculated to be 1 MHz, but a s u rfa ce wave of 1 MHz (A = 3 m m ) w ill not
215
give any o u tp u t as seen fro m the sym m etric deform ation s h o w n in the fig u re ,
w hereas a w ave corresponding to h / 2 = 9 m m , (f = 1 6 7 kHz) w ill couple
strongly.
To overcom e this problem th e disc can be made small; h o w e ver the capacit
ance of th e disc is then reduced considerably resulting in a m uch lower s e n s i
tiv ity w h e n loaded w ith e xte rn a l capacitances from cable etc. If a large w e a r
plate is used fo r coupling to the structure, it is d iffic u lt to e n su re coupling at
the point opposite the piezoelectric disc. If a sm all plate is used tiltin g may be
come a problem .
219
9 .6 . S IG N A L A M P L IF IC A T IO N
The transducer is fo llo w e d by a p re a m p lifie r - a m p lifie r com bination giving
up to 1 0 0 d B total a m p lific a tio n . The p re a m p lifie r stage, w h ic h has a lo w o u t
put im pedance and can th e re fo re feed long cables, m ay be b u ilt into the tra n s
ducer as show n in F ig .9 .5 . O therw ise a separate p re a m p lifie r, as sh o w n in
F ig.9 .8 , is placed near to the tra n sd u ce r. P re a m p lifie r gain is fixed (e.g.
4 0 dB) and the p re a m p lifie r may also be fitte d w ith bandpass filters.
For m eaningful com parison of p re a m p lifie rs, th e ir noise level should be
m easured and com pared w hen the in p u t is loaded by th e transducer im p e d
ance. If suitable m odern electronic com ponents are used, th e noise level w ill
be determ ined by th e th e rm a l noise of the piezoelectric disc. Because of
noise th e m inim um detectable AE a m p litu d e fo r a tra n s d u c e r in the 2 0 0 kHz
range is in the order o f 1 0 14 m at room te m perature.
The signal from th e p re a m p lifie r is fed in to a c o n d itio n in g am plifier w ith ad
justable gain, e.g. fro m 0 to 6 0 dB in 1 dB steps. F ig .9 .9 show s an a m p lifie r
w h ic h is especially s u ita b le for AE w o rk and w h ich has selectable fre q u e n cy
filte rs so as to suppress interference fro m low fre q u e n cy electrical noise and
m echanical vib ra tio n .
AE Preamplifier
Type 2637
9 .7 . D A T A A N A L Y S IS A N D P R E S E N T A T IO N
A variety of m ethods are available fo r analysis and presentation of th e a m
plified AE signal. These are sum m arised in Fig.9 .1 0 . Som e guidelines about
th e choice of the m ost suitable m easuring principle can be given as fo llo w s:
2 20
7 9 10 9 6
79 10 2 0
780722
9 .8 . A C O U S T IC E M IS S IO N S O U R C E LOCATIO N
M ultich a n n e l system s are used fo r AE source location, fo r exam ple on
large pressure vessels. W ith su ita b le control, location systems can d is c rim i
nate betw een signals coming fro m a certain area o f interest and s ig n a ls a rriv
ing from other areas.
W ith tw o transducers a source lo cation can be determ ined in one d im e n
sio n. W ith three transducers the source location can be calculated in tw o d i
m ensions. Usually an extra channel is added enabling th e system to a u to m a ti
cally ignore incon sisten t data.
The tim e diffe re n ce between th e arrival tim e o f the signal at tw o
transducers determ in es a hyperbola in a plane if th e propagation ve lo city is
kn o w n . The inte rse ctio n of hyberbolas obtained fro m other tra n s d u c e r pairs
de fin es the location o f the source. T h is is shown sch e m a tica lly in F ig .9 .1 4 .
224
9 .9 . FIELDS OF A P P L IC A T IO N
The study of plastic deform ation and crack form ation and extension are tw o
of th e freque ntly reported application areas of AE m easurem ent. F ig .9 .1 5
show s sketches of tw o typical results fro m plastic d e fo rm a tio n of test pieces.
________________________________________________________________________________________________________________________________________________ 7 9 0 4 5 2
Final te st
Pressure te s tin g of
vessels and pipelines
Heat tre a tm e n t
Hardening
C onstruction
Phase
transform ation
Bonding
Later Inspection
Proof testing
C ontinuous
m o n ito rin g
V essels and
p ip e lin e s
V essels and
p ip e lin e s
Bridges
B u ild in g s
b rid g e s
R otating
m a ch in e ry
M in e s
227
2 28
229
2 30
a linear scale w as used in the previous example. It can be seen th a t a c tiv ity
decreases w hen the pressure is kept con sta n t.
Fig. 9 .2 3 . Locations o f the sources o f em issions on the two ends o f the pres
sure vessel I folded out)
232
tio n , and at the crack tip. To be able to distin g uish between the tw o sources
of acoustic em ission, a theoretical analysis was firs t carried out to determ ine
the relative tim e of arrival at th e transducers, fo r signals o rig in a tin g at any
part of the test piece.
F ig.9 .2 5 (a) show s th e o re tica lly calculated regions w ith diffe re n t band n u m
bers corresponding to the various tim e differences. For practical m easure
m ents a pencil lead w as broken at various locations on the specim en to gen
erate an a rtificia l source and th e tim e differences fo r the signals to arrive at
the tw o transducer positions w e re measured. The results are s h o w n in
F ig,9 .2 5 (b) and are found to be in good agreem ent w ith the th e o re tic a l val
ues. The in stru m e n ta tio n set-up s h o w n in Fig.9 .2 6 w as used to sto re and dis
play the tim e differences.
The te st piece w as then clam ped and loaded above the expected fracture
load. The acoustic em ission w as m easured and fo u n d to be m ostly in channel
num bers 1 3 as show n in Fig 9 .2 7 (a) confirm ing th a t the activity w a s gener
ated at the loading points. To ensure th a t no defo rm a tio n and fric tio n occurred
at the loading points w hen the cla m p w as rem oved, the specimen w a s loaded
once again above the expected fra c tu re load w ith th e clamp on. F ig .9 .2 7 (b)
show s again the acoustic em ission in channels 13 and some in ch a n n e l 9.
The clam p was n o w removed and th e test perform ed by applying a suitable
load. The acoustic em ission m easured is show n in Fig.9 .2 7 (c). It can be
seen th a t most of it lies in ch annel 0 indicating th a t the activity is generated
at the crack tip.
233
_________________________________________________________________________________________________________________________________________________ 790456
234
Num ber
o f Signals
-100
-5 0
+ 13
Channel Number
-1 3
N um ber
I k o f Signals
-10
l.m
.........
+ 13
Channel Number
-1 3
Num ber
j ^ o f Signals
70
60
D urin g Test
50
40
30
20
-10
-1 3
+ 13
Channel Number
9 .1 1 . SELECTED B IB LIO G R A PH Y
ASTM:
ASTY, M .:
BECHT, J .,
EISENBLTTER, J.
and JA X , P.:
BOLIN, L.A .:
DROUILLARD, T.F.:
H A T A N O ,H .a n d
MORI, E.:
KAISER, J.:
NICHOLS, R .W ., (ed.):
SCHLEGEL, D.,
RUNOW P. and
FINK, W .:
TOBIAS, A.:
VIKTOROV, I.A.:
Proceedings:
236
10.
1 0 .1 . V IB R A T IO N TES TIN G
One of the e arliest w orkers to use a system atic fo rm of vib ra tio n testing
w a s W ohler, w h o used special apparatus for his experim ents on th e fatigue
o f m etals over 1 0 0 years ago. H ow ever, most m odern vibration te s tin g te c h
n iques have been developed since th e start of W orld W a r II.
A t th a t tim e it w as im portant to develop parts and equipm ent capable of
w ith sta n d in g the service e n v iro n m e n t in aircraft. Problem s w ere e ncountered
not only w ith m echanical s tru ctu ra l fa ilu re , but also w ith sophisticated elec
tro n ic and electrom echanical in s tru m e n ta tio n and co n tro l systems, w h o se per
form ance and re lia b ility was se n sitive to the vib ra tio n encountered in service.
F urtherm ore, it w a s appreciated th a t theoretical pre d ictio n of the v ib ra tio n re
sponse of a piece of equipm ent in service could be e xtrem ely d iffic u lt.
The developm ent of such te st te ch n iq u e s has ta ke n place p rim a rily in the
aerospace in dustrie s, but vib ra tio n te stin g has a m uch w ider a p p lica tio n to
day in other areas such as the a u to m o b ile , c o n stru ctio n , electronics, m achineto o l, packaging and ship building in d u strie s. The m ost common uses are for:
1.
2.
3.
Production C ontrol
Frequency Response / D ynam ic Performance Testing
E nvironm ental Tests
se lf. F urtherm ore voltage analogues of frequency and vibration variables and
m echanical para m e te rs can be used readily to p rovide hard copies o f test re
su lts using a n c illa ry recorder or p lo tte r equipm ents.
1 0 .1 .1 . The Exciter
A ny vibration te s t system requires a device w h ic h can be activated to sub
je ct the test o bject to the m echanical m otion re q u ire d . This tra n sd u ce r is the
exciter. Useful o p e rating regions fo r the tw o p rin c ip a l exciter te ch n iq u e s are
show n in Fig.10.1 .
0,1
10
100
10 0 0
10000
Frequency Hz
267002/1
M ovin g table
Upper
790502
Fre q u e n cy Hz
7g0s04
Measuring O b je c t:_______________
B ru e l &
K j r
Hz
10
20
50
R actjfia r.
100
DC
200
500
_H i
1000
2000
W riting Speed
5000
10000
m m /ie c .
20000
Paper Soaeri:
50000
100000
m m /i* < 7 6 0 5 4 5 /t
Fig. 10.4. Exam ple o f the a cce le ra tio n ch a ra cte ristic o f a p ra c tic a l excite r re
corded as a fu n ctio n o f frequency fo r c o n s ta n t drive voltage
240
tro lle d , the acceleration level available a w ill be lim ite d by the rated fo rce of
th e exciter F (N), th e mass of the te s t object M (kg), and the mass of th e m ov
ing elem ent M e (kg), such that:
F
ms 2
(1 0 .1 )
M + Me
The nom inal force rating for a given exciter design w ill be d ependent on
th e m axim um c u rre n t capability o f th e m oving co il. T h is is seen fro m th e ba
sic relationship
F = [B L ]I
(1 0 .2 )
w h e re
F = force in N
B = m agnetic flu x d ensity in air gap in W b m 2
/ = c u rre n t in m oving coil in A
L = length of conductor in air gap in m
The product BL is called the "h e a d c o n s ta n t" as it is fixed for a given de
sign. It has fu rth e r significance in consideration o f the "d y n a m o " p e rfo r
m ance of the m oving coil, w h ich produces a back e .m .f. according to:
Eb
a c k
= [ B L ]
(1 0 .3 )
w h e re
~ back e m -f- across m oving coil in V
v = velocity of conductor in air gap in m s1
eB A C K
Frequency
vibration
exciter
to g e th e r
w ith
interchangeable
F re q u e n cy (Hz)
264479
20
50
100
2 00
500
100 0
200 0
5000
F re q u e n cy (Hz)
790 50 3
1 0 .1 .4 . Sinusoidal Excitation
W hen it is re qu ire d to m aintain a constant v ib ra tio n level w ith a sinusoidal
drive signal, th e o u tp u t level fro m a vibration sensor mounted on the w ork
table may be used in a servo loop to control th e input level to th e exciter
(Fig. 1 0.9). In th is w a y the e ffe cts of the system resonances are regulated by
the control sensor, w h ils t behaviour of the te st objects is studied using a re
sponse accelerom eter connected th ro u g h a p re a m p lifie r to a m easuring a m p li
fie r or a graphic level recorder, or using a stroboscopic motion analyzer.
The circu it used to regulate th e excitation sig n a l is known as a com pressor,
and the servo-loop of w hich it fo rm s a part is k n o w n as a com pressor loop.
Fig. 10.9. Exam ple o f a se rvo -co n tro lle d sw e p t-fre q u e n c y test a rra n g em e n t
244
E xcite r C o ntrol
1047
Pow er A m p lifie r
2707
V ib ra tio n E xcite r
4 8 0 5 + Head
______________________ 780264/1
very selective filte rin g of the co n tro l signal reduces th e effective regulation
speed and the s ta b ility of the servo loop.
The most co n ve n ie n t vibration sensors for both co n tro l and response meas
urem ent are piezoelectric accelerom eters. This type of device generates an
e lectrical output signal proportional to instantaneous acceleration. If th is sig
nal (suitably am p lifie d by a c o n d ition in g a m p lifie r) is fed to the com pressor,
th e n it is the m ean acceleration level of the te st object w hich is kept con
sta n t. However, th e signal may, if required, be integrated w ith respect to
tim e , using an e le ctro n ic in te g ra to r. One stage of inte g ra tio n gives a signal
proportional to instantaneous velocity; a second gives a signal proportional to
instantaneous displacem ent. Thus th e incorporation o f one or tw o integrators
in th e com pressor loop perm its re g u la tio n of ve lo city or displacem ent respec
tive ly. If the effects o f m echanical or electrical noise cause the signal to be
n o n -lin e a r, filte rin g m ay be required.
It is som etim es required to cross-over from one required variable to
an o th e r as recom m ended in certain standardised vib ra tio n test program m es,
e.g. IEC 6 8 -2 -6 Test F. A uto m a tic fre q u e n cy-co n tro lle d cross-over sw itch in g
is then necessary (Fig. 10.1 1 ).
F re q ue ncy
V ib ra tio n
Program mers
Z H 0100
2626
I
I
1 ^ 1 ?
11 *
Fixture
C o n d itio n in g
A m p lifie r
1 "
'
E xcite r C ontrol 1047
Drive
I
?
*'
I
'
*
Signal
Pow er A m p lifie r
2707
V ib ra tio n E xcite r
4801 + Head
^Accelerometers
e.g. 4371
V ib ra tio n Programmers
Z H 0100
EB: * :E3
C o n tro l
Signals
C o ntrol
Signal
Charge A m p lifiers
2651
E x c ite r C o n tro l
1047
2307
nm
Level Recorder
Records actual
test level
T w o Channel
Tracking Filte rs
5716
Control Signal Selector
______ ^ _________
*
Pow er
S u pp ly
2805
W hen large, com plex test specim ens are bolted to the w ork ta b le , control
from one point may not be s u ffic ie n t, ow ing to irre g u la r m otion in diffe re n t
parts of the test object. It is desirable in such cases to be able to select the
control point required, or to average over a nu m b e r of points, in order to
avoid over-testing (Fig.10.1 2).
W here it is required to perform vib ra tio n tests up to high fre q u e n cie s, and
a very low exciting-force is acceptable, it is in some situ a tio n s advantageous
to use a piezoelectric accelerom eter as an exciter (Fig. 1 0.1 3). This is feasible
because the accelerom eter is a passive device obeying the re cip ro city th e o
rem . The u n ifo rm ity of its fre q u e n cy response as a sensor can be exploited
w h e n it is used, in reverse, as an exciter, to achieve excitation w ith o u t a re-
24 7
2713
Pow er A m p lifier
248
1000
100
O'
0
4 A
I <
o$
A
.
*
0 0
* a<*
A*
10
oI
100
1,0 k
Frequency (Hz)
Fig. 1 0 .1 4 . Exam ple o f the d e riv a tio n o f vib ra tio n test level specifications
from e n viro n m e n ta l data
W hen a random type of in p u t signal is applied to a vibration te s t system,
th e effect of system resonances can be avoided by using a su ita b le spectrum
shaper, consisting of a bank of p a ra lle l band-pass filte rs w ith in d iv id u a lly ad
ju stable atte n u a tio n (Fig. 1 0 .1 5 ). A real-tim e fre q u e n cy analyzer can be used
to compare the re su ltin g e xcita tio n spectrum w ith th e desired e x c ita tio n spec
tru m , and the atte nuato rs of th e spectrum shaper can be adjusted to achieve
th e spectrum desired. The same arrangem ent can be used to "lin e a riz e " the
response of the exciter system , using a random noise source as a reference,
w h ic h is replaced subsequently by a tape recorder to reproduce th e measured
e nvironm ent at th e w o rk table.
This system has the disadvantage th a t it is passive and provides no control
or regulation of th e test w h ils t it is in progress. Broad-band e q u a lize r / anal
yzer systems w h ic h include com pressor loops fo r many n a rro w band fre
quency com ponents have been m anufactured using analogue e le ctro n ics, but
th e y tend to be com plex and bu lky ow ing to th e large num ber o f electronic
com ponents used. C urrent g e n e ra tio n systems m ake extensive use of digital
techniques, w h e re com parison o f th e measured sig n a l w ith the d e sire d signal
a llow s the required signal to be synthesized.
A vibration testing technique w h ic h eases the problem s associated w ith the
249
P iezoelectric
A cccle ro m e te r 4 3 7 0
C o n d itio n in g
Pream plifier 2 6 3 5
D ig ital F re q ue ncy
A n a ly z e r 21 31
P iezoelectric
P ie zo e le ctric
A cce lero m e te r
4370
V ib ra tio n Meter
2511
Frequency
synchro nization
X -Y R e co rd e r
2308
and
Z = i/
(M echanical Im pedance)
(10 4)
(M echanical M o b ility )
(10.5)
w h e re F is the com plex force vector, and i/ the com plex velocity vector. Fur
th e r definition s apply to tra n s fe r m obility, w hen th e response is m easured at
a different point to th e applied fo rce .
Point Impedance
----- !----- /
Transfer Im pedance
-----
0
800374/1
(10.6)
= - 1
(10.7)
/ tom
'
(10.9)
w he re :
a = a0e i (Ot
v =
d =
m =
k =
- a cce le ra tio n
(1 0 .1 0 )
Mc=
(1 0 .1 1 )
(+)
(+) +j
+i
i^
1I 1 k
i kjwm
X
<
>
_c
1/c
<
>
c
05
E
c
Real Axis (+)
_k_ _ jk
' 1jcj
(jJ
jtom cjm
(-)
(-) -j
-j
(a)
271634
(a)
(b)
790500
Charge
Am plifier 2651
Force
Input
Acc. Input
w
Charge
Amp.
2651
O u tp u t
Mass
Compensation
Unit
5565
Measuring
Amplifier
2609
Velocity
Output
FI
J ii
Power Supply
2805
Compressor
Input
Measurinq Am plifier
2609
Frequency synchronization
| Sine Generator
1023
Power Am plifier
2706
m
Level Recorder
2307
Fig. 1 0 .2 0 .
F ig . 1 0 . 2 1 .
P o i n t i m p e d a n c e m e a s u r e m e n t s o n t h e c u t t in g t o o l o f a la t h e :
(a) T o o l in n o r m a l p o s i t io n
254
(b) T o o l in o u t e r p o s i t i o n
the object through an im pedance head (F ig.6 .24). This device incorporates
tw o sensors, one d e live rin g a signal p ro portion to the fo rce being applied to
the test object, and th e second m easuring the m otion o f th e point at w h ic h
the force is being applied. W here larger forces are re q u ire d to be tra n sm itte d
it is best to separate th e fu n ctio n s by using separate fo rce and m otion s e n
sors. S ince the m otion tra n sd u ce r is u s u a lly an accelerom eter, th e signal o u t
put m ust be integrated e le c tric a lly to produce a velocity s ig n a l.
For m echanical im pedance m easurem ents the velocity level of the m e a su r
ing point is kept co n sta n t, w hereby a recording of th e fo rce level indicates
the m odulus of the im pedance a s |Z | = | f i | / | v | (Fig. 1 0 .2 0 ). This representa
tion is p a rticu la rly u sefu l w h e re properties of sub-system s m ust be com bined
to give the impedance m a trix of a com plete system. F ig .10.2 1 shows a ty p i
cal application of m echanical impedance m easurem ent.
S im ila rly, to m easure th e m obility, th e driving force sh o u ld be kept c o n
stant and the velocity level recorded as \ M \ = \ v \ / \ F \ . This representation is
pa rticu la rly useful w h e n m odal param eters of a system are to be found.
The phase difference betw een force and velocity sig n a ls can be m easured
by m eans of a phase m e te r. W here sig n a l harm onics or extraneous noise in
terfere w ith m easurem ents, tracking filte rs can be used (F ig .1 0 .2 2 ).
A = Response signal
B = Fo rce co n tro l signal
A cce lero m e te r
4370
Force
\
J
Transducer
A d d itio n a l
measuring
Accelerom eter
p osition s
Charge
A m p lifie r
2635
m
T w o C hann el
T racking F ilte r
57 16
: < frs |
'
m
i
F ig . 1 0 .2 2 .
- -
Response
Signal
Phase Meter
2971
Force
C o n tro l
Signal
Pow er
A m p lifie r
2708
V oltm e te r
2425
Frequency
tracking
signal
E x c ite r C o n tro l
1047
E x a m p le
Phase difference
L B - A
G raphic
Level Recorder
2307
<?
##
o f a rra n g e m e n t
f o r m a k in g m o b i l i t y m e a s u r e m e n t s o n
t h e r u d d e r o f a s h ip
255
1 0 .2 .1 . Laboratory Testing
In general, it is im practicable to reproduce th e actual shock e n vironm ent in
a test It is better to ensure th a t the effects of the test shock upon the tested
equipm ent are sim ila r to th o se of the shock(s) occurring in p ra ctice , and to e n
sure that these shock e ffe cts are com pletely reproducible. The latter is of
prim e im portance w hen com parisons are m ade betw een te s t re su lts obtained
at diffe rent in s titu tio n s , or w h e n different product designs are evaluated.
One m ethod of specifying a
eration of th e shock pulse as
27), together w ith a llo w a b le
change produced by the pulse
Fig. 10.23.
256
___________________________________________________________________________________________________________________________________ 2 7 1 5 3 5
Fig. 1 0 .2 4 .
169094
Fig. 1 0 .2 5 .
2 6 90 66
Fig. 70 .2 6 .
25 kH z
Fre q u e n cy
271534
Fig. 1 0 .2 7 .
F ig .1 0 .2 8 .
Fig. 10.29 .
260
plex in operation. A s in the case o f broad band v ib ra tio n test system s, anal
ogue signal generators have la rg e ly been superseded by digital system s in
th is field.
O scillatory shock m otio n s, such as those described above, are o fte n term ed
com plex shocks. W h ile the p o ssib ility of producing a large variety o f complex
shocks is offered by a shock syn th e sis system, som e "s im p le r m achines do
exist w h ich produce specific com plex shocks. These are in general designed
according to the p e n dulum p rin cip le (F ig .1 0.30).
Fig. 1 0 .3 0 .
1 0 .2 .2 . Service Testing
The transport m edium or packaging method has to be specified fo r a piece
of vulnerable e qu ipm en t, so th a t it w ill not be dam aged by shocks occurring
d u rin g shipping or h a n dlin g . This s itu a tio n m ight a rise in the d e live ry of con
signm ents of delicate products su ch as fila m e n t devices, as w e ll as fo r the
261
790779
Fig. 1 0 .3 1 .
262
Fig. 1 0 .3 2 .
assem bly may be packaged together. The recorder may m o n ito r the progress
of the cargo over several w eeks, p rin tin g o u t on an in te g ra l p rin te r the tim e
and shock value on each occasion the preset threshold level is exceeded
(Fig.1 0 .3 3 ). The inclusion of an integrator in the device enables the shock v e
locity values to be obtained fo r use in th o se cases w h e re th e consignm ent is
pa rticu la rly sensitive to im pact velocity. A m uch more precise assessm ent of
the m axim um shocks o ccu rrin g in service and th e ir tim e of occurrence can
thus be obtained using an apparatus of th is kind.
D< V
y
V e loc ty (m/s)
H o ur
M inute
I
Acceleration (m /s2 )
|
7 6 0 6 1 4 /1
Fig. 1 0 .3 3 .
263
1 0 .3 . S ELECTED B IB LIO G R A P H Y
ANDERSON, D.:
BOOTH, G.B.:
BROCH, J.T.:
BROCH, J.T.:
BROCH, J.T.:
BROCH, J.T.:
BROCH, J.T.:
BROCH, J.T.:
BROCH, J.T.:
BROOKS, G .W . and
CARDEN, H.D.:
Bruel
Bruel &
&
Kj r
V ibration".
J.
CRANDALL, S.H. et al. : "R a n d o m V ibration". Jo h n W iley and Sons, Inc. New
York 1 9 5 9
CRANDALL, S.H. et al.
"R a n d o m
Vibration
M ass. 1 9 6 3
DAS, B.M.:
264
II".
MIT
Press,
Cambridge,
EWINS, D.J.:
FRANKLAND, J .M .:
IEC:
IEC:
IEC:
ISO DRAFT
INTERNATIONAL
STANDARD 5 3 4 4 :
JORDAN, J.C.:
KEEFFE, R E. and
BATHKE, E.A.:
KITTELSEN, K.E.:
LEVENSON, M. and
SUSSHOLZ, B.:
M cG r a t h , m b .:
M cG ra w -H ill
Book Com
265
METZGAR, K.J.:
MONROE, J.:
MORROW, C.T.:
MORROW, C.T.:
MORROW, C.T.:
MORROW, C.T.:
MORROW , C.T.:
M 0LLER
PETERSEN, P.E.:
OLESON, M .W .:
OTTS, J.V.:
PIERSOL, A.G..
PIERSOL, A.G.:
REGIER, A .A . and
HUBBARD, H.H.:
266
V ibration".
V ibration".
Noise
C ontrol.
V o l.4
SPANG, K.:
STAHLE, C.V.:
S0RENSEN, O B.:
TROTTER, W .D.:
TRULL, R.V.:
VIGNESS, I.:
VIGNESS, I.:
WITTE, A.F.:
WITTE, A.F.:
267
11.
1 1 .1 . IN T R O D U C T IO N
Unbalance in ro ta tin g m a ch in e ry has become an increasingly im p o rta n t fac
to r in the developm ent of m odern equipm ent especially w here th e needs for
speed a n d /o r re lia b ility are h ig h . Techniques fo r balancing su ch equipm ent
have advanced s ig n ifica n tly in recent years, e n a bling both p ro d u ctio n and
fie ld balancing to be executed system atically and w ith a m in im u m o f fuss and
experim entation. Balancing o f m achines is im p o rta n t to prevent fa tig u e fa i
lure in associated structures, to prevent excessive loading of su p p o rt bear
ings, to prevent transm ission o f excessive e xte rn a l noise and v ib ra tio n , and
to improve the d u ra b ility and u se fu ln e ss of the m achines in service.
1 1 .2 . U N B A L A N C E OF R IG ID R O TO R S
The sim plest case of unbalance can be considered for a u n ifo rm th in disc
of mass M (kg), ro ta tio n a lly sym m etrical about its axis o f rotation. W hen a
small m ass m (g) is fix e d to the disc at a distance r (m m) fro m that axis, th e
disc is said to be in a s ta te of "u n b a la n c e . This type of unbalance can be d e
tected by supporting th e axle on a p a ir of knife edges to fin d the "h e a vy
spot": fo r th is reason it is often called static (or sin g le -p la n e ) unbalance
F ig .1 1 .1(a)). The c e n trifu g a l force F generated by the uncom pensated mass,
w hen the disc rotates a speed
(ra d /s), is given by
10
F = mra>2*
H 1-1)
(1 1 .2 )
(1 1 3 )
It is also possible to e lim in a te F betw een (1 1.1) and (1 1 .2) to see that the e c
centricity term e in m a ls o represents th e specific unbalance
of the rotor in
g -m m /kg :
* F, r have a d e fin ite line of a c tio n w ith respect to th e geo m e try of th e ro to r, a nd are th u s d en o te d
as vector q u a n titie s ,
cj
= 2 n n / 6 0 * n /1 0 , w h e re n is ro ta tio n speed in r / m i n .
269
F =
/= 1
M any practical m achines, such as g rin d in g w heels, in d u s tria l blowers and
flyw he els, can be considered as though th e ir mass w ere concentrated in a
single disc, but the m a jo rity of rotors have mass d is trib u te d along th e ir
length. This gives the p o s s ib ility of a second form of u n balance. For the case
of a rig h t cylinder, it is possible to have tw o equal uncom pensated masses
sym m e trically placed about th e centre of m ass, but positioned at 1 8 0 to one
another. The rotor is in s ta tic balance, yet centrifugal fo rce s w ill produce a
m om ent about the centre of mass w hen th e rotor tu rn s. T h is type of u n b a l
ance is called couple u nbalance (Fig. 11 1(b)), and it results in a tiltin g or pend
ulum action o f the p rin cip a l in e rtia axis about the shaft axis at the centre of
mass. To counteract th e couple it is necessary to make co rre ctio n s on tw o
planes.
The general condition w h e n both static and couple unb a la n ce are present
is called dynam ic unbalance (F ig . 1 1.1(c)). The principal in e rtia axis is now in
clined to th e geom etric s h a ft axis, but th e re is also an e c c e n tric ity at the c e n
tre of mass. This unbalance condition can be resolved by s u ita b le in s tru m e n
tation for correction in tw o planes. C orrection is made by m ass addition (w e ld
ing, rivets, etc.) or mass rem oval (boring, p laning etc.), using a n cilla ry eq u ip
m ent.
1 1 .3 . R O TO R SUPPO R T S Y S T E M
In an assembled m achine, th e rotor is supported by a b e a rin g and base as
sembly. This m ust restrain th e unbalance-excited motion of th e rotor jo u rn a l.
For the case of a single b e a rin g , the ro to r/s u p p o rt assembly m ay be modelled
as a single degree of freedom system . The d iffe re n tia l e q u a tio n o f motion can
be expressed:
M x + c x + kx = m r w 2 sin cot
For sinusoidal motion the displacem ent x can be w ritten:
x - x 0 sin (tu t - <j>)
giving the solution:
270
(1 1 .5 )
and
= ro ta tio n speed in r / m in
R ig h t Support Plane
Unbalance 2
Unbalance 1
'L
7T7T7T7TT
'R
7ft7777777'
7905 19
Fig. 1 1.3. Unbalance of a rotating body vibrating with two degrees of freedom
*L 1 (XL2
_a /?7 a/?2
1 1 .4 . S E TTIN G THE S T A N D A R D S
Ideally, a com pletely balanced m achine w ould s h o w no unbalance at all. In
practice, th o u g h , ow ing to m a ch in in g tolerances, m echanical play, run-out,
m isalignm ent d isto rtio n etc. perfect balance can never be achieved. In any
production or m aintenance s itu a tio n an appropriate residual unbalance, "b a l
ance q u a lity " or vib ra tion tolerance m ust be selected w h ich depends on the
perform ance required from the m achine and the econom ics of th e balancing
process.
In the fiftie s m uch w ork w as done in W est G e rm a n y to collect th e experi
ences of engineers w o rkin g in th is field. VDI 2 0 6 0 , "B e urteilungsm a stbe
f r den A usw uch tzusta nd ro tie re n de r, starrer K rp e r", has now been adopted
in te rn a tio n a lly as recom m endation ISO Standard 1 9 4 0 , "B alance Q uality of
Rotating Rigid B o dies". The recom m endations re la te acceptable re sidual un
balance to the m axim um service speed of the ro to r, and associate various
types of representative rotors w ith ranges of recom m ended q u a lity grades
(Figs. 1 1 4 and 1 1.5). The q u a lity grade, G, (e q u iva le n t to the p ro d u ct ew for
an unrestrained rotor) is introduced, as it enables th e physically observed be
haviour of m achines ru nn in g at d iffe re n t speeds to be compared. The values
of G in the Standard are n u m e rica lly equivalent to th e eccentricity e in /um for
a rotor runn ing at 9 5 0 0 RPM. The q uality grade, o r unbalance, o f a rotor can
be assessed using a calibrated balancing machine.
For m achines in service, unbalance vibrations are influenced considerably
m <2)
mm/sec
G 4000
4000
Chrankshaft-drives
o f cy lin d e rs (4).
G 1600
1600
G 630
630
G 250
250
G 100
100
Crankshaft-drives-of fast diesel engines w ith six and m ore c y lin d e rs <4>. C o m p le te engines
(gasoline o r diesel) fo r cars, tru c k s and locom otives <5>.
G 40
40
Car w heels, wheel rims, w heel sets, d rive shafts. Crankshaft-drives o f ela stically m oun ted
fast fo u r-cy c le engines (gasoline o r diesel) w ith six and m ore cylinders*4). C ran ksha ftdrives fo r engines o f cars, tru c k s and locom otives.
G 16
16
Drive shafts (propeller shafts, card a n shafts) w ith special requirem ents. Parts o f cru shin g
m achinery. Parts of agricultu ra l m achinery. Individual com p on e nts o f engines (gasoline
or diesel, fo r cars, trucks and lo com otiv e s. Crankshaft-drives o f engines w ith six and
more c y lin d e rs under special requirem ents.
G 6.3
6.3
Parts o f process plant m achines. M a rin e main turbine gears (merchant service).
C e ntrifug e drums. Fans. A sse m b le d aircra ft gas turb in e rotors.
F ly wheels. Pump im pellers. M a c h in e -to o l and general m achin ery parts. N orm al
e lectrical armatures. Ind ivid ual com p on ents o f engines u n d e r special requirem ents.
2.5
G 2.5
G 1
G 0.4
0.4
o f rig id ly m oun ted slow m arine diesel engines w ith uneven num ber
Tape recorder and phonograph (gram ophone) drives. G rin din g -m ach ine drives.
Sm all e lectrical armatures w ith special requirements.
Sp in dles, discs, and arm atures o f p re cisio n grinders. Gyroscopes.
Notes:
1.
co = 2 7r n/60
2.
In general, for rigid roto rs w ith tw o co rre ctio n planes, one half of the recom m ended residual unbalance
is to be taken fo r each plane; these values a p p ly usu a lly for any tw o a rb itra rily chosen planes, b u t the
state o f unbalance m ay be im proved upon at the bearings. For disc-shaped rotors the fu ll recom m ended
value holds for one plane.
3.
A crankshaft-drive is an assembly w hich in clu d e s the crankshaft, a fly w h e e l, clutch, p ulley, v ib ra tio n
damper, rotating p o rtio n o f connecting rod, etc.
4.
F o r the present purposes, slow diesel engines are those w ith a piston v e lo c ity o f less than 9 m/s; fast
diesel engines are those w ith a piston v e lo c ity o f greater than 9 m/s.
5.
In com plete engines the ro to r mass com prises th e sum o f all masses b elonging to the cranksh aft-d rive
described in fo o tn o te 3 above.
800624
274
displacement e in pm
per unit of rotor mass in ------------------------ or centre-of-gravity
residual unbalance
Acceptable
1.6
2.5
10
16
25
100
160 25 0
500
1000
Hz
1600
800778
275
45
Not perm issible
28
0)
>
w
18
N o t perm issible
Just tolerable
11'2
7,1
Just tolerable
4,5
2,8
Just tolerable
A llow able
0,28
0,18
Good
Good
Good
0,71
0,45
A llow able
A llo w ab le
1,8
1,12
A llow able
Just tolerable
Good
Sm all machines, up to
15 kW.
M edium machines
1575 kW o r up to
3 0 0 kW on special
foundations.
1 1 .5 . B A LA N C IN G M A C H IN E S
A dynam ic balancing m achine consists o f a bed assem bly (Fig. 1 1.7) and an
associated m easurem ent u n it (F ig.1 1 .8 ). Instrum ented support pedestals
carry th e rotor to be balanced, w hich is driven at co n sta n t speed by a m otor
276
the
;a * r
m,
99
S i l l
Fig 1 7 8 A n example o f the console fo r a dynamic b a la n cin g m achine, B & K
Type 2 5 0 4
and drive system. M ost tw o -p la n e m achines operate w ith the rotor axis h o riz
o n ta l, and are described as ''u n iv e rs a l". Typically, d rive to the rotor is p ro
vided via an axial cardan shaft or a c irc u m fe re n tia l belt arrangem ent, de p e n d
ing on th e size and specification of th e balancing re q u ire m e n t. The support
pedestals can be set at any co nvenient position along th e bed to suit d iffe re n t
roto r geom etries w ith in the mass ra nge of the m achine, w h ils t the m e a s u re
m ent head, or console, is designed to accomm odate an y likely co m bination of
m easuring and correction planes presented to it. The electrical signals
sensed at the supports are analysed in the console to display d ire c tly the
am o unt and angle of unbalance to be corrected on th e tw o correction planes
selected for the rotor. The balancing bench and the console com plem ent one
another; if they are used correctly, an unbalance re d u ctio n ratio of 8 0
90% m ight be typical fo r a single ru n .
277
1 1 .6 . FIELD B A L A N C IN G
Som etim es access to a balancing m achine is not available; th e rotor to be
balanced is too large to suit a balancing m achine, or the rotor m u st be bal
anced in its no rm al service c o n d itio n s. In these cases " fie ld " or " in - s itu " bal
ancing can be carrie d out using suitable p o rta b le instru m e n ts. Generally,
such instru m e nts require more k n o w -h o w on th e part of the user th a n a bal
ancing m achine, but are more ve rsa tile in a p p lica tio n ow ing to th e ir portabil
ity and the range o f d iffe re n t tra n sd u ce r s e n s itiv itie s w hich can be used w ith
278
single plane is to use a sim ple vib ra tio n meter connected to an accelerom eter
m ounted on the bearing. First th e in itia l unbalance is measured. Then a trial
mass is used to introduce a k n o w n unbalance by attaching it to th e rotor at
th e same radius to be used fo r th e fin a l co rre ctio n mass. Three te s t runs are
carried out w ith th e tria l mass placed at 0 , 1 2 0 , 2 4 0 on th e ro to r. Geom
e trica lly, S ie b e rt's co n stru ctio n (F ig . 1 1.9) can be used to evaluate th e correc
tio n values. Three vectors of eq u a l length, corresponding to in itia l unbalance
V0 , at 0 , 1 2 0 , 2 4 0 respectively, are draw n o u t from the o rig in . Vectors
corresponding to V j (0), V j (1 2 0 ), V j (2 4 0 ) are constructed by centring a
com pass point on each of th e V0 vectors in tu rn : th e point o f in te rse ctio n of
these arcs enables the vectors corresponding to the trial m ass alone to be
constructed. The correction m ass can be calculated directly.
Usually it is im practicable to ca rry out three tria l runs; fu rth e rm o re it is ne
cessary to use a filte r synchronised to rotation speed, to ensure th a t the u n
balance com ponent of the vib ra tio n signal can be isolated from o th e r m echan
ical influences. The inclusion o f som e form of phase-m easuring device in the
system enables a m ore practical procedure to be im plem ented. O ne way of
d eterm ining th e phase is to tape o r mark a scale graduated in a n g u la r units
on the rotor, and illu m in a te th e scale during th e tria l balancing ru n s w ith the
lig h t from a stroboscope triggered by the filte re d vib ra tio n signal. A n example
o f such a system is show n in Fig .1 1 .1 0 .
P urpose-built fie ld -b a la n cin g sets, how ever, e m p lo y a non-contact tachom e
te r probe to trig g e r an a ll-e le c tro n ic phase in d ic a to r from a s in g le arbitrary
m ark on the ro tor. The example s h o w n in Fig. 1 1 .11 is designed to be equally
suitable for m o n ito rin g and analysis tasks as m e n tio n e d above.
For single-plane balancing, such as fo r a g rin d in g w heel, an in itia l reading
o f vibration a m p litu d e , \ V0 \, and phase angle, a 0 (w ith reference to a fixed
Fig. 1 1 .1 0 .
280
Fig. 1 1 .1 1 .
Fig. 1 1 .1 2 .
point on the rotor) are obtained (Fig. 1 1 .1 2 ). The m achine is then stopped and
a tria l mass M T is fixe d to the rotor at some a rbitrary p osition. R unning the
m achine at the same speed as before yields a new v ib ra tio n am plitude, \ V i\ ,
and phase angle, a^ , enabling a vector diagram to be constructed dire ctly.
281
Plane 1
Plane 2
V 2.0
V 1.1
V ,0
V 1.0
^2.1 V 2.0
_<l
o
V2.2 - V2.0
Fig. 11.13 .
Mc = ,
, Mr
\V ,~ V j
(11 .8)
22
20
position 1.
m easuring position
m easuring position
m easuring position
1.
2.
2.
283
rY U D A L
'
L I ST
1C DIM
, c<
2 )/D C 2
CC
D IM
D IM
K C 2 /2 )/L C /2 )/
SC2 , 2 ) / TC2 , 2 ) ,
F0P
Y=
READ
LET
) , C2 /
1(2/
) / PC 2 /
/ CC /
J*( 2,
1 T0
A (Y )jL (Y )
Y
C(
C(
C(
1
1
1 ) = AC
I )*C 0SC B C
2 ) =AC
1 )*S IM C B C
I )=
1) )
1) )
BrOel & Kjr
LET
LET
1)
C<
DC
2 ) =CC
1)=AC
C0SCBC
LET
DC
2 ) = AC
F I K C b c
LET
LET
DC
DC
1) = - D C
2 ) = DC
LET
EC
LET
EC
1 ) = AC
2 ) = AC
1)
3) )
3>*C 0SC BC
3 > *S IM C B C
3) )
100
105
LET
LET
EC
2/
1) = -E C
2)
EC
2/
2 ) =EC
1/
1 10
LET
FC
1 ) =AC
4>*C 0C bC
4) )
1 15
LET
FC
1/
2 ) = AC
4 > * S IM C bC
4) )
120
125
LET
LET
FC
FC
2/
2*
2)
1) = - F C 1/
2>=FC
1/
1)
130
LET
GC
1/
135
LET
GC
1/
1 ) = AC
2 ) =A C
140
LET
GC
2
2,
6100
1/
QP
LET
LET
1)
5 ) * C0 S Cb C
5 ) * S I M CbC
=- G C 1/ 2 )
=G C 1/
1)
1 ) = AC 6 > * C 0 S C B C
)
)
5)
5)
1)
2 )
LET
GC
LET
HC
155
LET
HC
1/
160
LET
HC
2.
2 ) = AC 6 ) * S I N C B C
1) = -H C
1/
2)
165
LET
HC
2/
2 ) =H C
200
MAT
1/
\>
145
150
6) )
6) )
1)
I =E- C
MAT
J = F -D
210
MAT
K = G -C
215
MAT
220
225
MAT
MAT
L=F-D
M=H- D
N
230
235
MAT
MAT
0 = D* I
P=C *J
240
245
MAT
MAT
250
255
MAT
MAT
T = Q-R
260
265
MAT
MAT
U =11JV
v=s*u
270
275
MAT
MAT
280
285
MAT
I = C *M
J =D *K
K= I - J
MAT
X=K*U
290
300
LET
LET
Y 1 = S CP CV C
Y2=SGRCXC
3 10
IF
320
LET
330
G0T0
Y3= 0
350
340
LET
Y3=
350
360
IF
XC 2 f
2) <
LET Y4=
0
370
380
G0T0
390
LF.T Y 4 =
ISC
390
LET
Y 5 = Y 3 + C AT fJ Cv C
1/
2)/V C
1#
l)) ) /A T r i(
1) +
LF.T
Y 6 - Y 4 + C A T N C XC
1/
2 )/X C
1,
1 ) ) ) / ATiJ C
1)*
6100
=E-
QP
205
0=K *L
R=M *N
S= 0 - P
VC
1/
CT
>
I X
1/
1/
1)
l) t
2+VC
2+XC
TH EIJ
340
THEM
330
1/
1/
QP
BrOel & K j r
6100
400
/ J C2 /
BC Y>=EC Y)*A TM (
NEXT
LET
r,
j2(2>
18C
P R IN T
P R ItJT
"M O D U LI :
"M 0DULUS
499
5 1C
RUIJ
DATA
17 C>
112/
AIJ2
AMD
5 3/
A P G U ! 1 EM T
APGME TJ T
73,
235/
dF
3F
r i * , Y2 *Y t
END
M0DULUT
A_'J D
A P G U ! I F ;J T
M0DULUS
AMD
A P G U : ! I:
0F
C l:
. 7f . 1 2 7
53C379
t.
17
1. r144
B r sl & Kjao
4 10
420
READY
Fig. 1 1. 14.
(1 1 9 )
W hen a phase meter and an accurate num erical c a lcu la tio n technique are
used, the residual unbalance is often reduced to acceptable levels after th e
firs t ru n . W here d iffic u ltie s are experienced, it may be necessary to verify th e
lin e a rity and phase re produceability betw e e n trial unbalance placed on th e
correction planes and vib ra tio n m easurem ents in the m easuring planes. In
th is case, some experim ents w ith p o sitio nin g of the sensors, or selection of
correction planes, may enable a technique to be evolved to give more sa tisfa c
tory results.
1 1 .7 . D IFFIC U LT B A L A N C IN G TASKS
1 1 .7 .1 . Fine Balancing
To achieve unbalance q u a lity grades as lo w as G1 and GO ,4 for equipm ent
such as record players and gyroscopes, special techniques are required. For
exam ple at GO,4 a roto r ru n n in g at 6 0 0 0 r/m in w ill re q u ire an ecce n tricity
of the ce ntre of mass of less than 0 , 6 3 /m. Considering th a t in stru m e n t ball
bearings may them selves be m anufactured to a radial ru n -o u t tolerance of
1 /jm , w h ils t the typical tolerance on fin e m achining (b o rin g , tu rn in g , g rin d
ing) is 2 5 /um, it is clear th a t the unbalance introduced at all stages of p ro d u c
tion and assembly m ust be controlled to achieve precision. This w ill necessi
tate balancing at the fin a l stage.
S im ple support ro lle rs, or prism atic blocks w hich are used to support th e
rotor jo u rn a ls for general-purpose tasks on industrial b alancing m achines, are
inadequate in these cases. For grades b e tte r than G1 th e rotor should be
m ounted in its own su p p o rt bearings on the balancing m achine; for grade
GO,4 the rotor should be driven as it w o u ld in service, so th a t the actual s e r
vice en viron m en t in te rm s of electrical and aerodynamic e ffe cts may be re p ro
duced. Examples are gyroscopes excited by a half-stator assem bly, and tu rb o
chargers driven by com pressed air.
1 1 .7 .2 . Flexible Rotors
A t high rotational speeds, the rotor can no longer be regarded as rigid. For
ro tational speeds greater th a n 50% of th e firs t critical speed it may be said to
be fle xib le (Fig. 1 1.15). The axial d is trib u tio n of unbalance along the rotor w ill
tend to excite the various mode shapes o f the rotor, depending on its speed
285
_________________________________
Fig.1T. 15.
s h a ft
790521
deflection
and
Fig. 1 1 .1 6 .
1 1 .7 .3 . Crankshaft Balancing
C rankshafts are used to convert the reciprocating m otion o f a piston into ro
tary m otion of a shaft For design purposes the moving e le m e n ts can be d i
vided into purely rotary com ponents (big-end pin, connecting rod big-end) and
purely reciprocating com ponents (piston assem bly, connecting rod sm all-end).
Therefore suitable counterbalance w e ig h ts can be devised and incorporated
in the crankshaft w ebs, w h ic h balance th e rotary com ponents plus a propor
tion (or factor) of the re cip ro ca tin g com ponents (Fig. 1 1 .1 7 ). In th e production
process, such crankshafts can be balanced on balancing m achines, so p h isti
cated autom ated handling and correction equipm ent is w id e ly used in the car
industry fo r th is purpose. In certain c o n fig u ra tio n s such as in V -4 and V -6 d e
signs, the com ponent of unbalance at tw ic e the rotation frequency of the
287
B ig-End
P in
Counterbalance
Pin
Fig. 1 1 .1 7 .
______ 790523
shaft, o w ing to the com plex m otion of th e connecting rods, can be unaccept
able. In these cases, balance can be achieved by provision of a contra-rotating balance shaft driven at tw ic e engine speed. These s h a fts are fabricated
w ith specially calculated counterbalance w ebs.
1 1 .7 .4 . M ultiple-Span S hafts
The m ajority of c u rre n t te ch n iq u e s have evolved for use w ith single-span ro
tors, supported in bearings at each end. The problems o f ro to r assem blies
supported in three or m ore bearings is th e subject of c u rre n t research
(F ig.1 1 .1 8 ).
288
For a rigid rotor, m easured values w ill depend not only on rotor unbalance,
but also on th e accuracy of the alig n m e n t of the bearings and on the ru n -o u t
at the jo u rn a ls . In contrast to unbalance fo rce s, the latter effects are p re d o m i
nantly speed-independent, causing a co n s ta n t centre-of-m ass eccentricity. Re
building of th e m achine (e.g. for lin e -b o rin g of the bearing housings and ro to r
balancing) w ould be necessary to correct su ch a fa u lt. For fle xib le rotors, it is
extrem ely d iffic u lt if not im possible to p re d ict how the bearing systems w ill a f
fect the mode shapes and vibration response at the critic a l speeds. It is c o m
mon to consider the roto r spans in tu rn , as separate sim ply-supported sys
tem s. The sh a ft-stiffn e ss is often such th a t interm ediate bearing assem blies
are assum ed to decouple separate rotor spans. A lte rn a tiv e ly , where lin e a rity
and phase fid e lity is observed betw een unbalance added on any co rrection
plane and the vibration m easurem ents on every support bearing, then th e in
fluence coe fficie n t m ethod may be used. In general, one extra co rrection
plane is required for every extra bearing m easurem ent. Portable m easuring
e quipm ent may be used fo r th is w ork, w ith sw itches to a llo w connection of
extra vib ra tio n sensors: b u t, as the n u m b e r of bearings increases, the s o lu
tion of th e re sulting m a trix equations becom es increasingly complex and ca lls
for su b sta n tia l com puting fa cilitie s.
1 1 .8 . S ELECTED B IB LIO G R A PH Y
ADKINS, F.E. and
GRAY, A .:
W orks".
Parsons,
BRUNNENGRBER, H.
and DRUST, P.:
BISHOP, R E D ,
PARKINSON, A .G :
BISHOP, R E D. and
GLADW ELL, G .M .L.:
EL-HADI, I.:
M essen
289
FEDERN, K.:
FEDERN, K.:
GRGIC, A.:
GUINS, S B. and
BURMIST, J.:
JUDGE, A .W .:
KELLENBERGER, W.
KOLBE, W .:
LAWRIE, G.C.:
LUND, J .W . and
T0NNESEN, J.:
MACDUFF, J.N .:
PETERMANN, J.E .:
290
Tool
Engineer,
Proc.
RIEGER, N.F.:
SCHNEIDER, H.:
THEARLE, E L :
TESSERZIK, J .M .,
BADGLEY, R.H. and
ANDERSON, W .J.:
THEARLE, E.L.:
T0NNESEN, J.:
W ILCOX, J.B .:
291
1 2 .1 . IS O LA TIO N OF V IB R A T IO N A N D S H O C K
Undesired vibration and shock may o rig in a te from a w id e variety of
sources, such as unbalance and reciprocating motion in m echanical m a ch in
ery, aerodynam ic tu rb u le n c e , rough sea m ovem ents, earthquakes, road and
rail tra n sp o rta tio n , rough h a n dling of e quipm ent, etc.
Even th ough ideally all undesirable v ib ra tio n s should be e lim inated at the
source it is obvious from th e above " lis t" o f sources th a t th is may be possible
only in very fe w cases. In o th e r cases, how ever, it may be possible to " is o
la te " the source by m eans o f shock and vib ra tio n isolators, or to reduce the
shock and vibration effects by means of effe ctive ly designed vibration absorb
ers, or the use of dam ping treatm ents.
On the other hand, " n a tu r a l" vibration sources like aerodynam ic tu rb u
lence, rough sea m ovem ents and earthquakes cannot be "is o la te d " in the us
ual sense of the w ord. The o n ly way to d im in is h undesirable vibration effects
o rig inating from these types of sources is to "is o la te " th e equipm ent to
w h ich the vibrations may cause serious dam age.
Now, w h e th e r it is the source or the equ ipm e n t that is going to be isolated,
the physical principles involved are sim ilar.
1 2 .1 .1 . Vibration Isolation
Fig. 12.1 show s the " u n iv e rs a l" so lution to isolation problem s, i.e. the
proper m ounting of the source (machine), Fig. 12.1 a), or the equipm ent,
Fig. 1 2.1 b), on springs and dam pers. (If th e springs consist o f cork or ru bber
like m aterials dam ping is a u to m a tica lly b u ilt-in to the spring in the form of in
ternal m aterial damping).
Consider firs t the vib ra tio n iso la tio n o f the source, Fig. 1 2.1 a).
The equation of m otion fo r th e mass, m, in the system, Fig. 1 2.1 a), was fo r292
Equipm ent m
M achine m
-[*)
- J * (
271506
ClX
F f = k x + c -d T
b)
a)
Fig. 1 2 .1 . Illu stra tio n o f th e basic p rin c ip le s involved in vib ra tio n isolation, i.e.
The m o u n tin g o f the m achine p ro d u c in g the vib ra tio n , or the e q u ip
m e n t to be iso la te d from the vib ra tio n , m, on sp rin g s and dam pers
m ulated and solved in Chapter 3, se ctio n
force, F0 e i2nft\
x { f ) = H {f)F 0 e i2*ft
1
,
4 n 2f02m
w h e re H [f) = ------------
In th e case of vibration isolation, one is not norm ally interested in x(f) but
in the force tra n sm itte d to the fo u n d a tio n . This force is the vector sum of
the force transm itte d thro u g h the sp rin g elem ent and that tra n s m itte d
th rough th e damper, i.e.:
F (f) = kx +
293
where T is the force tra n sm issib ility. M a n ip u la tio n of this equation results in
in practice.
I I 111 1 1
1
w
\ \
%
Hevea
(N atural ru bber)-------- V v \
fh
C5R-S----------
GR-
i\
\ :l
\ A
0,001
0,1
....................
1,0
R a tio -
l i
...........1
10
Forcing fre qu e ncy
\ i
100
f/t
27 15 08
296
m
~B
(12.3)
M achine m
1
Fo un d atio n, B
Fig. 12.5. Illu s tra tio n o f the approxim ation o f th e m achine fo u n d a tio n by a
mass, B
A p ra ctica l approach to vib ra tio n problem s of the type sketched in
Fig 12.1 a), is to frequency analyze the vibrations produced by th e m achine.
From the m easured (or estim ated) spectrum the lo w e s t frequency com ponent
to be "iso la te d can be determ ined.
By then u tilizin g the curves, F ig .1 2 .2 , the re so n a n t fre q u e n cy of the
m ounted system , as w e ll as the dam ping, necessary to provide s u ffic ie n t iso
lation, can be estim ated.
To find the stiffn e ss required fro m the isolation m o u n t (spring) w h e n the de
sired resonant fre q u e n cy has been determ ined th e fo rm u la :
k
3 9 ,2 P f02 N /m
(12.4)
791021
k = 3 9 , 2 - 2 -1 0 2 = 7840 N /m
From the m a n u fa ctu re r's catalogue it w as found, how ever, th a t he did not
supply a vibration isolator w ith e xactly th is stiffness, and use th e re fo re had to
be made of isolators w ith a stiffness o f 1 1 8 0 0 N /m .
298
P (m ax)
d [m ax)
W here P is th e w eight of th e m achine per isolator and d is the sta tic deflec
tion produced by this load. In th e above exam ple th e m axim um w e ig h t per iso
lator was given by the m a n u fa ctu re r to be 3 ,6 kgf, and the corresponding def
lection 3 mm th u s
k = 3,6 kgf = 3,6 ~9,S- N/m = 11800 N/m
3 mm
0,003
It is now necessary to check h ow th is in flu e n ce s the resonant frequency of
th e isolation system . R earranging Eqn. (1 2.4) gives:
. /
^
. / 1 1800
fn = V --------- = V ---------- = 12,3 Hz
0
V 39 2 P
y 39,2 2
BrUal
O O
dB
D O
D O
O D D O
D D O
O D O
Lm l: ------------------------------
M a c h in e V ib r a t io n
50000
0000
1 -------------- 1
^
M achine
I
| Center o f g ravity
^
<
V ib ra tio n
isolators
1
|
< |
>
/ / / / , F o u n d atio n V / / /
V / / / / / / / / / / / / / / Z / / / / / / / , ^// / / / / / / /
271510
300
Fig. 1 2 .1 0 .
Fig. 1 2 .1 1 .
S ketch sh o w in g h ow the a d d itio n o f mass is utilized in th e socalled " flo a tin g " floor
301
ances th a n does the "s im p le " system d iscussed above, see Fig. 1 2 .1 3 . The
design of such compound system s is, on th e other hand, m ore com plicated
and critica l th a n the design o f a "s im p le " v ib ra tio n isolator.
For readers w ho are fa m ilia r w ith e le c tric a l filte r theory and ele ctro -m e chanical analogies the design problem s in vo lve d may, how ever, not seem to o
form idable.
0,1
0,2
0,5
10
20
50 100
Freq uency ratio f/f-|
2 7 15 14
Fig. 1 2 .1 3 .
Transm issibility curves illu s tra tin g th e difference in tra n s m is s ib ility betw een the s im p le and the com pound system
R eturning n o w to the second "ca se o f vib ra tio n isolation, i.e. the case
w he re e q u ipm e n t is to be isolated from a vib ra tin g fou n d a tio n , Fig 12 1 b),
th e equation o f m otion for th e mass, m, is:
(1 2 .5 )
A gain solving the equation fo r an
xo = X 0 e i2n,t results in
302
jf_
Qfo
1
4n*f*
2nfc\
-,
+
m + 1 m
1 If
1 + \ To
2
1 If
= \T\
12.6)
+ o A fa
Fig. 1 2 .1 4 .
304
s w /fJ V
m *.
The m axim um force a ctin g on the fo u n d a tio n Fr, can n o w be found from
Fig 12 18 b) and Fig.1 2 .1 5 , provided th a t th e resonant fre q u e n cy, f0 , of the
system F ig . 1 2 .1 7 is know n. A ssum ing th a t th is is f0 =1 / 1 0 T
th e m axim um
force acting on the fo u n d a tio n is
found
fro m Fig.
1 2.1 5 to be approxim ate
0 ,6 tim es th e "e ffe ctive fo rce , Fig. 1 2 .1 8 b), i.e.:
M a xim um force "re s p o n s e " = 0 ,6 F0 = Fr
The m axim um displacem ent of the mass, m , is equal to th e force divided by
the stiffness, k, of the isolator:
*'
= Fl _ 0 ,6 Fp
k
k
305
271516
Fig. 1 2 .1 5 .
___________________________________________________________________________________________________________________________________ 2 7 1 5 1 7
Fig. 1 2 .1 6 .
Illu s tra tio n o f the concept o f "e ffe c tiv e " p u lse height
Fig. 1 2 .1 7 .
Fig. 1 2 .1 8 .
vr
and
= dxdt
ar =
d 2x
d t2
= 2 n f0 x r = 2 n f0 ~
= 1, 2 n f F
= (2 n f0)2x r = 4 n 2 f 2 = 2 ,4 k 2 f 2 ^
Fig. 1 2 .1 9 .
w h e re
*)
c
cc
is no longer s im p ly equal to
c
2 \/k m
tt
but
1
2Q
In e ffe ct th is d iffe re n c e in re sonant fre q u e n c y betw een d am p e d and undam ped re s o n a n c e s also
a pp lies to vib ra tio n is o la tio n .s y s te m s . H o w e v e r, th e dam p in g in c lu d e d in these s y s te m s in p rac
tice is often so s m a ll ( f 1) th a t th e re so n a n ce s h ift is n o rm a lly neglected.
308
Taking these factors into account the m axim um displacem ent, xr , o f the
mass, m , becomes:
x = -----
(1 2 7)
r k\/1 + {2^y
and
a' _ v
1-Z2
/ i W
F,
k
Fr
i
(1oo,
W hen the shock pulse d u ra tio n is no longer short com pared w ith one half
period of the isolation system m otion, u tiliza tio n of th e shock spectrum te c h
nique becomes som ew hat m ore com plicated.
It seems, how ever, that u tiliza tio n of th e "s p e c tru m ' show n in F ig .1 2 .1 9
and the method of e stim a tio n described above for dam ped systems m ay re
s u lt in reasonably good approxim ations to actual practical problems.
In conjunction w ith the practical a p p lication of shock isolators ce rta in facts
should be noted:
F irstly, as m entioned above, reduction in shock fo rc e tra n s m is s ib ility can
only take place by a llo w in g th e isolator to deflect, i.e. by m otion w h ic h a llo w s
the shock energy to be dissipated over a m uch longer period of tim e th a n that
occupied by the shock itself. Thus ce rta in space clearances must be allo w e d
fo r the isolated equipm ent.
Secondly, if the resonant frequency of th e isolation system is chosen in c o r
rectly the isolator may "a m p lify " the d estructive e ffe c ts of the shock rather
than provide the desired isolation. This requires th a t th e resonant fre q u e n cy
of th e isolation system is aw ay from all resonances w ith in the m a ch in e or
equipm ent to be isolated.
Thirdly, if the isolator tu rn s out to have unexpected non-linear c h a ra c te ris
tics (and many practical isolator m a te ria ls do p e rfo rm non-linearly) a great
num ber of "e x tra " response effects m ay take place a t harm onic or s u b -h a r
m onic frequencies.
In some cases isolators are, on purpose, designed to be non -lin e a r. If, for
instance, space lim ita tio n s do not a llo w fo r the re q u ire d (linear) m o tio n of
equipm ent one may be tem ped to em ploy non -lin e a r isolators of the "h a rd e n
30 9
in g spring type (see also C hapter 3, section 3 .2 ). This kind o f isolator w ill,
w h en deflecting into the n o n -lin e a r region, firs tly change the resonant fre
quency of the isolation system (f0 increases w ith increasing is o la to r deflec
tion), and secondly produce a num ber of h a rm o n ic force com ponents w hich
may excite in te rn a l resonances in th e isolated equipm ent. A lso th e peak ac
celeration of the equipm ent m ay be considerably increased by th e use of
"h a rd e n in g " spring type isolators.
On the other hand, i f the is o la tio n system c o n ta in s a fa ir a m o u n t o f dam p
ing, the d e terioratin g effects m entioned above are drastically reduced. Thus,
a heavily damped, "h a rd e n in g " sp rin g type is o la tio n system m a y in some
cases provide the appropriate s o lu tio n to a d iffic u lt iso la tio n p roblem .
BASIC TYPES OF ISOLATORS
Material
F re q u e n c y Range
O p tim u m Frequencies
Dam pin g
L im ita tio n s
R em ark s
M etal Springs:
H elical Com pression
Sprin gs
V e ry lo w
0.1% o f
critic a l
R e a d ily transm it
high frequencies
V e ry low
L it t le used
Lea f Springs
Low
Low
F a irly good
(due to fr ic tio n )
L im ite d t o s pecific
a p p lic a tio n s
B elleville Washers
Rubber:
(i) In Shear
(ii) In Com pression
(iii) Shear-Com pression
Hig h w ith
parallel
stackin g
S u b je c t to fatigue:
m o re com plicated
assem bly
C o m p a c t. S tiffn e s s
depends o n m eth o d o f
s tack in g . C o n tr o lle d
n o n -lin e a r stiffness
I lim ite d
> lo ad-carrying
| capacity
depends on
com position
and hardness
High
C o rk
H igh
L o w (6% o f
critical)
H ig h ly com p ressib le
w ith o u t lateral
ex pan sion
F elt
Depends on density
& thickn ess. Exten ds
in t o au dio -frequ en cy
range
H ig h (u su a lly above
4 0 Hz
High
Low
Fair
L o w stiffness with
h igh com p ressib ility
Used in th e fo rm of
m o u ld ed pads or
c u t slabs
Low
Fa ir t o high
L im ite d load-carrying
c a p a c ity
Used in fo r m o f pads;
also as in serts
Sponge Rubber
Steel Mesh
I
|
Increases
w ith rub ber
hardness
Low
R e la tiv e ly u ndeveloped
H ig h
Depends on
design
M o u ld e d ru b b e r mounting
pads w ith m e ta llic casings
a n d/o r in serts
W id e range depends
o n design
D ep end s o n desig n
Low
Rubberised F a b ric
10
6 8% ty p ic a l
Cork -R ub ber
High
M o d e ra te ly lo w
Steel-Bound C o rk
D ep end s on density
Hig h
12 H z typical
Low
depends
on rubber
hardness
P ro p e rtie s s im ila r to
s o lid r u b b e r b u t w ith
increased s ta tic
de fle c tio n s
U p to 6% o f
critica l
..
Table 12 1
310
C o r k c o m p o s itio n w ith
bo nde d m e ta l faces.
P a r tic u la r ly ap p licab le
fo r is o la tio n o f concrete
m o u n tin g b lo c k s
1 2 .2 . D Y N A M IC V IB R A T IO N C O N TR O L A N D V IB R A T IO N D A M P IN G
In th e previous section th e basic aspects involved in th e isolation of v ib ra
tions and shocks w ere o u tlin e d . There a re , how ever, practical cases w here v i
bration isolation is not suitable, or d iffic u lt to arrange, and other m ethods of
vibration reduction m ust be sought. O ne w a y of reducing th e vibration m ay
then be to utilize the p rin cip le of the d yn a m ic vibration absorber. This p rin c i
ple can, in general, only be used e ffe c tiv e ly w hen th e "o rig in a l vib ra tio n s
contain one major freq u e n cy com ponent o nly (or they co n sist of a very n a r
ro w band of frequencies such as a lig h tly dam ped, random ly excited single re
sonance).
If vib ra tio n reduction is to be achieved in cases of ra n d o m ly excited m u ltidegree-of-freedom system s (plates and beam s) the a p p lication of dynam ic v i
bration absorbers is n o rm a lly com plicated and use is th e n preferably made of
some sort of general dam ping treatm ent.
A b so rb e r system
m"
k
M achine
| P0 sin(27r ft)
x 2 = X 2 sin(27r ft)
x 1 = X t sin(27r ft)
Isolator
TV ^77
////////
/ / r oFoun
u n adation
a iio n
271521
Fig. 1 2 .2 0 .
~^2
12. 10)
A ssum ing th a t the sta tio n a ry solutions to these equations can be w ritte n
(w here X j and X 2 can be e ith e r positive or negative)
* i = X , sin (2 n f t )
and
x 2 - X 2 sin ( 2 n f t )
then
and
w h ere
By setting
312
(12 1 1 )
resonant frequency of th e attached (absorber) system
the m otion, X J , of the m a ch in e w ill be zero, i.e. the m a c h in e w ill not vibrate
at all. The m axim um a m p litu d e of the mass, m , is in this case:
269109
Fig. 1 2 .2 1 .
3 13
Frequency ratio (r )
To
269082
Fig. 1 2 .2 2 .
27152 2
Fig. 1 2 .2 3 .
314
high, tra n sm issib ility "p e a k s " are found. The shape of the curve is caused by
th e dynam ic coupling betw een th e m a c h in e /is o la to r system and th e absorber
system . C oupling effects of th is sort are quite com m on in m any branches of
physics.
If the absorber damping is in fin ite , the absorber mass is v irtu a lly clamped
to the m achine and the absorber system does not fu n ctio n at a ll. Fig. 1 2 .2 2 .
In practice, w h e n a damped vib ra tio n absorber is applied to a m a c h in e /is o la
to r system th e tra n s m is s ib ility curve m ust lie betw een th e tw o extrem es
sketched in F ig .1 2 .2 2 . This is illu stra te d in F ig .1 2 .2 3 for va rio u s values of
absorber dam ping ratio.
Theory has show n th a t w h e n dam ping is added to the absorber th e " o p ti
m u m " perform ance co n d itio n s*) are, in general, no longer obtained by tuning
the resonant frequency o f th e absorber system to equal th e re sonant fre
quency of th e m a c h in e /is o la to r system . A c tu a lly the most fa vo u ra b le tuning
depends upon the ratio b e tw e e n the absorber mass and the m ass o f the ma
chine i.e. m /M . It has been fo u n d th a t w h e n th e dam ping is of th e viscous
type then th e ratio betw een th e absorber resonant frequency, fa , and the ma
c h in e /is o la to r resonant fre q u e n cy f0 , should be:
269076
Fig. 1 2 .2 4 .
*)
" O p tim u m " co n d itio n s are a s s u m e d to be th o se w h ic h e n su re a m a x im a lly " f l a t pea k-n o tch region of th e tra n s m is s ib ility c u rv e . Fig. 1 2 .2 3 , to be o b ta in e d .
315
Fig. 1 2 .2 5 .
Fig. 1 2 .2 6 .
316
m ass-ratio th e "d e -tu n in g " o f th e absorber m ay become very sig n ifica n t. A lso
the "o p tim u m viscous d a m p in g factor depends upon th e m ass-ratio, see
F ig .1 2 .2 4 . Finally, F ig .1 2 .2 5 shows some theoretical tra n s m is s ib ility curves
calculated fo r various m a ss-ra tio s and "o p tim u m dam ping. Note the dec
rease in resonant a m p lific a tio n w ith increasing m ass-ratios.
As pointed out in section 12.1 the th e o re tica l tre a tm e n t of the vibration
tra n s m is s ib ility from a v ib ra tin g source (m achine) to its fo u n d a tio n , and th a t
Vibration of
specimen mass
afija:__
v 25 mm/sec
10000
V ib r a t io n o f
specim en mass
w ith absorber
attached_______
!5 mm/sec
1 mm/sec
QP1123
Fig. 1 2 .2 7 .
10000
269121
317
of the vib ra tio n tra n s m is s ib ility from a vib ra tin g fo u n d a tio n to a m ounted
equipm ent is more or less identical. This, of course, also applies w ith respect
to the use of dynam ic v ib ra tio n absorbers see, F ig. 1 2 .2 6 .
To illu s tra te this s ta te m e n t consider a rig id equipm ent (m ass: M) e lastically
m ounted on an e le ctro-dynam ic vibration m a ch in e (see also section 10.1). The
tra n s m is s ib ility curve fo r th is system w a s measured and a utom atically re
corded on a Bruel & K j r Level R ecorder, Fig. 1 2 .2 7 a). By attaching a
damped dynam ic absorber system (mass: m = 0 ,5 M ) to th e m ass M the tra n s
m issib ility curve was changed into the one show n in Fig. 1 2 .2 7 b). The e ffect
of the absorber is cle a rly noted. For th e sake of com pletion th e vibration o f
the absorber mass, m , w a s also m easured and recorded, Fig. 1 2 .2 8 . Before
fin ish in g th is brief discussion of the d yn a m ic vibration a bsorber it should be
m entioned th a t the p rin c ip le of the absorber may be used not only to reduce
resonance effects in v ib ra tio n and shock is o la tio n system s, b u t also to reduce
the vib ra tio n of beams and plates v ib ra tin g in one of th e ir fundam ental
modes. Thereby the acoustic radiation fro m , fo r instance, a plate, may be re
duced, m aking the dynam ic vibration absorber an e fficie n t to o l in the "b a ttle
against a coustic noise.
Fig. 1 2 .2 8 .
*)
A c tu a lly because in te rn a l m a te ria l d am ping o fte n increases w ith fre q u e n c y a certain " d a m p in g "
e ffe c t m ay also be a chie ve d by s h iftin g th e re s o n a n t frequencies.
319
Fig. 12.29.
Other types of ''d e a d e n in g '' materials are at present co m m e rcia lly avail
able. Common to all of th e m are that they are made from hig h polym er m ate
rials possessing optim um d a m p in g properties over certain fre q u e n c y and te m
perature regions. These re g io n s may, how ever, fo r some m astic m aterials, be
fa irly wide.
To obtain o ptim u m dam ping of the co m bination stru ctu ra l e le m e n t + dam p
ing m aterial, not only m ust th e internal loss fa cto r of the d a m ping material
be high, but so also must its m odulus of e la s tic ity (Young's m odulus).
An approxim ate form ula governing the dam ping properties of a treated
panel in practice is given by th e expression:
2
320
One fact, w h ich is im m ediately obvious from the above form ula, is th a t the
relative thickness of th e layer of dam ping m aterial, (d 2 / d j ) , plays a ve ry im
portant role in the re su lta n t dam ping. In practice th e ra tio is norm ally chosen
to be of the order of th re e to one. A ls o , it can be seen th a t it is g e n e ra lly ad
vantageous to apply one (thick) layer of damping m a te ria l rather than dividing
the layer in tw o by using doublesided coating.
A th ird method of applying dam ping to stru ctu ra l elem ents is th e use of
sandw ich stru ctu re s, F ig .1 2 .3 1 . S everal types of su ch con stru ctio n s exist:
The original stru ctu re may be supplied w ith a co n stra in e d viscoelastic layer,
i.e. the dam ping m ateria l is covered w ith a thin m etal sheet, F ig .1 2 .3 1 a): a
th in visco-elastic layer is placed b e tw een tw o e q u a lly th ick plates (adhesive
separator), Fig 12.3 1 b); or fin a lly use may be m ade of a thick visco-elastic
layer betw een the tw o plates, F ig. 1 2 .3 1 c).
A considerable nu m b e r of th e o re tica l and exp e rim e n ta l investigations have
been carried out to a llo w the p re d ictio n and com parison of dam ping p ro p e rt
ies of sandw ich stru ctu re s . The ge n e ra l results of th e se investigations in d i
cate th a t, con trary to the above discussed a p p lication of mastic deadeners,
the thickness of th e visco-elastic layer is not a fa c to r of prim e im portance. It
seems, how ever, th a t th e overall geom etry of th e c o n stru ctio n (sym m etrical,
unsym m etrical) is im p o rta n t, the sym m etrical c o n s tru c tio n show ing th e most
favourable overall dam ping p roperties. On the o th e r hand, w hen th e th ic k
ness of the visco-elastic layer is increased, the te m p e ra tu re and frequency
ranges w ith in w h ic h optim um dam ping can be o b tained also increases.
E 2 ;r)2
V iscoe la stic
d2
layer
:
Panel
Ej
Fig. 1 2 .3 0 .
2 7 15 25
T h in m etal sheet
Viscoelastic layer
Panel -
Panel
V is c o e la s tic layer
a)
Fig. 1 2 .3 1 .
Panel -
V iscoelastic layer
b)
c)
2.
The practical application of th e frequency response method n o rm a lly consits of cutting a bar-shaped sam ple from the m a te ria l to be te ste d , clamping
50
100
200
4 00
8 00
1600
Hz
3200
6400
Frequency, f
27 15 27
Fig. 12.32.
322
Fig. 1 2 .3 3 .
m e a su re
the sample bar at one end, or both ends, and e x c itin g it into b e n ding vibra
tio n s w ith a variable frequency, sinusoidal force.
The am plitude o f the response vib ra tio n s is then p lotted as a fu n c tio n of fre
quency, see Fig. 1 2 .3 3 . From such a curve, at a resonance peak, th e loss fac
tor can be calculated as
r, = A fn t f n
(1 2 .1 4 )
^ j 2 N /m 2
(1 2 .1 5 )
/
h
p
Kn
*)
3 23
n>3
one end free, one end clamped:
K, = 1,875; K 2 = 4 ,6 9 4 ; K 3 = 7 ,8 5 5
n>3
T his measuring m ethod is
0 ,0 0 1 . W hen the loss factor
plitu d e because no standing
resonance peaks w ill be too
w ith reasonable accuracy.
For the second m ethod m entioned above, i.e. th e decay rate m ethod, the
external exciting fo rce is tuned to a resonant fre q u e n c y of the sam ple w hich
w ill start a forced oscilla tio n w ith steady a m p litu d e when e q u ilib riu m is
reached. If the e xcitin g force is stopped instantly, th e vibration a m p litu d e de
cays exponentially w ith tim e. (Thus lin e a rly w ith tim e if plotted o u t lo g a rith m i
cally.) The loss fa c to r rj is found fro m
1 =
(1 2 .1 6 )
Fig. 1 2 .3 4 .
Fig. 1 2 .3 5 .
So far the decay-rate type of te sts discussed have been based on the inter
ruption of a steady state norm al mode (resonant) vib ra tio n , i.e. tim e decay
m easurem ents. O ther types of decay-rate m easurem ents are sometim es
used: the de te rm in atio n of the decay-rate of progressive w aves. T h is is based
on decay-rate d e term in atio n s in space rather th a n in tim e, as described be
low. W hen a long strip of m aterial is excited in transverse vib ra tio n at one
end, and te rm in a te d at the o th e r end in such a w a y th a t pra ctica lly no reflec
tio n takes place. F ig .1 2 .3 5 , one-dim ensional m echanical w aves progress
325
7 = 2 7 ,3
(1 2 .1 7 )
1 2 .3 . SELECTED B IB LIO G R A P H Y
BLAY, D.,
BOURGAIN, L., and
S A M S O N , G.:
CRANDALL, S .H.:
CREDE, C E.
CREMER, L. and
HECKL, M .:
GROOTENHUIS, P.:
GROOTENHUIS, P.:
GROOTENHUIS, P.
GROOTENHUIS, P.:
KUNICA, S.:
KURTZE, G.:
326
B erlin, H eidelberg /
MEAD, D.J.:
MUSTIN, G S.:
OBERST, H.:
O L E S E N ,H P . and
DELPY, D.T.:
ROSS, D ..U N G A R , E.
and KERWIN, E.M .:
RUZICKA, J E.:
SNOWDON, J.C .:
SNOWDON, J.C .:
SNOWDON, J.C .:
APPENDICES
A P P E N D IX A
,.
Hm
P(*) = j...
x >o
P(x)P(x+Ax)
Ax------- L
(A. 1 )
P (x ) P ( x + A x) = P ( x ; x + Ax) = J
p (x )d x
fx+
P (x ; x + A x) =
X
x 2 p (x )d x
(A .2)
th is m ay be expressed as follow s.
f*
(T2 =
Hm
x 2 p (x) dx
J oc
Ax*0
x
Yj X2 P {x , x + Ax)
oo
( A . 3)
328
( j 2 is, in the literature on sta tistics, com m only term ed variance and the
square-root of the variance is called the sta n d a rd deviation, a . However,
w h en the statistical phenom enon being studied is a sta tio n a ry tim e -fu n c tio n
as show n in Fig.A. 1, then
329
(A. 4)
w h ic h is nothing but th e expression used in engineering dynam ics fo r th e s ig
nal R M S -value. Thus, besides being related to the p o w e r involved in th e p ro
cess, th e RM S-value is also directly re la te d to the process sta tistics. This may
be best appreciated by considering th e fact th a t m ost probability d ensity
curves are expressed in te rm s of standard deviations (R M S -deviations), see
also F ig .2 .9 .
F inally, F ig .A .2 illu s tra te s the re la tio n sh ip between th e instantaneous a m
plitu d e values in a G aussian random vib ra tio n signal and th e Gaussian p roba
b ility density curve.
A P P E N D IX B
(B. 1)
H (f) =
330
(B.2)
(B.3)
Q 2Vo
d>(f) = tan 1
(B.4)
Q \j
v(t)
= dxd t
a(t) =
d 2x
W
331
V elocity of mass, m,
Fig. B. 1a)
A cceleration of mass, m ,
Fig. B. 1a)
/A
'Vv/
Hrl
K r : Ur)
*
- 4 n 2f 2
' 'a y >
k
'- ( 0 ] *b(f)
8 0 0 12 9
332
V e lo c ity
\H ( f) \ - %
Acceleration
= 2 n fD 2
|WW|
- J p D ,
V e locity
vx (t)
Acceleration
\H (f)\ =
lH W l * 2 M ,
\H {f)\ - 4 n l 2D'
\m n \ = ^
! ( /)! -
800130
w here
and
S im ila rly respijn se versus excitation fu n c tio n s fo r th e re la tiv e m otion betw een th e mass and the foundation are ta b u la te d in the th ird table:
Response
Q uantity,
Fig.B. 1b)
Relative
Displacement
x ( t) - x 0 (t)
Relative
V elocity
V e lo c ity ,
Acceleration,
x 0 (t)
vxo ^
ax o M
\H {f)\ - f L
' 0
\H (f)\ =
vx M - Vx o (t)
Relative
A cceleration
aJ t ) ~ ax o (^
\H {f)\ =
lH W l '
\H {f)\ -
2 ,/ k
fL
J0 U
|WOTl - J t i o ,
'
\H ( f) \ '0
2 , 'H D 2
[
J0 U
800131
333
A P P E N D IX C
271546
---------------------*)
D<
334
By d iffe re n tia tio n of the d isplacem ent signal w ith respect to tim e one finds
th a t the corresponding velocity of th e mass w ill have a w ave-shape as in d i
cated in Fig.C.2 b). Finally, a second d iffe re n tia tio n yields the acceleration
w ave-shape show n in Fig.C.2 c).
Bm.1 8, C,OK
Brtil 4 K|<w
Bftid&Klw
Fig C. 5. Frequency response curves for the motion of the second mass in the
system sketched in Fig.C.4. Curves are shown for different levels of
excitation of the foundation
336
A P P E N D IX D
Fig.D. 1. Example o f a shock excitation w aveform illu s tra tin g the p rin c ip le o f
superposition in the tim e d o m a in
337
h ( t x) a ( t ) d x
(D.1)
cos{2nf(t x)}]a(x)dx
. X
ff
a(x)dx\
X
I7
cos{2nf(t x))a(x)dx
J -
rt
338
a (r)
co s \ 2 n f ( t z ) \d z +const
(D.2)
This is also
th e expression fo r th e Fourier sp e ctru m of the acceleration
shock pulse except fo r the phase (see Chapter 2, s e ctio n 2.3). Thus
M ax
vR = \Fa (f)\
(D.3)
For each frequ ency com ponent in th e response "s p e c tru m " the re lationship
aR =2 n fv Ris valid, w hereby
Sa (f)
= 2 n f\F a(f)\
(D.4)
A P P E N D IX E
v = I a d t = I A 0e i2nft d t =
J
jZnt
ei2nft = V0 e<2nfl
Thus:
K)
(E.1)
j2 n f
J2nf
2nf)2
e i2nft = X 0e i2n,t
Thus:
Y
K>
j2 n f
^0
/c
O)
(.2nf)2
1 then
RC j2 n f
(E 3)
-o
a
oi I
7
7 + j2 n fR C
(E.4)
ec
ea
1
1 + j2 n fR C
a (t) = A 0 sin ( 2 7i f t ) ,
= 0,
0 < t < T ,
f = f
o th e rw is e
v ( t) = | a ( f ) c f r =
A 0 sin ( 2 n fx )d x
~ - ^ c o s (2 n fx )
'0
= 1 7 f [ - cos ( 2 H
fo r 0 < t <
(E.6)
342
a) Acceleration:
a(t) = A 0 sin 27rft; f = *~r
a(t)
;
fo r 0 < t < T
a(t) = 0
t< 0 ,
t > T
V/
b) V elocity:
i k 2A
v(t)
v (t) =
27rf
(1
- cos 27rft)
A
T
.
t
c) Displacement:
(t)
' . A 0T
2nf
---------------
x(t) =
X ,/
2 ttt
(t -
si n 27rft)
2irf
t
800465
2nf
x -r^.sin (2 n fx )
2 n f t - ~2^fsm ( ^ n f t)
( 7 )
800440
A P P E N D IX F
(F. 1)
w he re W0 is a reference pow er. H ow ever, as the p o w e r measured across a
certain impedance is related to the square of the force a ctin g upon th is im p e d
ance, Z, a more com m on ly used d e fin itio n is
345
dB = 7 0 /o ^ 10( ^ | ) =
2 0 lo g ,
where F and Fa are the RMS (root mean square) values of th e forces.
A ctually, as long as the m easurem ents are re la te d to one a n d the same im
pedance the d e cib el-n o ta tio n in the form 2 0 lo g w ( X /X 0 ) m a y be used as a
convenient re la tive m a g n itu d e scale for a variety o f q u a n titie s, not only fo r
ces. X may, fo r instance, be an RMS displacem ent, a velocity o r an accelera
tio n . X0 , how ever, m ust a lw a ys be a reference q uantity of th e same type as
X. That is w h e n X represents a displacem ent th e n X 0 re p re se nts a reference
displacem ent, and w hen X represents an acceleration th e n X0 represents a
reference acceleration.
A useful application of th e decibel scale is evident in th e fre q u e n cy analy
sis of m echanical vib ra tio n s w h e re large changes in a m p litu d e s occur at re
sonance. The use of decibels (dB) compresses a range of disp la ce m e n t or ac
celeration m agnitudes of 1 : 1 0 0 0 to 0 - 6 0 dB, at the sam e tim e attaining
constant re la tive accuracy in th e graphical presentation. A fu rth e r conse
quence th a t arises from th e decibel scale is th a t m u ltip lica tive fa c to rs become
additive te rm s in th e ir lo g a rith m ic equivalents.
Also, w h e n th e ratio X / X 0 is sm aller th a n 1 ,0 the lo g a rith m to the base
ten of the ra tio , and th u s th e decibel value, becomes negative. In the table
given below only ratios X / X 0 larger than u n ity have been considered. The
same table can, how ever, also be used in cases w here X /X 0 is sm aller than
u n ity rem em bering that
2 0 /w , 0( ^ ) - 2 0 to s , 0( ^ ) - - 2 0 / w , 0 )
i.e. by fin d in g the decibel va lu e for the in ve rse ratio (X0/ X ) and assigning to
it a negative value.
Exam ple
Find the n u m b e r o f dB corresponding to:
346
From the table the corresponding decibel value is found to be 6,0 2 1 dB,
and because of the inverting operation the actual decibel value is 6 ,0 2 1 .
To fin d X / X 0 w hen th e decibel value is g iven, the nearest dB -value is
sought in the table and the corresponding ratio read off the le ft hand and top
colum ns.
E xam ples
a)
b)
= 3 0 2 3 7 dB
= 2,5 1 x 10 3
Table F. 7.
348
(ir )
.00
.01
.02
.03
.04
1.0
1.1
1.2
1.3
1.4
.000
.828
1.584
2.279
2.923
.086
.906
1.656
2.345
2.984
.172
.984
1.727
2.411
3.046
.257
1.062
1.798
2.477
3.107
.341
1.138
1.868
2.542
3.167
1.5
1.6
1.7
1.8
1.9
3.522
4.082
4.609
5.105
5.575
3.580
4.137
4.660
5.154
5.621
3.637
4.190
4.711
5.201
5.666
3.694
4.244
4.761
5.249
5.711
2.0
2.1
2.2
2.3
2.4
6.021
6.444
6.848
7.235
7.604
6.064
6.486
6.888
7.272
7.640
6.107
6.527
6.927
7.310
7.676
2.5
2.6
2.7
2.8
2.9
7.959
8.299
8.627
8.943
9.248
7.993
8.333
8.659
8.974
9.278
3.0
3.1
3.2
3.3
3.4
9.542
9.827
10.103
10.370
10.630
3.5
3.6
3.7
3.8
3.9
.05
.06
.07
.08
.09
.424
1.214
1.938
2.607
3.227
.506
1.289
2.007
2.671
3.287
.588
1.364
2.076
2.734
3.346
.668
1.438
2.144
2.798
3.405
.749
1.511
2.212
2.860
3.464
3.750
4.297
4.811
5.296
5.756
3.807
4.350
4.861
5.343
5.801
3.862
4.402
4.910
5.390
5.845
3.918
4.454
4.959
5.437
5.889
3.973
4.506
5.008
5.483
5.933
4.028
4.558
5.057
5.529
5.977
6.150
6.568
6.966
7.347
7.712
6.193
6.608
7.008
7.384
7.748
6.235
6.649
7.044
7.421
7.783
6.277
6.689
7.082
7.458
7.819
6.319
6.729
7.121
7.495
7.854
6.361
6.769
7.159
7.532
7.889
6.403
6.809
7.197
7.568
7.924
8.028
8.366
8.691
9.005
9.308
8.062
8.399
8.723
9.036
9.337
8.097
8.432
8.755
9.066
9.367
8.131
8.465
8.787
9.097
9.396
8.165
8.498
8.818
9.127
9.426
8.199
8.530
8.850
9.158
9.455
8.232
8.563
8.881
9.188
9.484
8.266
8.595
8.912
9.218
9.513
9.571
9.855
10.130
10.397
10.655
9.600
9.883
10.157
10.423
10.681
9.629
9.911
10.184
10.449
10.706
9.657
9.939
10.211
10.475
10.731
9.686
9.966
10.238
10.501
10.756
9.714
9.994
10.264
10.527
10.782
9.743
10.021
10.291
10.553
10.807
9.771
10.049
10.317
10.578
10.832
9.799
10.076
10.344
10.604
10.857
10.881
11.126
11.364
11.596
11.821
10.906
11.150
11.387
11.618
11.844
10.931
11.174
11.411
11.641
11.866
10.955
11.198
11.434
11.664
11.888
10.980
11.222
11.457
11.687
11.910
11.005
11.246
11.481
11.709
11.932
11.029
11.270
11.504
11.732
11.954
11.053
11.293
11.527
11.754
11.976
11.078
11.317
11.550
11.777
11.998
11.102
11.341
11.573
11.799
12.019
4.0
4.1
4.2
4.3
4.4
12.041
12.256
12.465
12.669
12.869
12.063
12.277
12.486
12.690
12.889
12.085
12.298
12.506
12.710
12.908
12.106
12.319
12.527
12.730
12.928
12.128
12.340
12.547
12.750
12.948
12.149
12.361
12.568
12.770
12.967
12.171
12.382
12.588
12.790
12.987
12.192
12.403
12.609
12.810
13.006
12.213
12.424
12.629
12.829
13.026
12.234
12.444
12.649
12.849
13.045
4.5
4.6
4.7
4.8
4.9
13.064
13.255
13.442
13.625
13.804
13.084
13.274
13.460
13.643
13.822
13.103
13.293
13.479
13 661
13.839
13.122
13.312
13.497
13.679
13.857
13.141
13.330
13.516
13.697
13.875
13.160
13.349
13.534
13.715
13.892
13.179
13.368
13.552
13.733
13.910
13.198
13.386
13.570
13.751
13.927
13.217
13.405
13.589
13.768
13.945
13.236
13.423
13.607
13.786
13.962
5.0
5.1
5.2
5.3
5.4
13.979
14.151
14.320
14.486
14.648 i
13.997
14.168
14.337
14.502
14.664
14.014
14.185
14.553
14.518
14.680
14.031
14.202
14.370
14.535
14.696
14.049
14.219
14.387
14.551
14.712
14.066
14.236
14.403
14.567
14.728
14.083
14.253
14.420
14.583
14.744
14.100
14.270
14.436
14.599
14.760
14.117
14.287
14.453
14.616
14.776
14.134
14.303
14.469
14.632
14.791
5.5
5.6
5.7
5.8
5.9
14.807
14.964
15.117
15.269
15.417
14.823
14.979
15.133
15.284
15.432
14.839
14.995
15.148
15.298
15.446
14.855
15.010
15.163
15.313
15.461
14.870
15.026
15.178
15.328
15.476
14.886
15.041
15.193
15.343
15.490
14.902
15.056
15.208
15.358
15.505
14.917
15.072
15.224
15.373
15.519
14.933
15.087
15.239
15.388
15.534
14.948
15.102
15.254
15.402
15.549
6.0
6.1
6.2
6.3
6.4
15.563
15.707
15.848
15.987
16.124
15.577
15.721
15.862
16.001
16.137
15.592
15.735
15.876
16.014
16.151
15.606
15.749
15.890
16.028
16.164
15.621
15.763
15.904
16.042
16.178
15.635
15.778
15.918
16.055
16.191
15.649
15.792
15.931
16.069
16.205
15.664
15.806
15.945
16.083
16.218
15.678
15.820
15.959
16.096
16.232
15.692
15.834
15.973
16.110
16.245
(continued)
Table F. 1 .
R a tio
.00
.01
.02
.03
.04
.05
.06
.07
.08
.09
6.5
6.6
6.7
6.8
6.9
16.258
16.391
16.521
16.650
16.777
16.272
16.404
16.534
16.663
16.790
16.285
16.417
16.547
16.676
16.802
16.298
16.430
16.560
16.688
16.815
16.312
16.443
16.573
16.701
16.827
16.325
16.456
16.586
16.714
16.840
16.338
16.469
16.599
16.726
16.852
16.351
16.483
16.612
16.739
16.865
16.365
16.496
16.625
16.752
16.877
16.378
16.509
16.637
16.764
16.890
7.0
7.1
7.2
7.3
7.4
16.902
17.025
17.147
17.266
17.385
16.914
17.037
17.159
17.278
17.396
16.927
17.050
17.171
17.290
17.408
16.939
17.062
17.183
17.302
17.420
16.951
17.074
17.195
17.314
17.431
16.964
17.086
17.207
17.326
17.443
16.976
17.098
17.219
17.338
17.455
16.988
17.110
17.231
17.349
17.466
17.001
17.122
17.243
17.361
17.478
17.013
17.135
17.255
17.373
17.490
7.5
7.6
7.7
7.8
7.9
17.501
17.616
17.730
17.842
17.953
17.513
17.628
17.741
17.853
17.964
17.524
17.639
17.752
17.864
17.975
17.536
17.650
17.764
17.875
17.985
17.547
17.662
17.775
17.886
17.996
17.559
17.673
17.786
17.897
18.007
17.570
17.685
17.797
17.908
18.018
17.582
17.696
17.808
17.919
18.029
17.593
17.707
17.820
17.931
18.040
17.605
17.719
17.831
17.942
18.051
8.0
8.1
8.2
8.3
8.4
18.062
18.170
18.276
18.382
18.486
18.073
18.180
18.287
18.392
18.496
18.083
18.191
18.297
18.402
18.506
18.094
18.202
18.308
18.413
18.517
18.105
18.212
18.319
18.423
18.527
18.116
18.223
18.329
18.434
18.537
18.127
18.234
18.340
18.444
18.547
18.137
18.244
18.350
18.455
18.558
18.148
18.255
18.361
18.465
18.568
18.159
18.266
18.371
18.475
18.578
8.5
8.6
8.7
8 8
8.9
18.588
18.690
18.790
18.890
18.988
18.599
18.700
18.800
18.900
18.998
18.609
18.710
18.810
18.909
19.007
18.619
18.720
18.820
18.919
19.017
18.629
18.730
18.830
18.929
19.027
18.639
18.740
18.840
18.939
19.036
18.649
18.750
18.850
18.949
19.046
18.660
18.760
18.860
18.958
19.056
18.670
18.770
18.870
18.968
19.066
18.680
18.780
18.880
18.978
19.075
9.0
9.1
9.2
9.3
9.4
19.085
19.181
19.276
19.370
19.463
19.094
19.190
19.285
19.379
19.472
19.104
19.200
19.295
19.388
19.481
19.114
19.209
19.304
19.398
19.490
19.123
19.219
19.313
19.407
19.499
19.133
19.228
19.323
19.416
19.509
19.143
19.238
19.332
19.426
19.518
19.152
19.247
19.342
19.435
19.527
19.162
19.257
19.351
19.444
19.536
19.171
19.266
19.360
19.453
19.545
9.5
9.6
9.7
9.8
9.9
19.554
19.645
19.735
19.825
19.913
19.564
19.654
19.744
19.833
19.921
19.573
19.664
19.753
19.842
19.930
19.582
19.673
19.762
19.851
19.939
19.591
19.682
19.771
19.860
19.948
19.600
19.691
19.780
19.869
19.956
19.609
19.700
19.789
19.878
19.965
19.618
19.709
19.798
19.886
19.974
19.627
19.718
19.807
19.895
19.983
19.636
19.726
19.816
19.904
19.991
10
20
30
40
20.000
26.021
29.542
32.041
20.828
26.444
29.827
32.256
21.584
26.848
30.103
32.465
22.279
27.235
30.370
32.669
22.923
27.604
30.630
32.869
23.522
27.959
30.881
33.064
24.082
28.299
31.126
33.255
24.609
28.627
31.364
33.442
25.105
28.943
31.596
33.625
25.575
29.248
31.821
33.804
50
60
70
80
90
33.979
35.563
36.902
38.062
39.085
34.151
35.707
37.025
38.170
39.181
34.320
35.848
37.147
38.276
39.276
34.486
35.987
37.266
38.382
39.370
34.648
36.124
37.385
38.486
39.463
34.807
36.258
37.501
38.588
39.554
34.964
36.391
37.616
38.690
39.645
35.117
36.521
37.730
33.790
39.735
35.269
36.650
37.842
38.890
39.825
35.417
36.777
37.953
38.988
39.913
100
40.000
( )
V xo 1
R atio
(=)
Table F.2.
349
A P P E N D IX G
C O N V E R S IO N C H A R T S , TABLES etc.
Conversion of Length
m
cm
mm
ft
in
100
1000
3,281
39,37
0,01
10
0,001
0,1
0,3048
30,48
304,8
0,0254
2,54
25,4
0 ,0328
0,3937
0,00328
0,03937
12
0 ,0833
1
8 0 0 18 4
Conversion of Velocity
m/s
km/h
ft/m in
mile/h
3,6
196,85
2,2369
0,2778
54,68
0,6214
5,08 x 10 3
1,829 x 10 2
0,4470
1,6093
1,136 x 10 2
88
1
8 0 0 18 5
Conversion of Acceleration
9
m/s2
cm / s2
ft/s 2
in /s 2
9,81
981
32,2
386
0,102
100
0,00102
0,01
0,03109
0,3048
30,48
0,00259
0,0254
2,54
3,281
0,0328
1
0,0833
39,37
0,3937
12
1
80 0 18 6
350
Conversion of Area
m2
cm 2
ft2
in 2
yd2
1
10 4
9,29 x 1 0 - 2
6,452 x 10 4
0,8361
104
1
929
6,452
8361
10,764
1550
0,1550
144
1
1296
1,196
0,0011
0,1111
0,0 0 0 8
1
1,0764 x 1CT3
1
6,9 4 4 x 10 3
9
8 0 0 18 7
Conversion of V olum e
m3
I = (dm )3
ft3
gal (UK)
gal (US)
yd3
1
10 3
2,8317 x 10 2
4,546 x 1 0 " 3
103
1
28,317
4,546
3,785
35,315
0 ,0 35315
1
0,1 6 0 5
219,98
0,21998
6,2290
1
1,308
0 ,0 0 1 3
0 ,0 3 7 0
0,0 0 5 9
0,13368
27
0,8326
168,16
264,28
0,26428
7,4805
1,2011
1
201,97
3,785 x 1 0 - 3
0,7646
764,56
0,0 0 5 0
1
8 0 0 18 2
Conversion of Mass
kg
tekma
lb
oz
0,102
1000
2,2046
35,274
9,807
9807
21,6205
345,93
10-3
1 ,0 2 x 10 4
2,205 x 1 0 " 3
3,527 x 10 2
0,45359
4 ,6 2 5 x 10-2
453,59
16
2,835 x 1 0 - 2
2,8908 x 10~3
28,35
6,25 x IQ 2
1
8 0 0 18 3
lb / f t 3
6 ,2 4 3 x 10-2
16,0185
1
8 0 0 18 1
351
Conversion of Force
Conversion of Pressure
Pa
m bar
mm H 2 0
atm
1
100
9,807
1,013 x 105
249,10
6908,9
1 0 -2
1
9,8 0 7 x 1 0 -2
1013
2,491
69 ,0 8 9
0,102
10,917
1
1,0332 x 104
25,4
704,49
9.869
9.869
9,678
1
2,453
6,805
x IO 6
x 1 0 -4
x 1 0 -5
x 10 3
x 10~2
in W G
Ib f/in 2
4,0 2 x 10 -3
0,402
3,937 x 10-2
406,77
1
27,736
1.4 5 0 4 x 1 0 - 4
1 .4 5 0 4 x 1 0 -2
1 ,4 2 2 3 x 1 0 -3
14 ,6 9 6
3 ,6 0 5 x 1 0 - 2
1
800630
kW h
kpm
kcal
Btu
ft Ibf
1
3,6 x 106
9,807
4187
1,055
1,3558
2 ,7 7 8 x I O - 7
1
2,7241 x 1 0 " 6
1,1 6 3 x 103
2,9 3 x 104
3 ,7 6 6 x 1 0 - 7
0,1020
3 ,6 7 1 0 x 105
1
427
107,59
0,1383
2,39 x 1 0 - 4
860
2,3423 x 1 0 - 3
1
0,25198
3,239 x 10 4
9,4 8 x 1 0 "4
34 13
9,2949 x 1 0 -3
3,9685
1
1,285 x IO " 3
0 ,7 3 7 6
2 ,6 5 5 x 106
7 ,2 3 3
3 0 8 7 ,4
7 7 7 ,9 7
1
800629
Conversion of P ow er
352
kW
kpm /s
hk
kcal/h
f t Ibf/s
hp
1
9,81 x 1 0 -3
0,735
1,16 x 1 0 -3
1,36
0,745
2,93 x 1 0 - 4
3,52
102
1
75
0,1 19
0,13 8
76
2,99 x 1 0 -2
35,9
1,36
1,33 x
1
1,58 x
1,84 x
1,014
3,99 x
0,479
860
8,44
632
1
1,17
642
0,252
3024
738
7,23
542
0,858
1
550
0,216
259
1,34
1,32 x
0 ,9 8 6
1,56 x
1,82 x
1
3 ,9 3 x
0,471
1 0 -2
1 0 -3
IO 3
1 0 -4
1 0 -2
IO "3
1 0 -3
10 4
Tem perature:
F = - C + 32
5
C = - ( F - 32)
9
Resonant frequency
w he re
g = g ravitationa l acceleration
A st = static d eflection of the mass
V elocity
Displacement
a0 cos 2 n ft
-J ?an sin 2 n ft
2nf u
- 4 n 2 f2 a o c s 2 * f t
- 2 n fv 0 sin 2 n ft
v0 cos 2 n ft
1
o ,
-v n sin 2 n ft
2nf 0
- 4 n 2f 2d 0 cos 2 n ft
- 2 n fd 0 sin 2 n ft
d0 cos 2 n f t
80 01 80
353
10
too
354
1k
F re q u e n c y
Hz
10 k
A P P E N D IX H
Standards Related to
Vibration and Shock M easurem ents
H . 1. G e n e ra l
ste rre ich isch e s N o rm u n g s in s titu t
L eopoldsg 4
1 0 2 0 W ien
M e ch a n is c h e S c h w in g u n g e n ; E rs c h tte
ru n g e n
A u s tr ia
S 9 00 1
S c h w in g u n g s le h re ; K in e m a tis c h e B e
g riffe
S 9100
C .S .S .R .
S N 01 1 3 1 2
1975
SN 01 1 3 9 0
M e th o d s of
m e c h a n ic a l
M e c h a n ic a l
e q u ip m e n t.
S N 01 1391
M e a s u re m e n ts of
vib ra tion s
vib ra tio n m e a s u rin g
G en e ra l te rm in o lo g y
SN 01 1 4 0 0
1975
S N 01 1401
W irk u n g
m e c h a n i
TGL 2 2 3 1 2 / 0 3 A rb eitsh ygien e ;
scher S c h w in g u n g e n a u f den M e n
1 97 1
schen; M e m e th o d ik f r G anzkrper
sch w in g u n g e n
TGL 2 2 3 1 2 / 0 4 A rb e itsh yg ie n e ;
W irk u n g
m e c h a n i
1972
scher
S c h w in g u n g e n
a uf
den
M enschen; G re n z w e rte f r T e ilk rp e r
s c h w in g u n g e n
TGL 2 2 3 1 2 / 0 5 A rb eitsh ygien e ;
W irk u n g
scher S c h w in g u n g e n a u f
1972
schen; M e m e th o d ik f r
s c h w in g u n g e n
G r e a t B r ita in
BS 3 0 1 5
1958
H i in n a r v
"
Y
SN 02 8 9 0 2
SN 12 3 0 6 2
M .S z 4 9 0 0 / 2
-7 8
V ib ra tio n s et chocs m c a n iq u e s
v o ca b u la ire
M th o d e s d 'essais Essais g n rau x
c lim a tiq u e s et m ca n iq u e s. G uide pour
l essai de chocs sp ectre s de chocs et
a u tre s c a ra ct ristiq u e s des fo rm e s de
ch ocs
B e u th V erlag G mbH
B u rg g ra fe n s tr 4 10
1 0 0 0 B erlin 3 0
S c h w in g u n g s le h re
Teil 1 K in em a tisch e B e g riffe
Teil 2 E infache S c h w in g e r
Teil 3: S chw ing system e
Teil 4 S chw ing e nd e K o n tin u a
F ra n c e
NF E 9 0 -0 0 1
1972
C 20408
Ja n . 1 9 7 4
G e rm a n y
(BR D )
DIN 131 1
E rs c h tte ru n g e n im B a u w e s e n , G ru n d
stze
K la ssie rve rfa h re n f r das E rfassen
DIN 4 5 6 6 7
re g e llo s e r S c h w in g u n g e n
A m t f r S ta n d a rd is ie ru n g , M e w e sen
G e rm a n y
u nd W a re n p r fu n g der D e u tsch e n De
(D D R )
m o kra tis c h e n R epublik
1 0 2 6 B erlin
W a lls tra e 1 6
TG L 0 - 1 3 1 1 / 0 1 S c h w in g u n g s le h re ; B e n e n n u n g e n
1963
DIN 41 5 0 T
TGL 0 -1 3 1 1 /0 2 S c h w in g u n g s le h re ; E in fa c h e S c h w in
ger
1963
S c h w in g u n g s s y
TGL 0 -1 3 1 1 /0 3 S c h w in g u n g s le h re ;
ste m e m it e nd lich vie le n F re ih e itsg ra
1963
den
W irk u n g
m e ch a n i
TG L 2 2 3 1 2 / 0 1 A rb e its h y g ie n e ;
s c h e r S c h w in g u n g e n a u f den M e n
1 971
sch en ; B egriffe
W irk u n g
m e ch a n i
TGL 2 2 3 1 2 / 0 2 A rb e its h y g ie n e ;
s c h e r S c h w in g u n g e n a u f den M e n
1971
sch en ; G re nzw erte f r G an zk rp er
sch w in g u n g e n
m e c h a n i
den M e n
T e ilk rp er
N e t h e r la n d s
NEN- ISO 2 0 4 1
Sam e as (ISO 2 0 4 1 - 1 9 7 5 )
P n la n H
PN 7 5
M -5 3 5 2 7
Sw eden
SS 4 0 1 1 4 0 1
M C A
S1 1 - 1 9 6 0
(R 1 9 7 6 )
c; R
G ost 1 3 7 3 1 - 6 8
G ost 1 6 8 1 9 -7 1
K o m ite t S ta n d a rto v ,
L en in sky P ro sp e kt 9 b,
1 1 7 0 4 9 M o skva M - 4 9
M e ch a n ica l v ib ra tio n . G eneral
re q u ire m e n ts fo r m e a su re m e n t
perfo rm an ces
V ib rom ete rs.
Term s and d e fin itio n s
355
S tandardization.
1. Rue de V a re m b ^ ,
G eneva. S w itz e rla n d
50
5 0 (2 0 ) (1 9 5 8 )
5 0 (4 1 1) ( 1 9 7 3 ) R otating m a chines
I n t e r n a t io n a l
(I S O .)
R .1 9 2 5 -1 9 7 4
B alancing V o c a b u la ry
R 2 0 4 1 -1 9 7 5
DP 4 8 6 3
Flexible co u p lin g s G u id e fo r
specifica tio ns, s e le c tio n and
a pplica tio n
DP 4 8 6 5
A n a lo g analysis a n d p re s e n ta tio n of
vib ra tio n and shock d ata
DP 4 8 6 6
DP 5 3 4 5
DIS 5 8 0 5
DIS 5 9 8 2
E 9 0 -6 0 0
1970
E 9 0 -6 0 1
1973
G e rm a n y
/O D i-M
(b n U )
DIN 4 1 5 0
E rs c h tte ru n g e n im B au w e sen
T 2: E in w irk u n g a u f M e n sch e n
T 3: E in w irk u n g e n a uf b a u lic h e A n la
gen
VDI 2 0 5 6
B e u rte ilu n g s m a s t b e f r m e c h a n is c h e
S c h w in g u n g e n von M a sch in e n
VD I 2 0 5 7
G uide to s a fe ty aspects of h u m a n
v ib ra tio n e xp e rim e n ts
DD 3 2
G uide to th e e va lu a tio n of h u m a n
expo sure o f w h o le body v ib ra tio n
DD 4 3
H u n q a rv
H . 2 . V ib r a tio n R a tin g R e c o m m e n d a tio n s
A u s t r ia
S 9010
S 9020
C .S .S .R .
O ffice fo r S ta n da rd s a nd
M e a sure m e n ts.
1 1 3 4 7 Praha 1.
V clavsk N m sti 1 9
CSN 01 1 4 1 0
1 971
D e n m a rk
A rb e jd sm ilj fo n d e t
V esterbrogade 69
1 6 2 0 Kbenhavn V
Hvide Fingre
V ibratio n sskad e r fra h n d v rkt je r
F in la n d
F ra n c e
E 9 0 -4 0 0
1971
356
KG S T 7 1 5 -7 7
N e th e r la n d s
NE N- ISO 1 9 4 0 S am e as (IS O 1 9 4 0 - 1 9 7 3 )
NEN- ISO 2 3 7 2 S am e a s (IS O 2 3 7 2 - 1 9 7 4 )
NEN 2 2 3 7 3
S am e as (IS O 2 3 7 3 - 1 9 7 4 )
S am e as (IS O 2 6 3 1 - 1 9 7 4 )
P o la n d
P N -7 3
E -0 6 0 2 0
U .S .A .
Y le iset ko n e te k n illis e t
tu rva llis u u s o h je e t
A N S I S 2 .19
1975
U .S .S R
B eu th V e rla g G mbH
B u rg g ra fe n s tr 4 10
1 0 0 0 B e rlin 3 0
G o st
G ost
G o st
G o st
G o st
GOST 2 3 5 5 2
-7 9
In te r n a tio n a l
(I.S .O .)
R .1 9 4 0 -1 9 7 3
2 3 7 2 -1 9 7 4
2 3 7 3 -1 9 7 4
R 2 6 3 1 -1 9 7 4
DP 4 8 6 7
Code fo r th e m e a su re m e n t fo r
re p o rtin g of sh ipb o ard V ib ra tio n
data
DP 4 8 6 8
E 9 0 -2 1 0
1970
E 9 0 -2 1 1
1970
E 9 0 -2 1 3
1970
G e rm a n y
(B n U )
B eu th V erlag G m bH
B u rg g ra fe n str 4 10
1 0 0 0 B e rlin 3 0
DIN 4 5 6 6 1
S c h w in g u n g s m e g e r te ; B egriffe,
K e n n g r e n , S t rg r e n
DIN 4 5 6 6 2
E ig en scha ften vo n S c h w in g u n g s m e
g e rte n, A n g a b e n in Typ en b ltte rn
DIN 4 5 6 6 4
A nko p plu n g vo n S c h w in g u n g s m e g e r
ten und b e rp r fu n g a u f S t re in fl sse
DIN 4 6 6 6 6
S c h w in g s t rk e m e g e r t; A n fo rd e ru n
gen
DIN 4 5 6 6 9
Teil 1
E n tw u rf
M e ssu ng von
S c h w in g u n g s im m is s io n e n
(S c h w in g u n g sm e sse r)
G e rm a n y
Amt
S ta n d a rd is ie ru n g ,
M e w e se n
und
W a re n p r fu n g
der
D eutsch en D e m o kra tisch e n Republik
1 0 2 6 B erlin
W a lls tra e 1 6
DP 5 3 4 3
DIS 5 3 4 9
(D D R )
DP 6 8 9 7
G uide to th e e va lu a tio n o f th e
re sp on se o f o ccup an ts o f fixe d
stru c tu re s , e spe cia lly b u ild in g s , and
o ff-s h o re stru c tu re s to lo w fre q u e n cy
h o riz o n ta l m o tio n ( 0 ,0 6 3 Hz to 1 Hz)
N e t h e r la n d s
H .3 . V ib r a tio n M e a s u r in g E q u ip m e n t
NEN 1 0 2 1 7
C .S .S . R .
O ffic e fo r S ta n da rd s and
M e a su re m e n ts.
1 1 3 4 7 Praha 1
V clavsk N am sti 19
CSN 01 1391
1970
M e c h a n ic a l v ib ra tio n m e a su rin g
e q u ip m e n t G en e ra l te rm in o lo g y
SN 3 5 6 8 5 0
1972
In s tru m e n ts fo r v ib ra tio n
m e a su re m e n ts. T e ch n ica l
re q u ire m e n ts
SN 3 5 6 8 5 8
1975
F in la n d
S u o m e n S ta n d a rd is o im is liitto
PL 2 0 5 , 0 0 1 2 1 H e ls in k i 12
SFS 2 8 8 2
S am e as IEC 1 8 4 ( 1 9 6 5 )
SFS 2 8 8 3
S am e as IEC 2 2 2 ( 1 9 6 6 )
F ra n c e
E 9 0 -1 0 0
1973
In s tru m e n ts de m e su re des
c a ra c t ris tiq u e s d in te n s it vib ra toire
des m a ch ine s to u rn a n te s ou
a lte rn a tiv e s
E 9 0 -1 1 0
1973
E 9 0 -2 0 0
1970
S am e as (IEC 2 1 7 - 1 9 6 7 )
Sam e as IEC 3 4 8 - 1 9 7 1 , 3 4 8 A -1 9 7 4 )
S p a in
21 3 2 8 7 5 ( 1 )
21 3 2 8 7 5 ( 2 )
Clases de c a p ta d o re s de vib ra ci n y
e lem en tos
se n sib le s
em pleados en
estos capta do re s
21 3 2 8 7 5 (3)
U S A
A m e rica n N a tio n a l
S ta n da rd s In s titu te ,
1 4 3 0 B ro a d w a y,
N e w Y ork, NY 1 0 0 1 8
S hock
357
A N S I S 2 .4 1976
M e th o d o f S pe cifyin g th e C h a ra cte ri
stics of A u x ilia ry E q u ip m e n t fo r Shock
and V ib ratio n M e a s u re m e n ts
A N S I S 2 .5 1 9 6 2 (R 1 9 7 6 )
A N S I S 2 .1 1
1 9 6 9 (R 1 9 7 8 )
ANSI S 2 .1 4
1 9 7 3 (R 1 9 7 8 )
A N S I S 2 .1 5 1 9 7 2 (R 1 9 7 7 )
A N S I Z 2 4 .2 1 1957
(R 1 9 7 8 )
DIS 5 3 4 4
E le ctro d yn a m ic te st e q u ip m e n t fo r
g e n e ra tin g v ib ra tio n M e th o d s of
d e scrib in g th e c h a ra cte ristics o f th e
e q u ip m e n t
DP 5 3 4 7
S ta n da rd m e th o d s fo r th e c a lib ra tio n
of v ib ra tio n and shock pickups
DP 5 3 4 8
DIS 6 0 7 0
H .4 . M e a s u r e m e n t of Vibration of M a c h i n e s
Austria
IS A S 3 7 .2
1964
U.S.S.R.
K o m ite t S ta n da rto v,
L en in sky Prospekt 9 b,
1 1 7 0 4 9 M oskva M - 4 9
O A L -R ic h tlin ie
N r. 4
O A L -R ic h tlin ie
Nr 7
S ch w in g u n g s g e d m m te M a s c h in e n a u f
ste llu n g
Bulgaria
G ost 5 .2 9 5 - 6 9
B DS 5 6 2 6 - 6 5
G ost 5 .3 0 4 - 6 9
C.S.S.R.
V ib ro m e te rs
B asic param e ters
International
(I.E.C.)
In te rn a tio n a l O rg a n iz a tio n fo r
S ta n da rd iza tion ,
1, Rue de V arem b6 ,
G eneva, S w itz e rla n d
1 8 4 (1 9 6 5 )
M e th o d s fo r sp e cifyin g th e
ch a ra cte ristics o f e le c tro -m e c h a n ic a l
tra n sd u ce rs fo r sh ock and vib ra tio n
m e a sure m e n ts
2 1 7 (1 9 6 7 )
E le ctro n ic vo ltm e te rs
2 2 2 (1 9 6 6 )
M e th o d s fo r sp e cifyin g the
c h a ra cte ristics of a u x ilia ry e q u ip m e n t
fo r shock and v ib ra tio n m e a su re m e n t
2 2 5 (1 9 6 6 )
3 4 8 (1 9 7 8 )
International
(I.S.O.)
In te rn a tio n a l O rg a n iz a tio n fo r
S ta n da rd iza tion ,
1 ' Rue de V a re m b e ,
G eneva, S w itz e rla n d
R 2 3 7 1 -1 9 7 4
R .2 9 5 3 - 1 9 7 5
R 2 9 5 4 -1 9 7 5
DIS 3 7 1 9 2
358
S N 0 2 8 9 0 2
1 97 1
CSN 12 3 0 6 2
1 96 1
S N 12 3 0 6 3
1971
France
E 9 0 -3 0 0
1973
V ib ra tio n s m ca n iq u e s des m a c h in e s
ayant u ne vite sse de fo n c tio n n e m e n t
co m p rise e n tre 1 0 et 2 0 0 to u rs par
seconde v a lu a tio n de l'in te n s it
vib ra to ire
E 9 0 -3 1 0
1973
V ib ra tio n s m ca n iq u e s des m a ch in e s
le c triq u e s to u rn a n te s de h a u te u r
d'axe c o m p ris e s e n tre 8 0 et 4 0 0 m m
v a lu a tio n de l'in te n s it v ib ra to ire
H 0 0 -0 4 2
1971
H 0 0 -0 4 3
1971
Essai de v ib ra tio n
NFC 2 0 6 1 6
AVR 1 9 7 3
Germ any
(BRD)
DIN 4 5 6 6 5
DIN 4 5 6 6 8
A n k o p p lu n g f r S c h w in g u n g s a u fn e h
m er zu r b erw a chu ng von G ro m a
sch ine n
DIN ISO 2 3 7 3
M e c h a n isch e S ch w in g u n g e n vo n
u m la u fe n d e n e le ktrische n M a s c h in e n
m it A c h s h h e n von 8 0 4 0 0 m m
VDI 2 0 5 9
E n tw u rf
VD I 2 0 6 0
B e u rte ilu n g sm a st b e f r d e n A u s
w u c h tz u s ta n d ro tie re n d e r s ta r re r K r
per
G r e a t Britain
B S .4 6 7 5 : 1971
R 1 9 4 0 -1 9 7 3
R 2 3 7 2 -1 9 7 4
R .2 3 7 3 - 1 9 7 4
R 3 0 8 0 -1 9 7 4
3945 - 1977
DIS 5 4 0 6
M .S z 1 9 4 2 6 -7 7 Forgogpek rezgserssgnek m r se
KGST 7 1 6 -7 7
Netherlands
M o to ro s kziszerszm ok
zinek m r si mdszerei
H . 5. V i b r a t i o n T e s t i n g
Belqium
re z g s je lle m -
C 0 6 -2 0 6
PN 7 3
E 04255
Roumania
STAS 7 5 3 6 6 6
U.S.A.
A F B M A S ta n d a rc
No. 1 3 ( 1 9 6 8 )
U.S.S.R.
Fu nd a m n tele k lim a to lo g is c h e en
m echanische b e p ro e vin g sm e th o d e n :
Deel II: Proeven; P ro ef Fc:
(S in usvo rm ige ) t rillin g e n
C 06 229
France
UTE C 2 0 - 4 0 8
1974
NF C 2 0 - 5 0 8
1975
NF C 2 0 - 5 2 3
1975
G ost 1 2 3 7 9 -6 6
NF C 2 0 - 5 2 4
1975
G ost 1 6 5 2 9 - 7 0
NF C 2 0 - 5 2 6
1975
NF C 2 0 - 5 2 7
1975
NF C 2 0 - 5 2 8
1975
NF C 2 0 - 5 2 9
1975
NF C 2 0 - 6 0 8
1974
359
NF C 2 0 - 6 1 6
1973
NF C 2 0 - 6 2 3
1973
NF C 2 0 - 6 2 4
1973
NF C 9 0 - 1 6 3
1967
E 9 0 -2 0 0
1970
E 9 0 -2 1 0
1978
H 0 0 -0 4 3
1 97 1
B euth V e rla g G m bH
B u rg g ra fe n s tr. 4 10
1 0 0 0 B e rlin 3 0
G erm any
/O D n \
( M U )
DIN 4 0 0 4 6
DIN 5 0 1 0 0
D a u e rs c h w in g v e rs u c h e . B e g riffe , Z e i
chen. D u rc h f h ru n g . A u s w e rtu n g
DIN 5 1 2 2 8
D a u e rs c h w in g p r fm a s c h in e n
G r e a t Britain
2106
360
Nagy
fo rg g p e k
rezgserssgnek
helyszini vizsg la ta s m in sitse
Indian S ta n d a rd s In s titu tio n ,
M anak B ha va n
9 B ahadur S h a h , Z a fa r M arg,
N ew D e lh i 1 1 0 0 0 2
E n viro n m e n ta l te s ts fo r e lectron ic and
e le ctrica l e q u ip m e n t:
Part VII B um p T e st (1 9 6 4 )
Part VIII Im p act o r sh ock test (1 9 6 4 )
Part IX D rop te s t (1 9 6 4 )
Part XIV C o n s ta n t a ccele ra tio n test
(1 9 6 6 )
Part XVI V ib ra tio n te s t (1 9 7 1 )
Japan
J a p a n e s e S tandards A s s o c ia tio n
1 -2 4 , A kasaka 4 ch m e, M in a to -k u
T okyo
JIS C 0 91 1
(1 9 6 0 )
JIS C 0 9 1 2
(1 9 6 0 )
JIS 1601
(1 9 6 7 )
JIS B 6 0 0 3
(1 9 6 2 )
JIS W 6 05 1
(1 9 5 4 )
JIS W 6 0 5 3
(1 9 5 5 )
JIS W 6 0 5 4
(1 9 5 5 )
Poland
P N -7 5
0 -7 9 1 6 6
Sw eden
SEN 4 3 1 6 0 0
M ilj p ro v n k ig av
e le k tro n ik k o m p o n e n te r, a llm n t
SEN 4 3 1 6 0 5
M ilj p ro v n in g av
e le k tro n ik k o m p o n e n te r, ska kprov
SEN 4 3 1 6 0 6
M ilj p ro v n in g av
e le k tro n ik k o m p o n e n te r, v ib ra tio n
U S.A.
M IL
STD 81 OC
M IL
E 4158 E
E le c tro n ic E qu ip m e nt G ro un d ; G e n e ra l
R e q u ire m e n ts for
M IL
E 5272 C
M IL
E 5400 P
MIL
T 5422 F
M IL
E 16400 G
M IL
T 21200 L
M IL
STD 2 0 2 E
Test M e th o d s fo r E le ctron ic a nd
E le c tric a l C om ponents
NATO
STAN AG 3 5 1 8
E n v iro n m e n ta l Test M e th o d s fo r
A ir c r a ft E q u ip m e nt and A sso cia te d
G ro u n d E qu ip m e nt
International
(I. E.C.)
68
68 -1 (1 9 6 8 )
P a rt 1. G e n e ra l
6 8 -1 A (1 9 7 4 )
6 8 -2
Part 2. T e sts
6 8 - 2 - 6 (1 9 7 0 )
6 8 - 2 - 7 (1 9 6 8 )
G r e a t Britain
BS 1 8 8 1 : 1 9 7 0
BS.AU 125:
1966
DD 4 7
Netherlands
Roumania
6 8 - 2 - 2 7 (1 9 7 2 ) Test Ea: S ho ck
6 8 - 2 - 2 9 (1 9 6 8 ) Test Eb: B u m p
STAS 8 0 4 8 - 6 7
H .6.
Measurement
Dynamic
of
Properties of M aterials
F ra n rp
U.S.A.
ASTM
C 2 1 5 -6 0
ASTM
C 5 9 7 -7 1
NFS 3 1 0 5 1
F E V .1979
ASTM
D1 5 7 7 -6 6
NFS 3 1 0 5 2
FEV 1 9 7 9
IDEM - M e s u re en laboratoire de la
tra n s m is s io n du b ru it de choc par le s
pla n ch e rs
IDEM - M e s u re de la r d uctio n de la
tra n s m is s io n du b ru it de choc par les
re v te m e n ts de sol et les dalles
flo tta n te s
ASTM
D 2 2 3 1 -7 1
NFS 3 1 0 5 3
F E V .1979
NFS 3 1 0 4 9
FEV. 1 9 7 9
M e su re du p o u v o ir d 'iso latio n
a co u stiq u e des l m e n ts de
co n s tru c tio n et de l'iso le m e n t des
im m e u b le s, (fid lit )
NFS 3 1 0 5 0
FEV 1 9 7 9
G e rm a n y
,D D m
(B R D )
B eu th V e rla g G m bH
B u rg g ra fe n s tr 4 - 1 0
1 0 0 0 B e rlin 3 0
DIN 5 2 2 1 4
B a u a ku stisch e P r fu n g e n B e s tim
m ung der d yn a m is c h e n S te ifig k e it von
D m m s c h ic h te n f r s ch w im m e n d e B
den
DIN 5 3 4 2 6
DIN 5 3 4 4 0
P r fu n g vo n K u n s ts to ffe n und s c h w in
g u n g sg e d m p fte n g e sch ich te ten S y s te
men
B ie g e sch w in g u n g sve rsu ch .
Teil 1: A llg e m e in e G rundlagen
Teil 2: B e s tim m u n g des kom plexen
E la stiz it ts m o d u ls
Teil 3: M e h rs c h ic h ts y s te m e
DIN 5 3 4 4 5
P r fu n g vo n K u n ststo ffe n ;
T o rs io n s s c h w in g u n g s v e rs u c h
DIN 5 3 5 1 3
DIN 541 19
G u id e fo r D escribing the
C h a ra c te ris tic s o f R e s ilie n t M o u n tin g s
ANSI S2 9
1976
International
(I. S O . )
2 0 1 7 -1 9 7 2
DP 5 4 0 5
361
INDEX
tu n a b le f i l t e r .................................................................................................................. 1 3 0
tu n in g , absorber .......................................................................................................... 3 1 5
unbalance ...................................................................................................................... 2 7 3
ve lo c ity ....................................................................................................... 9 7 , 3 3 5 , 3 3 9
vib ra tio n absorber .......................................................................................... 3 1 1 3 1 8
vib ra tio n calibrator ........................................................................................ 1 1 6 1 2 0
v ib ra tio n exposure c rite ria ......................................................................... 9 0 , 9 1 , 93
vib ra tio n m easurem ent scheme (s u m m a ry )......................................................... 15 8
v ib ra tio n meter .............................................................. 1 2 9 13 5, 1 7 5 , 2 0 6 , 2 8 0
vib ra tio n testing .............................................................................................. 2 3 7 2 5 5
v ib ra to r, e lectrohydraulic ........................................................................................... 2 3 8
vib ra to r, electrodynam ic .............................................................................. 2 3 9 2 4 2
viscoelastic m aterials .................................................................................... 3 1 9 3 2 2
voltage am plifier .......................................................................................................... 112
voltage sensitivity ......................................................................................................... 112
w ave effects .................................................................................................................. 2 9 5
w ave-shape ............................................................................................... 5 1 , 3 3 4 , 3 3 5
w ave-shape d isto rtio n ................................................................................................ 3 3 4
w a ve fo rm analysis ........................................................................................ 1 4 3 14 4
w h ite fin g e r ...................................................................................................................... 94
w hole -b o d y vibration ......................................................................................... 8.8, 132
w id e band test .............................................................................................................. 2 4 8
W ie n e r-K h in ch in re la tio n s ........................................................................................... 3 0
W o h le r Kurve ....................................................................................................... 7 4 , 2 3 7
X-Y recorder ................................................................................................................... 157
zero s h ift ......................................................................................................................... 106
zoom analysis ................................................................................................... 1 8 1 , 1 87
370