DYNAMIC MODULUS ESTIMATION AND

STRUCTURAL VIBRATION ANALYSIS*

A. Gupta, S. Khandaswamy, S. Yellepeddi

Department of Mechanical Engineering
Northern Illinois University
DeKalb, IL 60115

T. Mulcahy, J. Hull
Energy Technology Division
Argonne National Laboratory
Argonne, IL 60439

The submitted manuscript has been created by the

University of Chicago as Operator of Argonne Nstional
Laboratory (“Argonne”) undw Contract No. W-31-109-ENG-
38 with the U.S. Department of Energy. The U.S.
Government retsina for itself, and others acting en R behaif,
a paid-up, nonexclusive, irrsvo+eble worldwide Iicanss in
said artiite 10 rspmdwca, prepare derivative works, distribute
copies to the public, and perform pubfiily and display
JMiafv, bv or cm bshati of the Govammant.

Distribution
R. W. Weeks
R. B. Poeppel
U. Balachandran
R. A. Valentin
Authors
ESA Section
ET Division File
F. Y. Fradin
H. Drucker
S. Lake

To be presented at 17th International Modal Analysis Conference, Kissimmee, FL,

February 8-11, 1999.

*Work partially supported by the U.S. Department of Energy, Energy Efficiency and
Renewable Energy, as part of a program to develop electric power technology, under
Contract W-31 -109-Eng-38.
 DISCLAIMER

This report was prepared as an account of work sponsored

by an agency of the United States Government. Neither the
United States Government nor any agency thereof, nor any
of their employees, make any warranty, express or implied,
or assumes any legal liability or responsibility for the
accuracy, completeness, or usefulness of any information,
apparatus, product, or process disclosed, or represents that
its use would not infringe privately owned rights. Reference
herein to any specific commercial product, process, or
service by trade name, trademark, manufacturer, or
otherwise does not necessarily constitute or imply its
endorsement, recommendation, or favoring by the United
States Government or any agency thereof. The views and
opinions of authors expressed herein do not necessarily
state or reflect those of the United States Government or
any agency thereof.
 DISCLAIMER

Pofiions of this document may be illegibie

in electronic image products. Images are
produced from the best available original
document.
 .
. .

DYNAMIC MODULUS ESTIMATION AND STRUCTURAL

VIBRATION ANALYSIS

A. Gupta, S. Khandammnzy, and S. YeIlepeddf T. Mukahy and J. Hull

NorthernIllinois University Argonne National Laboratory
DcKdb, IL 601 M ~gonne, IL 60439

4, = half-power bandwidth for modes of ownpositc bar, Hz

/n E rexmance fiequmcy for made n of Obast bar, Hz
/s = resonaha i%equsnoy f6r modes of coqmsite bar, Hz
ABsrmcr s = indanum&. 1, Z3 . . . . ..(5sn)
! = kqth of bar, m
Ofb, the dynamic el~-c modulus of a ~“al with frequency
dapmdent propettics is diffioult to CSlmstC
- Ill- Uncetraincics
. am compounded in any struotuxai viion analysis udng the I. lNTRODUCTXON
-“al propmics. HZ different G%PemM@ “ ?CehrlquCs *
Uscdwcsthate the pmpcrties of a particular elastomeric mmerial Viscoeksic materials due to their low elastic tnduhis end high
ev~a btoedftcqueq rage. k fhepropertiesaredetemind loss modulus ate wed in many cases. However, modulus of a
various Suucturq incorpocaring the aIastom arc Soalpd by an viaoOCl*”c rnsmridis highlydcpcndsmon ntsoypsmncms Suoh
itemtivcfinite c19ncotmcdlcci mdetcmdnenmai ffequcnciesd as fqlency. tempmtwc main amplitudq prdoad etc. [1]. Due
mode shapes. Then, the finite ekmsnt nsuits are arrelated with to m“daapread use of viscdastic matdda in various indusaics
resuIts obtained by axperhncnml modal anelysk inoludiig automoti% biimechanid, and StmCtum4 WIixls test
mcthodsbavcbecn pmposcdtocha=Kzaizethsacmataialsandno
single or test ntctld has &n adopted univaraalIy.
NoMENcLAnJRE
.
Onead method i Wets bar method[2]. !%= modulus of tic
A = (MY(2+DTW?) metm”aI beiig inWStigMd is vety low, a sandwich ap&rnco as
Ac = aoas-seuion of cyiindtical qAmen, * shown in FW I was tested in a Bmel & Kjaarqmratus P].
B = 1/6(1+7)2
C. = c4cfIioient for mode n, of claped-f~ (Oberst) bar r’
c, = 0-s5959
C2 - 35~9
G = 9.8194
D = p, /p, density ratio
E = Young’s modulus of Obemt bar, Pa
EC _ complut rrmdulua R E +jq
G ~ ah- m~~]~ of ~wl~~ ~, pa
~ = thj~~f~~,m
HZ = thkbl= of ~~c ~~, m
L ==length ofoylinddcal ~“ntcn, m
M _ mass of metal dd on top of speck% Kg
M, = M of qmcimcn = p, AL Kg
T = H, ~ thickn~ ratio
~ = loss ftior of viscoelaatic matctial. dimensionless
% = M, /~, Ios fa~or of sandwiched spaiman, dimensionless
P - density of Obcrst bar, kg/ml
p, = density of damping matciial, kg/m’ Fig. 1. Sandwich Beams fw Obcrst exdment

“9N3 “H23W 8S :6a 366 T-SI-130

 . . .
,-.

Nahmd ftcquencics and damping ratb were WSSUrCdfirstfor the and loss factor is given by
bare specimen with antikvcr end conditions and then for the
sandwiched specimen. Magnaic excitors and pickups were used
A ~,
to avoid msss Iosdiig of the bearm This method is apparently
popukar in automotive ind~ wh~ diffkrent mstaial.s are
~= (2)
(A- B)-2(A-E12-2(AV)
cxplomd for thm”rmud d viirarion damping applications.

Another method (hcreafler mf~ to as tripd nvmhod) has bean

proposed by Niel*rI cd 54]. k this mctbod oylindrioel wst For tie uipod method alwm in Fitguq% the complex modulus EC
specimens * mccd plates at one end is compressed by an can bc solvd from the mfs urement of amplian= [4] as
inverwd shaker as shown in Figure 2. Measuring the force and
accderation (an impedance hesd was ussd to messure both) the
AL(Q) = Sin(p)
rnoduhta and loss fsctor oan be calculated using a numerical compliancc(~) =
fo rcc(u) Q:~,$XCO@-MXS~~) (3)
itomtivc ~ Strain gages were bondsd to the S~irncn to
monitor strain due to prehwi as well as dynamic strain. 71ds
method (so cdlcd non-resonant method) is used to compute
~ and loss f-or continuously over a wide frsaumcy rarwe.
.

RESULTS

l%e dimsmdons of wo sandwich - for the Obcrsl mctid

=40~rn R~l. ~etimtirnf~(I)d@)
are based on cantilever end conditions. In order to verify &o end
editions natural fkqrenciesof the bare bars in O&m appmtus
wete meaaurd and ate prasentai in Teblc I along wish the
theoretical values. It maybe notsd that the first mode rssults ere
ignored as suggostd in the test procedure @].

*
Beam Thickness Theoretical Ex#mental
Fig. 2, ?Yipod experimental setup (m) Frequency < Frquancy &
(HZ) (HZ)
h was decicki that s structure mode of the same viscoelasrk
0.9144 66.8 65.6
material -uld be t~td by cxaml modal analysis to 187.4 184.3
determine its natural ~eneies. The same stmctura would also
3673 361.3
be modelled by finite element method using ANSYS [5] and the
stifhcss of theviwelam “cmaterielwouldbe djxtd tornmch the 1.524 111.4 111.2
expa%emal result 312.4 311.8
I 612.1 611.8
I
Table 1: Nanml ikquencics of bare hers
2. THEORY
So it seems that cantilever bwtdary conditions is satistkl
approximately for the thinner beam whmeas ~1 is vq wctl
for the thicker beam.

Next naturalfrequenciesand damphg ratios of the ssndwich

2xC EHH2 specimens were reared by u$iu the o~~ aPPMUW ~d
[(A -B) - 2(A - ~)> -2 (AT@] -
G. P .(1) magnetic aenso~ connected to Bruel & Kjacr 3550 frsquency
analyzer and wansfeting the data to a computer where STAR
MODAL [6] is used to estirnatc natural frquencics and demping
ratios. The values obtained am prcsmted in Tabte I
 . .
“..

Sandwich Natuml Damping

S~imut # Frequency ~ Ratio q,
(Hz) ●/*

1 200.4 15.0
418.4 20.5

2 228.8 13.0
493.9 13.6
783.7 2.24
Table 2: Sandwich -

Combmingrsaults of both sandwich spdnans, bawd on the

information presentedin Tables 1 and z shsar modulus and loss
factor (twice the damping ratio) were estkmsl usiag formulae
(1) and (2). Assuming a Poisson’s ratio of 0.45, elastic modulus
was also estimated. Fig. 3. E for sample 3 ac E-+ -in amplitude

Frequency

‘
(f@

200.4
228.8
418.4
493.9
shear
Modulus G
(MPa)

2.987
3.691
8.679
10.483
IAss
Factor q

0.408
0384
0S48
0.4s4
8.66
I
10.70
25.17
30.40
Elastic
Modulus E
(MPa)

“A“
783.7 13.4 0.1 38.86
Table 3: Propertiesof viscoelaetic materiel by Oberst
Cxpcriment
II
ll\]

Thcra Wem four awnpks reared to eatunare

. dynamic modulus by
m-pod method folIowing equation (3). W samples have
dimensio~ as ~ Table 4.

Sample # Length L Diamcts Msas of

(mm) Qnrn) End Plate M
m)
1 49.68 I 8.5 2.2
. Fig. 440ss factor fOr a&lc 3 at 403M5stniin amplitude
2 45.43 26.54 &s

3 41.85 34.73 133

43
Table 4: Specimens for mipod tedng

Moddii were obtainrd over frequency range ~icaIly up to 450
Hz as shown in Figuru 3 and 4. Beyond 450 HZ the cestdts WGR
cmoncous(probablyefkctai by thercsotmna$. In order to
comparethis tes dara wirh other test methods, the rnodulii
ohind for diff.t specimens wi& diflcmnt ~stmins end
strain amplitudes et four ficquenciss ( 100 HZ 200 Hz 300 Hz
and 400 Hz) are presented in Table 5.
h maybe rrwmtioncd that in the Obcrst tcs method, srmin
amplitude is not catatant and it varies along the lengthdueto
oantilevcr gaonwtry whereas for the tripod test stminis castant
throughout the sample which is uncles oompeion. In od= ~0
~e -ul~ !hn both of these methods, avemge vaiues of
elastic modulus and 10SSt%tor (without consisting eff= of
pre-stmin and stmin amplitude) arc prescntdin Tablc 6.
 T

Wm >cMPLXJ4C)LS REAL MEAS MAIN ‘f: ::.;;

Y: IOOM [1 x:
x: IOOHZ + 800Hz LXN *

#AVM 10 YMAX : 89.5M

i
P
I

-50E6
1---- —--—. —..—.—. — . .. — . . ..-. -. ——— —.. ——.. .— - —— —— .-. .— .- --- .. . . .. . . .
--i

-iooE6

setup w
Measurement: DUAL-CH . SPECTRUM AVERAGING 800 lines
Trigger: GENEFiATOH

Averaging: L314 10

;;:; . SPAN 800Hz df: IHz l-: 1!3 dt: 977us

., Centre Freq: 500Hz

W&ixhting: ~;~F&NGLJLAR
. DIRECT /-0 , 71+z -\oN 22. 5mV/un i t
ch:B; - DIRECT /-0 . 7Hz -\oN Ss 10, 21nv/m/ea
Generator: &TXSINE O.OdB ref 1.00V WHITE DC Ouv
Repetition: CO14T
 Wm >LOSSJACTOR MAIN Y: 435mlJ
Y: i.00IJ RMS LIN [1 x: 400Hz
x: 100tiz + 800t-lz LIN
#AVM 10 YMAX : 43.OU
1.0 ~A-.-d..d

T
{ L
I
000 1--- -—.—.. —. — ..- —— -—.—. -—.. -——..... — . . ..— .— -4

;
I
+
J

—. ——
...—.—
i
I

0.6
1---
R
I

O*4 ..-— -------

0.2 —— .

1
0 L- ~—--”---” , ~ —-v-----i
200 300 400 !500 600 700 800 900

Setup W
Measurement: OUAL-CH, SPECTRUM AVERAGING 800 lines
Trigger: GENEBATOR

Ave~aging: LIN 10

;:;:. SPAN 800HZ df: ll+z n 1s dt: 977us

.
● Centre Freq: 500Hz

Weighting: RECTANGULAR
ch.A: 200mV DIRECT /-0.7Hz -\oN 22.5mV/unit
ch,B: Iv - DIRECT /-0.7Hz --\oN 10.2mV/m/52
Generato~: MULTISINE O.OdB ref i.00V &TE DC: C)U’$’
Repetition: CONT

F’”,
ii”
9’
.,-”

Sample # pre- Stmin Elastic Modulus E b’s Factor q

strain Amplitude MPa
x lo< x 10*

iOOHz 200 300 400 100 200 300 400

Hz Hz Hz Hz Hz Hz Hz

1 60 10 36.9 43.1 45.8 48.0 0.447 0-476 0.519 0.531

80 40 39.3 465 49.2 51.6 0.462 0.474 0s02 0519
100 60 39.3 46.5 48.8 51,3 0.449 0.473 0.499 0.517

2 80 10 37.1 44.8 49.2 52.3 0.436 0.457 0.469 0.47s

100 30 43.7 48.6 55.7 58.7 13.439 0.462 0,454 0,466
100 60 395 52.1 55.5 58.1 0.431 0.425 0.432 0.442

3 80 10 52.0 64.1 69.4 72.7 0.409 ., 0.416 0.423 0.434

100 20 52.6 62.1 68.2 74.0 0.442 0.418 0.42 I 0.434
120 40 53-2 62.3 6S.7 74.5 0.402 0.419 0-426 0.43s

4 60 10 42.3 50.6 55,9 60.7 o.4i4 0.414 0.409 0.406 ~

100 20 45,2 5s.1 60.2 65.1 0.427 0.440 0.402 0.398 ~
ilo 30 43.1 55.0 60-3 65.6 0.499 0.409 0.387 0.382
Y.LI. G n-.-:. .-AI... L... T.AJ .-

frequenciesand mode ahapcs. The inner disk is a sofid disk of

Fscquency ElasticModLI1ua
E Ma
tadii 119.38 rnmand outsrdisk isananmtlardisk of inside radius
Hz ( MPa) F-or
of 122.17 mm and the outside mdius of 1524 mm The space in
betweenthe disks is filled with the viscoetzsricmatcrid under
100 43.7 coasidemtion. The disk is suspendal by an elastic cord and is
0.438
200 52.6 0.440 impacrai by a hammer in hotizmntal direction. T?ISnatutad
300 57.2 I 0,445 frequencyfm 2 nodaldiameterOnodalcircle modewas observal
at 1110 Hz Ntuctthe finite elementsolution was obtained with
400 I
I Table 6. Average value 61.1 I 0.454
of dynamicmodulusfiotn Tripod test
modulus of aluminum as 73 GP~ Poisson’s ration of O-33, and
dens@ of 2700 IC#tT? and.Poisson’sratioof viacoelasticnaacerial
as 0.45, and density of 103?.1 K@. The modulus E of
Now a s&uoture made of two ahurtinum disk are joinssl by visooelastic material was varied as shown in Table 7 until the
vkc@astic - (SSshown in Figure 5) andis td for natural
natuml frequency of 2 nodal d~nodal oirck modeconcspond
totheexperimentvalueof lllOHz

Ekstic hhduhls E Frsquency and 2 nodal

(MPz) diameter maic
(w
10 919.1
20 995.1
40 1092.4
45 1109.4
I
Table 7. Elastic modulus of viscoelastic material used in FEM

CONCLUS1ON

[n the Oberst rnerhod, the specimens MCin cantilever configuration

resulting in non-uniform stmitt. Thus any suain dc~en~
phenomenon cannot be obsetvcd in the OberSCmethod. HOWCV*,
Fig. S. Aluminum disks bonded by viscods~c tttSCeri8i comparing average values, the methods do not agmc with ach
orher. Also the tripod method gives data continuously over a mnge

.s@”d 9T17ElKLs13 .9N3 .H331d t5:6El 366T-ST-130

as opposed [O the obcmt method which gives data u only a few
frequencies Shce a viacoeltic materiid has diffaent charwm.sits
over different foquencc range (@May, glass tubk rransitlo~
rubbay and flow region) any interpolation or extmpolation of data
may bc questionable. Ptcscmly a DMA tsst is being catried out to
determine is characteristh such as transition tmpmture along with
modulus aa a function of tampemmre and frequency, The only
natural frequency of the composite aluminum disk that could be
excitd fall outside the mnge of mossurd properties. A newample
is being prepared to have natural fkequancy widdn the mngc for
which pro~s are known. Ovemll, the tripod method seams the
prcfemble method though it involves a IOKmore instrummtation than
the Oberst method and also only V* small pre-suain or SUain-
amplinide oould be applied,

ACKNOWLEDGEMENTS

Partof this work was suppmed by the Summer Faculty Rt%mrch

Participation Program of &onne National &boratory.
Appreciation is also exrended to Mr. David Larson of S & V
Solutions for asaistanoc with tic Oberst experiment.

REFERENCES

[1] Nashi~ A.D., Jones, DJ.G.and HenderaomJ.P., VIMon

Damping, John WiIcy and Sons, 1985.

[2] ASl%i Iks.igmrion E 756-93, Standard Test Method fir

Mesuring Vibration-Damping PropaR&s of Mataii%
Annual Book of ASTM Standards,vol. 04.06, pp 827-833.
1993.

[3] Complex MOdUIUSAppiUZUUS

TYPG3930,Bxucland Kjacr, pp
I-u 1964.

[4] Nic~ L.F., Wismor, N,J. and Gad% S., Improved Method
for Gmnplex Modulus Eatimtim Bruel & Kjaet Application
Note, pp 1-7, 1997.

[s] ANSYS version5.4, ANSYS Inc, PA,1998.

[6] STAR MODAL version 5.23, Spectral dynami% IIIC.,CA.

1996.

9t3”d 9T P0ZSLS~3 “CIN3 “H33U 2S :60 366 T-ST-130