Sei sulla pagina 1di 7

MODAL ANALYSIS OF PLATES WITH PARTIAL CONSTRAINED-LAYER DAMPING TREATMENTS

Karl K. Stevens, Professor Department of Mechanical Engineering Florida Atlantic University Boca Raton, Florida USA

Rajendra A. Bhat, Graduate Student Department of Ocean Engineering Florida Atlantic University Boca Raton, Florida USA

ABSTRACT Application of modal analysis techniques in the determination of the modal parameters and mode shapes of an edge-fixed rectangular plate with a contstrained-layer viscoelastic damping treatment extending over a portion of the surface is desThe test specimen and test procedures are cribed. discussed, and experimentally-determined values of the system natural frequencies and loss factors for varying degrees of damping treatment are pre; sented and compared with predicted values. Information also is presented on the influence of the damping treatment on the flexural mode shapes of the plate. NOMENCLATURE - natural fn i -+iQ - quality T - kinetic At
D us

frequency factor energy

(in

Hz)

factor per cycle strain energy

decay time energy dissipated maximum stored logarithmic

decremeti

of critical damping 5 - fraction system loss factor 17 w circular fnequency w system natural circular frequency n INTRODUCTION Many structures are susceptible of this vibration fatigue failures is accomplished treatments. Viscoelastic free-layer ing layer strained-layer is overlaid layer. In are bonded viscoelastic include plates and panels, which to flexural vibration. Control often is important in reducing and noise problems, and frequently by the use of viscoelastic damping

deformations induced by the flexural vibration of the underlying structure. The advantages and disadvantages of both types of damping treatment are well established and are discussed in the extensive literature which now exists on this subject. A corn rehensive review of this literature is given in(l, !i ) and a good discussion of the current stateof-the-art may be found in (3). Suffice it here to say that free-layer treatments are effective for thin plates and panels, while constrained-layer treatments are more effective for thicker members. Only constrained-layer treatments are considered in this paper. Cost-effective design of damping-layer treatments for plates and panels frequently requires that material costs and added weight be held to a minimum. This requirement can be met only by judicious tradeoffs between the amount of damping treatment used and the amount of damping achieved. It is known intuitively, and from experience, that the damping achieved by complete coverage of a plate with a viscoelastic damping treatment usually is not significantly greater than that achieved by a partial coverage extending over some lesser fraction of the surface area. However, the designer needs information that is more specific if the necessary tradeoffs are to be made intelligently. This information is most useful when presented in the form of plots, nomographs, or similar design aids. Generation of design aids for plates with constrained-layer damping treatments is made difficult by the number of variables involved and by the ranges of parameter values which must be considered. Because of the large number of different cases which must be investigated, use of finite element methods is not economically feasible. Accordin ly, an approximate analytical method was developed( B) for predicting the effectiveness of partial and complete constrained-layer damping treatments in controlling the lower flexural modes of vibration of rectangular plates of finite extent. This paper describes the use of modal analysis techniques to obtain experimental verification of the procedure. A series of experiments was conducted on an edgefixed rectangular plate with a constrained-layer viscoelastic damping treatment applied over a
201

damping materials are used either as a treatment consisting of a single dampapplied to the structure or in a contreatment wherein hhe(idampling layer with a stiffer elastic constraining both configurations, the various layers together and energy is dissipated in the material as a result of the cyclic

concentric rectangular region of varying size on one side of the member. Impulse testing techniques were used to determine the system natural frequencies, loss factors, and mode shapes for the first five flexural modes of the plate. The test specimen and test procedures are described in detail, and results are presented and compared with predicted values for two of the modes. The mode shape data obtained confirm the validity of the assumption that addition of the damping treatment does not alter significantly the flexural mode shapes of the plate. As this assumption is commonly used in the analysis of plates with damping layers, this result should be of particular interest. ANALYTICAL APPROACH The analytical approach is described in detail in (4). Here, we give only a brief summary as background for the discussion to follow. The analysis is based upon an energy for the system which, for sinusoidal with frequency 6.1, yields the result w* = Us + i(l/*s)UD T (11 balance oscillations

herein. This same assumption is commonly used in the analysis of constrained-layer damping treatments for beams and plates; e.g., see (839). TEST SPECIMEN The test specimen was a 1.59mm (1/16th in.) thick rectangular aluminum plate with overall dimensions of 0.46m (18 in.) by 0.51m (20 in.). When installed in the support frame, the plate had a working area of .30m (12 in.) by .34m (13.4 in.), which gives an aspect ratio of 1.11. This aspect ratio was chosen to provide reasonable separation between the resonant frequencies for the modes of interest, thereby minimizing problems with modal overlap in the data reduction. The working area of the plate was covered on one side with a constrained-layer damping treatment consisting of a 0.254mm (10 mil) thick layer of #468 Adhesive Transfer Tape manufactured by 3M Corporation and a 0.381mm (15 mil) thick constraining layer of dead-soft aluminum. The viscoelastic adhesive bonds the various layers together and provides the damping. The plate specimen and the mechanical properties of the damping layer as functions of frequency and temperature were provided by Anatrol Corp. , Cincinnati, Ohio. Space does not permit inclusion of these properties here; they are given in (4, loI. The specific weight of the damping material was 10.2kN/m3 (64.81b/ft3), and it was assumed to be incompressible. In separate tests, the elastic modulus and Poissons ratio of the plate and constrainpg layer were found to be 67.6GN/m2 (9.8 x 10 lb/in ) and 8.33, respectively; the specific weight was 26.3kN/m (1671b/ft3). A test frame used in a previous study of s uare ? 7) kas plates with free-layer damping treatments modified to accommodate the rectangular plate specimen. The frame was made of 25.4mm (lin.) thick by 76.2mm (3in.) wide steel bar stock and consisted of two parts - an upper and lower half. Each part was fabricated from four lengthsof bar machined smooth on their mating edges and bolted together to form a rigid rectangular unit. The mating surfaces of the two halves of the frame were machined flat to provide good surface contact with the plate. Some of the rounded edges of the rolled bar stock were not completely removed in this process, however, and did contribute to a very slight variation of the boundary along the extreme edges of the plate. Clamping of the specimen was accomplished by sandwiching the plate between the upper and lower halves of the frame and bolting the two halves together with sixteen bolts around the periphery of the plate. To provide uniform clamping, all bolts were tightened to the same torque using a torque wrench. The plate specimen exhibited some initial curvature so the test specimen actually was a slightly curved plate constrained to a flat edge condition by the steel frame. A photograph of the test frame, with is shown in Figure 1. specimen installed, TEST PROCEDURE Impulse testing techniques (11-14) and a HewlettPackard 5423A Structural Dynamics Analyzer were used to determine the mode shapes and modal parameters for the first five flexural modes of the plate specimen. The plate was excited by impact2u2

In this expression, is the maximum system kinetic energy, US is the maximum stored strain per cycle. energy, and UD is the energy dissipated This result is the counterpart of the Rayleigh quotient for elastic systems. Since the Rayleigh quotient has a stationary valve in the neighborhood of the syssem eigenfunctions(5), accurate estimates for w can be obtained from Eq. (1) if reasonable estimates of the eigenfunctions are available. The natural frequency, wn, and loss factor, n, of the damped system are obtained from Eq. (1) using the relations wn = -+1 1 11
rl(wn) = D -

rum11

(2)

and

**us

The frequency dependence of the loss factor stems from the frequency dependence of the mechanical properties of the viscoelastic damping material. If desired, the system damping can also be expressed in terms of other damping parameters, such as the logarithmic decrement, A, the fraction of critical damping, 5, the amplification at resonance, Q, or the decay time At. These parameters are related to the loss factor via the expressions (4) where f, is the natural frequency in Hz.

It was assumed that addition of the damping treatment does not alter significantly the mode shapes of the undamped plate and that the latter can be used as an approximation to the flexural mode shapes of the system. The validity of this assumption has been demonstrate f layer damping treatments t6aqj pt~,e~a:i~~a?efound to be valid for the plate specimen with constrained layer treatment used in the experiments described

ing it with an impulse hammer with an attached PCB Piezotronics Model 3338 force transducer to sense the force imparted to the plate. The plate response was determined using a PCB 303A miniaturized accelerometer. Signals from both the force transducer and accelerometer were amplified by PCB 48ODO6 amplifiers and fed into the two channels of the HP 5423A analyzer to determine the system transfer function, from which the modal parameters are determined. A photograph of the overall test set-up is shown in Figure 2. In order to define the point of excitation and position of the accelerometer, a 6x6 mesh of equal-sized elements was drawn on the side of the plate without damping treatment (Figure lb). The twenty five mesh nodal points not located on the plate boundary were used as the points of excitation and accelerometer locations during the experiments. For identification, the mesh points were numbered as indicated in Figure 3. The x and y-coordinates of these nodal points were used in a coordinate table to define the plate configuration in the analyzer. A thin strip of damping treatment was removed from all four edges of the fully-coated plate to provide a narrow margin between the damping treatment and test frame. The purpose of this margin was to prevent the edges of the damping treatment from rubbing against the frame during vibration, thereby distorting the measured values of system damping. This resulted in a damping treatment covering 97.3% of the surface of the plate. The plate was then excited at a selected point by the impulse hammer and the response at a different point monitered by the accelerometer. For each pair of excitation and response points, data from fifteen repetitions of the impact were averaged to determine the transfer function. To obtain the mode shapes, the tests were repeated with the accelerometer located at each of the mesh nodal points not on the plate boundary, with the point of excitation kept the same. Cross-correlation of the data was checked by repeating the experiment with different points of excitation and the same accelerometer location. A coherence of 0.99, or better, was required for data acceptance. A previous study (15) revealed that, of the several procedures available on the analyzer for determining the modal parameters, the X-band procedure is the more reliable. Accordingly, this method was used in the present investigation. In this procedure, the modal parameters are determined by the analyzer from a curve fit to the transfer function over a frequency interval spanning the resonant frequency for the mode of interest. To increase the reliability of the data, experiments for determining the frequencies and damping factors were repeated four times each and experiments to determine the mode shapes were repeated twice. This was done for each of the five modes investigated. The second and third modes of the plate specimen were found to be relatively close, with their resonant frequencies approximately 15 Hz apart. Adjacent modes are accounted for in the curve-fit routines of the analyzer, However, the interaction
203

between these two modes was minimized in the tests for the modal parameters by locating the point of impact or response measurement on an anti-nodal line for the second mode and on a nodal line for the third mode, or vise versa. After all testing was completed for the fullytreated plate, a strip of damping treatment 25.4mm (1 in.) wide was removed from around the outer edge of the plate, leaving a concentric rectangular partial damping treatment centered at the middle of the plate, as illustrated in Figure la. Modal parameters and mode shapes were then determined for the partially-treated plate. The process was repeated until a bare plate condition was reached. The percentages of the plate surface covered by damping treatment during the tests were 97% (full coverage), 71%, 47%, 27%, 13%, 4% and 0% (bare plate). Temperatures in the laboratory ranged from 22C (7ZF)to26C (78F) during the course of the experiments, with most tests conducted at 24C (75Y). Potential problems arising from the different coefficients of thermal expansion of the steel frame and aluminum plate were alleviated by releasing the plate and reclamping it to the same bolt torques before each set of experiments. Careful investigation revealed that release and reclamping of the plate had negligible effect on reproducability of the data. Measurement tables were set up on the analyzer for each of the five modes of vibration with each of the different degrees of damping treatment. A typical table is shown in Figure 4. These tables list the natural frequency, percent critical damping, test temperature, impact and accelerometer location, and degree of damping treatment. The Point No. listings refer to points on the plate grid in Figure 3 and the residues of the transfer function are proportional to the vibrational amplitude at these points. The shape listings are normalized modal displacements, which define the mode shapes. MODE SHAPES Figures 5 and 6 give overall views of the experimentally-determined mode shapes for the second and fourth flexural modes of the plate with complete (97%) damping treatment coverage. Figures 7-10 show these same mode shapes along traverses in the x and y-directions through an anti-nodal point for three different degrees of damping treatment: full treatment (97%), 47% coverage and bare plate. For comparison purposes, the theoretical mode shapes also are shown. In these figures, the numbers listed in the captions correspond to plate grid points (Figure 3) included in the traverse. These data, which are representative of the results obtained for the other three modes investigated, indicate clearly that the theoretical mode shapes for the undamped plate are reasonable approximations to the mode shapes of the damped member, as assumed in the analysis. The zero slope condition associated with a clamped edge is not evident in Figures 5-10, but could be made more apparent by taking more data points in the vicinity of the plate boundary.

MODAL PARAMETERS Figures 11-14 and measured loss factors modes of the tween theory for the other not shown. give a comparison of the predicted values of natural frequencies and for the first and second flexural Agreement beplate test specimen. and experiment was at least as good three modes, for which results are

Given the difficulties involved in obtaining identical conditions between theory and experiment and the scatter normally encountered in damping data, the agreement between theory and experiment in the present case is considered excellent. CONCLUDING REMARKS It has been demonstrated that impulse testing and modal analysis techniques, when used with proper attention to detail, can provide values of system damping with sufficient accuracy and consistency to evaluate analytical damping models. The mode shape information obtained confirmed the assumption that addition of the damping treatment had little effect on the flexural mode shapes of the plate. While this result is for the particular plate specimen and modes of vibration investigated, it is believed to hold for other constrained-layer configurations for while the flexural rigidity of the constraining layer is not significantly larger than that of the underlying structure, which usually is the case. Some variation in damping values obtained from different sets of measurement points was observed. This variation often exceeded the expected scatter band, and seemed to indicate some path dependence As might be expected, this of the damping values. effect was more noticeable for the smaller perTime did centages of damping treatment coverage. not permit a more detailed investigation of this point.

The properties of the damping material are rather dependent upon frequency and temperature in the ranges involved in the experiments. For each mode, the values of these properties used in the computations were determined at the natural frequency of the undamped plate. Since the natural frequency did not vary significantly with the percentage of damping treatment, the properties of the damping material were assumed constant for each mode of vibration. Predicted results based on the damping material properties at three different temperatures over the range encountered in the experiments are presented. The resulting set of three curves for temperatures of 72F, 75F and 78F gives some indication of the scatter that could be expected in the experimental values due to temperature variations encountered during the experiments. The loss factors for the bare plate ranged from about 0.005 to 0.010 for the modes considered. These values include radiation losses and the inherent damping in the plate and within the supporting frame. Since the loss factor for the bare plate is small, it can be added directly to the loss factor resulting from the dam ing treatment to obtain the total system damping (67. Accordingly, the loss factor measured for the bare plate was added to the predicted values in Figures 12 and 14 in order to obtain a more realistic comparison between theory and experiment. It is difficult to achieve a fully clamped edge condition in experiments, and this investigation proved to be no exception. It is evident from Figures 11 and 13 that the measured natural frequencies of the bare plate were consistently below the theoretical values, which is indicative of flexibility in the supports. To compensate for this effect, and to obtain a more realistic comparison between theory and experiment, the predicted results were adjusted, as follows. In the the maximum stored energy, US, was computations, reduced by an amount AU to account for the lack of complete edge fixity. ? he quantity AUS was chosen to bring the predicted and measured natural frequencies for the bare plate into agreement, and was assumed to be constant for all percentages of damping treatment. This adjustment yields the results indicated in Figures 11-14, and brought the predicted and measured frequencies and loss factors into substantial agreement for all five The fact that this adjustment modes of vibration. yields good agreement for all percentages of damping treatment in strong evidence that the initial differences between the theoretical and experimental results were associated with test system flexibility and not with some aspect of the damping treatment. 204

REFERENCES 1. Nakra, elastic Digest, Control with ViscoB.C., Vibration Materials, The --- Shock and Vibration Vol. 8, No. 6, pp. 3-12, 1976.

2.

3.

Nelson, F.C., Techniques for the Design of Highly Damped Structures, The Shock and --Vibration Digest, Vol. 9, No. 7, pp. 3-11, 1977. Rogers, L., ed., Conference on Aerospace Polymeric Viscoelastic Damping Technology for the 1980s, Air Force Flight Dynamics Laboratory,rt No. AFFDC-TM-78-78-FBA, July, 1978. Hsu, H.Y., Vibration Analysis of Rectangular Plates with Complete and Partial ConstrainedLayer Damping Treatments, M.S. Thesis, Department of Ocean Engineering, Florida Atlantic University, 1983. Meirovitch, C., Analytical tions, The MacMillan Co., Methods Nm,T967. in Vibra-

4.

5.

6.

Stevens, K.K., Kung, C.H. and Dunn, S.E., "Damping of Plates by Partial Viscoelastic Coatings-Part I-Analysis", Proceedings NoiseCon. 81, Raleigh, N.C., pp. 445-448, June, 1981. Dunn, S.E., Kung, C.H., Jaising, V. and Stevens, K.K., "Damping of Plates by Partial Viscoelastic Coatings-Part II-Experimental," Proceedings Noise-Con. 81, Raleigh, N.C., pp. 449-452, June, 1981. Kerwin, E.M. and McQuillan, R.J., "Plate Damping by a Constrained Viscoelastic Layer: Partial Coverage and Boundary Effects", Bolt Beranek and Newman Report No. 760, 1960.

7.

8.

9.

Johnson, C.D. and Kienholz, D.A., "Finite Element Prediction of Damping in Structures with Constrained Viscoelastic Layers", AIAA J., Vol. 20, No. 9, pp. 1284-1290, 1982. "Control of Noise and Vibration 10. Nashif, A.D., with Damping Materials", Sound and Vibration, -Vol. 17, No. 7, pp. 28-36, 1983. 11. "Dynamic Testing of Mechanical Systems Using Impulse Testing Techniques", Hewlett Packard Application Note 140-3, 1972. Ramsey, K.A., "Effective Measurements for Structural Dynamics Testing, Part l", Sound Vol. 9, pp. 24-35, 1975. -and Vibration, "Effective Measurements for Ramsey, K.A., Structural Dynamics Testing, Part 2", Sound Vol. 10, pp. 18-31, 1976. -and Vibration, Halvorsen, W.G. and Brown, D.L., "Impulse Technique for Structural Frequency Response Testing", Sound and Vibration, Vol. 11, pp. -8-21, 1977. of Damping Layer Jaising, V.R., "Analysis Treatments for Plates using Experimentally Determined Mode Shapes", M.S. Thesis, Department of Ocean Engineering, Florida Atlantic University, 1981.

Figure

2.

Experimental

Set-Up.

12.

13.

14.

15.

FIGURES

Figure

3.

Plate

Measurement

Grid.

Figure 1. Plate Damping Treatment

Specimen with (a) Partial and (b) Measurement Grid.


205

FIRST MODE (46.57% Frequency: Damping: Point 9 10 11 12 13 16 17 10 19 20 23 24 25 26 27 30 31 32 33 34 37 38 39 40 41 No. 124.597 3.290% Hz.

COVERAGE) Accelerometer Temperature: Shape


I

at 25 74F

Residue 1.403 2.881 3.463 2.896 1.606 2.843 5.655 6.653 6.167 3.423 3.373 6.545 8.851 6.775 3.871 2.690 5.779 6.437 6.160 3.231 1.212 2.605 2.937 2.607 1.305 K K K K K K K K K K K K K K K K K K K K K K K K K

374.034 768.292 923.414 772.134 428.162 758.037 1.5079 1.7740 1.6442 912.685 099.407 1.7451 2.3600 1 JO66 1 SO320 7 17.234 1.5410 1.7164 1.6426 861.482 323.291 694.694 783.150 695.127 348.096

m m m m m m

m m

Figure

6.

Plate

Mode Shape

(4th

Mode).

97.3% CovEffACE -Doflsction/Cantar

46.6% DANE COVCRAGE _-. - -PIATE ___-.

THEORETICAL

Deflection

m m m m m m

Point

of impact

was

moved

to different

locations.

Figure

4.

D a

Measurement

Table. Figure 7.

Dlatanca SECOND

MaoE X-DIRECTION

from

platr

sdpa

In Inohae (8.9.10.11.12.l3.14)

Second Mode-X

Traverse.

COVEFADE --

97.3x

CO%&E --.

- -P&!!$ __ __.

THEORETICAL

Dcflsction/Daflection I.5

at 4.50

inch.

1 .s 0 -3 -I

Figure

5.

Plate

Mode Shape

(2nd Mode).

I
-1.5 L 0.M) 2.25 4.50 Distance from SECOND MDDE. 6.75 plats ed9* Y-DIRECTLON 9.00 in inches (4.11.18.25.32.39.46) 11.25 13.375

Figure

8.

Second Mode-Y Traverse.

206

97.3x CovERA0E -Deflection/Deflection 1.5

46.6X COVERAOE --.


at 4.0 inch.

THEORETICAL PEGi - _ - - __ _

I Adjusted , 9.67 . :

*..9
..95
..a4 ..93 6.99 ..6, 9.99 6 1. 96 3. 4. si 9; 9; A

PER CENT DAMPING TREATMENT COVERAGE


2.00 4.00 6.00 6.00 in Inchew (36.37.30.39.40.41.42) 10.00 12.00 Distance from pIat* edga FOURTH MODE, X-DIRECTION

Figure

12.

Loss Factor

(1st Mode).

Figure

9.
97.3% COERAOE --

Fourth

Mode-X -Traverse.
THEORETICAL

C&%?OE ---

P%i ____--_

Deflsction/Dafkction 1.5

ot 9.0

inch.

I 9

*I

69

4;

5;

9:

9;

PER CENT DAMPING TREATMENT COVERAGE Figure


2.25 4.50 Dlstoncs ram FOUR, MODE. 6.75 plate edge Y-DIRECTION

13.

Natural

Frequency

(2nd Mode).

9.00
in lnchsa (5.12.19.26.33.40.47)

11.25

13.375

Figure
9.9 ,I. 199 nr 19.

10.

Fourth

Mode-Y Traverse.

9.14

..,I

. .

-! . ..I 1

I I I

I I

I I . i d

..

19.
14. 139 II. Lt. 1.9 . 69 3. 4. 69 66 79 9. 9, I.. 9 1, 6, 3. 49 9. 66 79 96

9.

1.9

PER CENT DAMPING TREATMENT COVERAGE Figure 14. Loss Factor (2nd Mode).

PER CENT DAMPING TREATMENT COVERAGE Figure 11. Natural Frequency (1st Mode).

207

Potrebbero piacerti anche