You are on page 1of 12

Journal of Sound and Vibration (1978) 57(4), 471-482

AN ANALYSIS

OF BINARY

FLUTTER

OF BRIDGE

DECK

SECTIONS

Y. NAKAMURA
Research Institute for Applied Mechanics, Kyushu University, Fukuoka, Japan (Received 20 September 1977)

This paper is concerned with an analytical and experimental study of binary flutter of bridge deck sections. A set of analytical formulas giving the frequency and rate of growth of oscillation, the position of the equivalent center of rotation and the phase difference between bending and torsion near the critical flutter point is presented. The formulas provide an analytical basis for the previously proposed method of classification of binary flutter of bluff structures. The results of wind tunnel experiments on models with simple geometrical shapes confirm that the present formulas are applicable to a variety of structures ranging from a flat plate to much more bluff bridge deck sections.

1. INTRODUCTION

Flutter is a flow-induced, self-excited oscillatory instability of elastic structures. Since the spectacular collapse of the Tacoma Narrows Bridge in 1940, flutter of long-span suspension bridges in wind has been a subject of serious engineering concern, and prediction of the critical flutter speed is currently one of the most important design procedures for modern long-span suspension bridges. Most bridge deck sections are not streamlined so that the flow around them is necessarily separated. Unfortunately, however, there is no satisfactory aerodynamic theory for any separated flow, particularly an unsteady one, and hence no complete theory of flutter of bluff structures is now available. It is also true, on the other hand, that most bridge deck sections are not very bluff so that they still behave as efficient lifting surfaces. Accordingly, there is some resemblance between aerofoil and bridge deck flutter, and a number of authors have dealt with the two phenomena as similar, if not actually identical. In the measurement of oscillatory aerodynamic forces and moments of bridge deck sections [l], Scanlan and Tomko, while benefiting from the aerofoil study as much as possible, yet showed its distinct limitations as regards applicability to bridge deck sections. In a previous paper [2], it was shown that flutter of bluff structures in many coupled degrees of freedom can be classified into three types depending upon the mode of energy transfer from fluid to structure; the three types of flutter mentioned are herein referred to as the classical type flutter, the single degree of freedom type, and the intermediate type, respectively. The classification could shed new light on the underlying physical mechanisms of flutter of bluff structures in many degrees of freedom that had often remained unclarified in the past investigations. In the case of binary flutter, in particular, a more practical method of classification was also proposed and the experiments could confirm its usefulness. This method of classification was based on the observation of the rate of growth of oscillation rather than that of the energy transfer from fluid to structure, on an intuitively correct assumption that the former would be proportional to the latter. 471 0 1978 Academic Press Inc. (London) Limited 00222460X/78/0422-0471 $02.00/O

472

Y. NAKAMURA

In the present paper, the analytical counterpart of the above-mentioned method of classification of binary flutter-that is, to predict the rate of growth of oscillation with use of given aerodynamic coefficients-is considered. The aerodynamic coefficients required for the analytical approach may be obtained experimentally either by a free oscillation technique [l, 31 or with a more sophisticated forcing apparatus [4, 51. First, a discussion of the aeroelastic characteristics of a bridge deck binary system is presented, and following from this a set of useful analytical formulas is derived for the solution to the flutter equation: namely, for the frequency and rate of growth of oscillation, the position of the equivalent center of rotation and the phase difference between bending and torsion. The formulas obtained are remarkably simple so that they are capable of clear physical interpretation. In particular, the formulas can be rewritten in forms suitable for the previously proposed classification of binary flutter. The results of some wind tunnel experiments are also presented to confirm the applicability of the formulas to all the three types of flutter mentioned above; in short, they are applicable to flutter of a variety of structures ranging from a flat plate to much more bluff bridge deck sections. It is thus shown that the present formulas are very useful not only for such practical applications as trend studies in the design of long-span suspension bridges, but also for academic purposes of obtaining clear insight into the complicated physical nature of flutter of bluff structures. It is well known that most bluff structures are susceptible to vortex excitation at wind speeds where the frequency of the von Karman vortex street approaches one of the natural frequencies of the structure. However, structures of the type herein considered are not, in general, very bluff, as was mentioned earlier. Their aeroelastic responses are then characterized by violent flutter instabilities occurring at relatively high wind speeds, which are sometimes preceded by very weak vortex excitation at much lower wind speeds. In this paper, attention is focused on the flutter instabilities at higher wind speeds with vortex excitation being left outside of consideration.

2. EQUATIONS OF MOTION Consider a two-dimensional section of a bridge deck which is endowed with two degrees of freedom, y and 0, and balanced mechanically about its midchord rotation axis. The model is suggested in Figure 1. For this model, the linearized equations of motion are

mji + m(2nfy)2y = L[y] + L[B],

ze + z(2nf)2 8 = M[y] + M [e],

(La

feare the still-air frequencies in the y and 0 degrees of freedom, respectively, L[ ] and M[ ]
are the aerodynamic lifts and moments about the midchord, respectively, in which [ ] means that the lifts and moments are functionals of the arguments, and the dot denotes the differentiation with respect to the time 1. For simplicity, structural damping is neglected in the present analysis.

where m and I are the mass and mass moment of inertia per unit span, respectively,& and

Figure 1. Bridge deck section in a uniform flow.

8 -9%
IY
V

BINARY

FLUTTER

OF BRIDGE

DECK

SECTIONS

473

The aerodynamic lifts and moments are respectively expressed in terms of the aerodynamic coefficients as

UYI + Jwl = +PW2b) {CJYPI + CJm


A4bl

(3) (4)

+ M PI = 4P?2bY {chf[Y/b1 + CM [W,

where p is the air density, V is the horizontal wind speed, 26 is the deck width, and C,[ ] and C,[ ] are the aerodynamic lift and moment coefficients, respectively. The equations of motion are then non-dimensionalized for convenience: II+ R2 ~9 q = ( r2/47r2 P) {C,[V] + CJOI], 6 + 02 0 = (P/27? v){C&] + C,[Q]}, (5)

y(t)/b

where T = 2Tft is the reduced time, in which f is the frequency of oscillation in wind, q(T) = is the reduced vertical displacement, p = m/(pb2) is the reduced mass, v = Z/(pb4)is the reduced mass moment of inertia, R =fv/fe is the uncoupled frequency ratio, 0 = fe/f is the frequency ratio, P= V/(fb) is the reduced wind velocity, and the dash denotes the differentiation with respect to T. The solution of equations (5) and (6) is assumed in a form of exponentially modified oscillations :

q(T) =
with

rlo exp

O(T)=Ooexp[(&+i)T+$],

(7,8)

lo/e0 = X

(9)

where q. and O. are the amplitudes, b is the logarithmic rate of growth of oscillation, &J is the phase angle between q(T) and O(T), i = (-1)1 2, and, as was pointed out in reference [2], X represents the position of the equivalent center of rotation on an assumption of 141< 1. It is further assumed that the aerodynamic lifts and moments for exponentially modified oscillations may be replaced by those which correspond to purely sinusoidal oscillations. That this assumption is correct at least for ]/II @ 1 was shown by Scanlan and Tomko [I] and this evidence is supplemented by the results of the experiment reported in what follows. For example, CJ?] is given by c,[v] = (CL,, + i&i) V(T), (10)

where CL,, and CL,, are the real and imaginary parts of the frequency response function, both of which are functions of the reduced wind velocity l? Introduction of equations (7) to (9) into equations (5) and (6) yields the following four equations, when the exponential functions are eliminated from both sides of the equations and then the real and imaginary parts are separated : [@ /4x2) - 1 + R2 g2] X = ( P2/47r2p) (CL,, X + CL,,
cos t$ -

CL,, sin b),

(11) (12) (13) (14)

(b2/47r2)- 1 + c2 = ( V2/27r2v) {(CM,, cos 4 + CM,, sin 4) X + CMeR}, j?X = ( Y2/4np) (CL,, X + CUR sin 4 + CL,, cos 4), B = ( P2/271v) K-C,,, sin 4 + CM,, cos 4) X + CMBI}.

Equations (11) to (14) can be solved to give rr, A ,#Iand 4, but such a solution would no doubt be complicated as too many parameters would then be involved. The derivation of simple

474

Y. NAKAMURA

formulas for c, X, /I and $J in which only the most essential parameters are retained was the aim of the work reported here. This is effected by observing some of the characteristics of the aerodynamic coefficients on the one hand and imposing some conditions on the solution on the other. 3. ANALYSIS OF THE AERODYNAMIC COEFFICIENTS AND THE SOLUTION First of all, the range of the reduced wind velocity of interest is well above that of vortex excitation; say, P> 10, approximately, is assumed. For this range of the reduced wind velocity, C,,, is much smaller in magnitude than CLBR for typical bridge deck sections as well as for a flat plate [4, 51. Thus IC,,,/ 6 ICLsRI. (15)

Second, the analysis of torsional flutter of bluff structures [6] indicates that major portions of C,[8] and C&I] are the contributions of the angle-of-attack motion, those of the angular velocity motion being much less. That is, the following relations are obtained to the first order of approximation :
(CL,, + GJ tl = (CL,, + iCLOd (24 PI 4, + iCd(W PI v . (16)

(CM,, + iCM,d II = (Go,

(17)

Or, alternatively, C lI = (27rV) CLtW C hfVR = - (27t/I? ChftW Chf,, = (270) C&ft?R, C,,, = - WY) C&V, (18)

According to the Theodorsen unsteady aerofoil theory, the four relations in equation (18) hold for a flat plate at high reduced wind velocities. It should be mentioned, however, that for bluff structural sections, all these relations may be valid over a much wider range of the reduced wind velocity, although the degree of the approximation and the lowest reduced wind velocity for which they are valid depend on the bluffness of a section. Regarding the solution, the stability of the torsional branch is considered in the subsequent analysis, whereas the bending branch is assumed to be stable : that is, IX/< 1, (19)

which states that the equivalent center of rotation should not be very far away from the midchord. In addition, the phase angle between bending and torsion is assumed to be small in magnitude: that is,

I91< 1.

(20)

Experience has shown (for example, see reference [2]) that these two conditions are satisfied for most bridge deck sections for which R @ 1. A further assumption to be imposed on the solution is

IBI 4 1.

(21)

Thus, oscillations near the critical flutter point, either decaying or growing, are herein considered. 4. THE USEFUL APPROXIMATE FORMULAS A considerable simplification of equations (11) to (14) is now feasible in the light of the analysis of the preceding section. Equation (12) is first taken up. It turns out that the first term on the left-hand side and those associated with the brace of the right-hand side are

BINARY FLUTTER OF BRIDGE DECK SECTIONS

415

relatively small and can be discarded. Equation (11) is then considered, in which the first term in the brace of the left-hand side is discarded whereas on the right-hand side only the second term in the brace is retained with cos 4 replaced by unity; similar analysis is appl.ied to equations (14) and (13). The simplified expressions for (T, X, /I and 4 are thus obtairred in the following form : d = 1 + (P/2$ v) C&&9& ( 22)
X = { az/47r~(-1 +

Rz a')} CLoR,

(23)
(24) (2.5)

P = (P2/2nv) (CM,, + CM,, X), /IX = ( P2/47rp) (C,,, X + CLe, + C,,, sin 4).

It is important to mention that equations (22) to (25) are presented in the order of their successive determination. That is, when the value of reduced wind velocity is prescribed, the frequency ratio ~7 is first determined by specifying the value of CMeR, which is a function of l? The evaluation of the position of the equivalent center of rotation X follows by specifying the value of C,,, together with r~just determined: similar procedure is applied for /3 and 4. Substitution of the first two relations of equation (18) into equations (23) and (24) leads to the following alternative expressions for X and /!:
CL,,,, X = {P3/8n3 ~(-1 + R2a')} (26)

J = (P2/27rv){C&@, + (27c/P)C,,,

X}.

(27)

For practical applications, specification of the phase angle is not always required. Then, use of equations (22), (26) and (27) is more advantageous with the aerodynamic coefficients involved being reduced to only three in number. Because of the simplicity of the present formulas, several useful conclusions may be obtained. First, consider equation (22) for the frequency ratio (T. Since the value of C,W,, for most bridge deck sections is positive, 0 > 1 holds for P > 0; in other words, as the wind speed is increased from zero, the frequency of oscillation decreases monotonically from,&. Second, equation (26) indicates that under the condition of R < 1 -1 +R202<0,

(28)

as the wind speed is increased from zero. This, combined with C,,, < 0, leads to X > 0: that is, the equivalent center of rotation of a bridge deck section moves toward the leading edge with an increase in l? Third and lastly, by multiplying equation (25) by X and adding it to equation (24), one obtains B = P2(C &, + CL,,, sin 4 (X/2)}/27r(v + pX2), where C MB1 = Chf,, + CM,, X + (C,,, + C,,, X) X/2. (30) Equation (29) is important because a classification of binary flutter is possible on this basis. Let the aerodynamic moment coefficient for the sinusoidal torsional oscillation corresponding to X be C,&,, + iC,& and consider that ChB, is expressed in terms of the aerodynamic coefficients corresponding to the midchord. After a little manipulation, one finds that C,&,, is identical with equation (30). One notes also that v + pX2 in equation (29) represents the reduced mass moment of inertia with respect to X. Then, writing (29)

476

Y.

NAKAMURA

where /3i = P2 C~Bi/Zrc(V + /LP), fi2 = P2 CL,, sin 4(X/2)/27r(v + pXz), (32933) one easily finds that the rate of growth of oscillation, /?, consists of two physically different terms, one representing the rate of growth of oscillation for the system in a single degree of freedosn in torsion with respect to X, and the other being that proportional to sin 4, where C#J is the phase angle between bending and torsion. The classification of binary flutter is now feasible as (1) the classical type flutter, if p2 9 fir ; (2) the single degree of freedom type flutter, if /?i B pz; (3) the intermediate type flutter, if fil 2: Bz. Thi:s method of classification of binary flutter is identical with what was proposed earlier [2], but analytical expressions have now been formulated, as equations (31) to (33), for the first time. One finds, by going back to equation (27), that the stability of a bridge deck section is directly controlled by the pitch damping (-CM,,) regardless of the type of flutter. That this is characteristic of bridge deck flutter will be mentioned later. The assumptions made about the aerodynamic coefficients and the solution in deriving the present formulas are by no means seriously restrictive, and as will be exemplified by the experiment described in section 5, they may be satisfied for typical bridge deck sections as well as for a flat plate provided these have mass and mass moment of inertia of sufficiently large values. In short, the present formulas are applicable to flutter of a variety of structures ranging from a flat plate to much more bluff bridge deck sections. It may be added that equation (24), when it is reduced to p = 0, is basically similar to the formula of reference [7] which had been obtained on the basis of the Theodorsen unsteady aerofoil theory. Further, consider application of the Theodorsen unsteady aerofoil theory to equation (27) for the case where the high frequency approximation is made. Here, the high frequency approximation implies that in the circulation function, which is C(k) = F(k) + iG(k), where k = 2~17, F(k) = 3 and G(k) = 0 are assumed. The aerodynamic coefficients for a flat plate in this approximation are, with the virtual mass terms neglected, CL@R = - n, Chf,, = nr/4 and C,,,,, = - 7r2/41? (34)

Inserting equation (34) into equations (22), (23) and (27), and letting j = 0 in equation (27), one obtains VF= d27r(l - R2)l{(l/P) + (1/4v)], (35)

for the critical flutter speed. By replacing the arithmetic mean, [,u- + (4~)~ l/2, in equation (35) by the geometrical mean, (~,LLv)- /~, equation (35) is reduced to (36) It is interesting that equation (36), if (27~)~= 2.506 . . ., is replaced by a slightly different constant of 2.623, is identical with Selberg s empirical formula [8] which is known to be exceptionally accurate for a flat plate.
5. EXPERIMENTAL SET UP AND TECHNIQUE

FF = d27r(l - R2) 6.

The experiments were conducted in a low-speed wind tunnel with a 3 m high by 0.7 m wide working section. The basic model which was used consisted of a thin flat plate, 6 mm in thickness, 0.66 m in span, with a uniform chord of O-4m, and with stiffening trusses. It had a large thin plate mounted at each end to ensure two-dimensional air flow near the ends. The model is illustrated in Figure 2.

BINARY FLUTTER OF BRIDGE DECK SECTIONS

411

Aluminium brackets attached to the end plates were elastically hinged with cruciform supporting frames by the use of flexure springs. The entire system was suspended by four equal coil springs which were clamped at one end on the horizontal rod of each supporting frame. Horizontal restraining wires of sufficient length were also attached to the top and bottom of the vertical rod of each supporting frame. Thus, the model was allowed to give both vertical and torsional motions with downstream and spanwise displacements prevented. The vertical displacement was detected with a contactless optical displacement follower, whereas the torsional one was measured by the use of strain-gauges cemented on the surfaces of the flexure springs. In addition to the basic model, three simple deck section models were chosen in the present experiment: that is, a flat plate with six equal-spacing stringers, and that with stringers and a center barrier, both of a height to deck width ratio of U.075, and an H-section with a girder to deck width ratio of 0.09. Any of these three models was easily obtained by attaching segments of light polystyrene plate with negligible weights to the basic flat plate model. The free oscillation experiment was divided into two parts. The first part of the experiment was concerned with the model experiment : i.e., (1) both the vertical and torsional responses 1
/--End plote

Figure 2. Flat plate model with end plates. All dimensions are in millimeters.

occurring simultaneously; (2) the torsional response occurring with the vertical response prohibited at X. This resulted in curves of b and fil together with X versus J? The second part of the experiment was concerned with the following responses : (1) the torsional response with the vertical response prohibited at the midchord; (2) the vertical response with the torsional response prohibited. The measurements of the frequency and growth rate of oscillation yielded curves of CWBR, CM,, and CL4, versus v which are required for equations (22), (26) and (27). In the first experiment, the still-air frequencies were f, and fe= 1.60 Hz and 3.20 Hz, respectively, from which R = 0.50 was obtained. In addition, p and v = 150 and 100, respectively, whereas the mechanical dampings, either vertical or torsional, were of an order of 0.03. The experimental wind speed ranged from 5 m/s to 18 m/s, approximately. The system used in the second part of the experiment was identical with that used in the first part of the experiment except for the case of the H-section model where the lighter system was necessary for the precise measurements of small frequency variations at low wind speeds. In most cases, oscillations at small amplitudes were found to be of exp [;.t] type where 1 is a constant, so that the measurements of b were done at 8, = 3 degrees, approximately. All the experiments were made with the model at zero mean incidence : i.e., with the deck surface parallel to the incident flow. Change in the mean incidence during the tunnel runs occurred due to the asymmetry of the flow caused by the presence of the stiffening trusses, but it was small and neglected; the maximum change encountered was about 0.5 degree for the Aat plate model.

478
6. EXPERIMENTAL

Y. NAKAMURA

RESULTS

AND

DISCUSSIONS

Figures 3(a) to 3(c) show the results of the binary flutter experiment for the flat plate, the flat plate with stringers and the H-section where the logarithmic rates of growth of oscillation /3 and /I1 together with the position of the equivalent center of rotation X are plotted as functions of the reduced wind velocity I?
(0)
0.5

X
a

2.0

n 0.25 a F 0 : 0 0
.--_::

I.0 P

-0.25

25
I I
I I

30 v

35
I

40
I

(b)
0.5

Cc)
I i

-0.25

A
n
I
1

.
I 0

2.0

0.

P,

1.0

I5

20

25

30

IO

15

20

25

Figure 3 .I osition of equivalent center of rotation X and logarithmic rates of growth B and jll us. reduced wind veloc:itx y. X is measured from midchord (positive towards leading edge); n, experimental; heavy line, theoretical: b represents the logarithmic rate of growth for binary system; 0, experimental; heavy line, theoretical: /31 represents the logarithmic rate of growth for torsional system with respect to X; 0, experimental. (a) Flat plate; (b) flat plate with stringers; (c) H-section.

According to Figure 3(a), the instability of the flat plate obviously belongs to the classical type flutter. The instability of the flat plate with stringers in Figure 3(b) still belongs to the classical type flutter, but the presence of the stringers results in an increase in PI and a decrease in the critical flutter speed. The H-section results in Figure 3(c) show that the critical flutter speed is further decreased, and because /? N j?r over the entire velocity range tested, flutter is now of the single degree of freedom type. However, there is also an indication that flutter at still higher wind speeds would tend to be of the classical type.

BINARY FLUTTER OF BRIDGE DECK SECTIONS

479

The results of measurement of the aerodynamic coefficients for all the four models are presented in Figures 4(a) to 4(c). Figure 4(a) indicates that all the data for CL,, roughly collapse on a single curve except for a sharp drop exhibited by the H-section results at low wind speeds. The sharp drop suggests the proximity to the critical wind speed for vortex excitation which is P= 7.0, approximately. Figures 4(b) and 4(c) show rather spectacular variations of both CM,, and CM,, with Pfor the H-section, which appear to be characteristic of structures with sufficient bluffness [6] ; again, the effects of vortex shedding at around P = 7.0 are evident.

I
v r
plate; c , flat plate

Figure 4. Experimental aerodynamic coefficients vs. reduced wind velocity. ?,? Flat with stringers; X, flat plate with stringers and a center barrier; a, H-section.

7. EXPERIMENTAL

CHECK

OF THE VALIDITY FORMULAS

OF THE APPROXIMATE

The theoretical estimates for j? and A , which were obtained by applying equations (22), (26) and (27), are presented as heavy lines in Figures 3(a) to 3(c); the theoretical estimates for /? include allowance for the experimental mechanical damping of O-03. The agreement between theory and experiment is reasonably good through all the three figures for both /I and X except for the case of /3 in Figure 3(b) where there is a slight discrepancy due to some unknown reasons. It is worth pointing out that the present formulas are valid not only for the classical type flutter (Figure 3(a)), but also for the single degree of freedom type (Figure

480

Y. NAKAMURA

3(c)). In other words, the validity of the formulas does not depend on the type of flutter. Note also that they are valid for any of the three oscillations (decaying, steady state and growing); in particular, this supports the validity of the aforementioned assumption that the steady state oscillation aerodynamic coefficients may be replaced by those corresponding to decaying or growing oscillation, and vice versa. It was shown [9] that a center barrier is sometimes effective in augmenting the aeroelastic stability of a bridge deck section. The present investigation provides a good example for this favourable effect of a center barrier. For example, compare the aerodynamic coefficients

2.0

I.0

4 0

I 20

25

30

35

Figure 5. Position of equivalent center of rotation and logarithmic rates of growth vs. reduced wind velocity for a flat plate with stringers and a center barrier. See Figure 4 for key to symbols.

between the flat plate with stringers and that with stringers and a center barrier, which are shown in Figures 4(a) to 4(c). It is obvious that the attachment of a center barrier yields a remarkable increase in pitch damping, whereas the rest of the aerodynamic coefficients indicate little change. It follows from equation (27) that this should result in a considerable increase in the critical flutter speed. That this prediction is correct is shown in Figure 5, where, again, the agreement between theory and experiment is reasonably good.

8. SOME

REMARKS

ON THE

CHARACTERISTICS

OF BRIDGE

DECK

FLUTTER

It may be worth mentioning the following points that characterize bridge deck flutter in comparison with the aircraft counterpart. First, the pitch damping can directly control the stability of a bridge deck section even for the case of the classical type flutter. In sharp contrast to this, the in-phase aerodynamic coefficients have dominant contributions in most aircraft flutter [lo, 111. In other words, the critical flutter speed for an aircraft wing is often determined, though approximately, by applying the so-called frequency coalescence criterion which states that the frequencies of the bending and torsional branches become coalesced as the critical flutter speed is approached. Because of the presence of CM,, in equation (27), however, the phenomenon of frequency coalescence is not found in bridge deck flutter.

BINARY FLUTTER OF BRIDGE DECK SECTIONS

481

According to Zimmerman [lo], the frequency coalescence criterion can be applied to a short-time explosive flutter where the variation of the phase angle becomes steep near the critical flutter point. The classical type flutter of bridge deck sections, however, appears to be rather mild, as exemplified by Figure 3(a) and also because the phase variation can be assumed small. Another point of interest is the weakness of the couplings of equations (22) to (25) in bridge deck flutter.,Again, this is not the case for aircraft flutter; one then finds that the equations for r~and X, and also those for b and 4, are both strongly coupled. A close examination of the equations of motion reveals that all the above-mentioned characteristics of bridge deck flutter may be attributed to the absence of the inertial coupling. A numerical experiment for a flat plate with an uncoupled frequency ratio of 0.5 was designed

1
X6/b = 0. I

I.0

p.2

-1

$=

0.75

T;;;
04
0.2 / 0.5 I 0.75 1 I.0 1.25

0
0.3 0.3

0.5 0

X,/b is measured from midchord;

Figure 6. Effect of the position of center of gravity X,/b on the natural mode frequencies for a flat plate. numerical experiment is based on Theodorsen unsteady aerofoil theory.

to demonstrate the effect of the inertial coupling on the variations of the natural mode frequencies with wind speed, where the position of the center of gravity, X,, measured from the midchord, is varied for 0 < X,/t, < 0.3. The analysis was based on the Theodorsen unsteady airfoil theory and followed the method of Goland and Luke [12], where the aeroelastic modes of exp [At] type, in which 1 is a complex number, were assumed, and the Wagner lift deficiency function was utilized to compute the unsteady aerodynamic forces corresponding to the exp [At] type motion. The results are shown in Figure 6, which indicates that as X,/b increases from zero the frequencies of the two branches become closer at around the critical flutter speed V,, so that the applicability of the frequency coalescence criterion becomes restored. 9. CONCLUDING REMARKS A set of analytical formulas applied to binary flutter of bridge deck sections has been presented in this paper. The frequency and rate of growth of oscillation together with the position of the equivalent center of rotation near the critical flutter point are obtained by solving three simple equations where the aerodynamic coefficients involved are only three in number. The rate of growth of oscillation consists of two terms of physically different origins; one represents the rate of growth of oscillation corresponding to a system in a single degree of freedom in torsion about the equivalent center of rotation, and the other is that propor-

482

Y. NAKAMURA

tional to sin 4, where 4 is the phase angle between bending and torsion. A comparison of which of these two is predominant provides an analytical basis for the previously proposed method of classification of binary flutter of bluff structures. The simplicity of the formulas may be attributed to the absence of the inertial coupling in bridge deck flutter. The assumptions on which the present formulas are based are not seriously restrictive, and the formulas are expected to be applicable to a variety of structures ranging from a flat plate to much more bluff bridge deck sections.

ACKNOWLEDGMENT The author is indebted to Messrs K. Watanabe ducting the experimental work. and T. Yoshimura for their help in con-

REFERENCES 1. R. H. SCANLAN and J. J. TOMKO 1971 Journal of the Engineering Mechanics Division, Proceedings of the American Society of Civil Engineers 97, EM6, 1717-1737. Airfoil and bridge deck flutter derivatives. 2. Y. NAKAMURA and T. YOSHIMURA 1976 Journal of the Engineering Mechanics Division, Proceedings of the American Society of Civil Engineers 102, EM4,685-700. Binary flutter of suspension bridge deck sections. 3. N. SHIRAISHIand K. OGAWA 1975 Proceedings of the Japan Society of Civil Engineers 244, 23-35. An investigation on aeroelastic responses of suspension bridges due to the non-linear aerodynamic forces (in Japanese). 4. N. UKEGUCHI and H. SAKATA 1965 Journal of the Japan Society for Aeronautical and Space Sciences 13,27-36. An investigation of aeroelastic instability of suspension bridges (in Japanese). 5. H. TANAKA and M. ITO 1969 Transactions of the Japan Society of Civil Engineers 1,209-226. The characteristics of the aerodynamic forces in self-excited oscillations of bluff structures. 6. T. YOSHIMURAand Y. NAKAMURA 1977 Proceedings of the Japan Society of Civil Engineers 264, 33-40. A study of indicial moment responses of bridge deck sections with special reference to torsional flutter (in Japanese). 7. H. SAKATA and Y. WATANABE 1976 (to be published). An approximate calculation for bendingtorsion flutter. 8. A. G. FRANDSEN 1966 Proceedings

of Symposium on Suspension Bridges, Lisbon, Portugal, Paper No. 43, 609-627. Wind stability of suspension bridges-application of the theory of thin

airfoils. 9. A. HEDEFINE and L. G. SILANO1971 Journalof the Structural Division, Proceedingsof the American Society of Civil Engineers 97, ST1 1, 2653-2678. Newport bridge superstructure. 10. N. H. ZIMMERMANand J. T. WEISSENBURGER 1964 Journal of Aircraft, American Institute of Aeronautics and Astronautics 1, 190-202. Prediction of flutter onset speed based on flight testing at subcritical speeds. 11. G. T. S. DONE 1969 Aeronautical Research Council, Reports and Memoranda No. 3554. A study of binary flutter roots using a method of system synthesis. 12. M. GOLAND and Y. L. LUKE 1949 Journal of the Aeronautical Sciences, American Institute of Aeronautics 16,389-396. A study of the bending-torsion aeroelastic modes for aircraft wings.

You might also like