Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Journal of Wind Engineering

and Industrial Aerodynamics 74—76 (1998) 913—921

Active damping and flutter control of cable-stayed bridges


Y. Achkire, F. Bossens, A. Preumont*
Active Structures Laboratory, Universite& Libre de Bruxelles,
CP165, Av. F.D. Roosevelt 50, 1050 Brussels, Belgium

Abstract

This paper presents a strategy for active damping of structures with guyed cables connected
to active tendons; the control algorithm has guaranteed stability, including at the parametric
resonance. The theoretical results are confirmed by experiments. The strategy is applied to the
flutter control of cable-stayed bridges. ( 1998 Elsevier Science Ltd. All rights reserved.

Keywords: Flutter; Active damping; Cable-stayed bridges; Tendon control

1. Introduction

Because of their increasing span, cable-stayed bridges have become increasingly


sensitive to flutter instability as well as to wind and traffic-induced vibrations. The
problem is difficult, because of the highly nonlinear behaviour of cables with sag, and
the coupling between the cables and the bridge deck. Several cases of parametrically
excited cable vibrations have been recorded on actual bridges.
Several strategies have been proposed for the active tendon control of the global
modes of the bridge, as well as for the in-plane and out-of-plane cable vibrations
[1—4]; all of them use noncollocated actuator/sensor configurations. These strategies
use different control algorithms for the various vibration modes, and they have been
found to be prone to instability when the interaction between the cable and the
structure is large.
An alternative strategy has been proposed by the present authors [5,6], which is
based on a displacement actuator (active tendon) collocated with a force sensor; this
approach does not rely on a model of the system and enjoys guaranteed stability
properties (assuming perfect actuator and sensor dynamics). The approach is reviewed

* Corresponding author. E-mail: apreumon@ulb.ac.be.

0167-6105/98/$19.00 ( 1998 Elsevier Science Ltd. All rights reserved.


PII: S 0 1 6 7 - 6 1 0 5 ( 9 8 ) 0 0 0 8 3 - X
914 Y. Achkire et al./J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 913–921

in the next section; it is subsequently extended to the decentralized damping of a cable


structure involving several active cables (e.g. guyed tower) and its potential to increase
the critical flutter velocity of cable-stayed bridges is demonstrated.

2. Active damping of cable structures

2.1. Control strategy

It is widely accepted that the active damping of linear structures is much simplified
if one uses collocated actuator—sensor pairs [7]; for nonlinear systems, this configura-
tion is still quite attractive, because there exist control laws that are guaranteed to
remove energy from the structure. The direct velocity feedback is an example of such
“energy absorbing” control. When using a displacement actuator (active tendon) and
a force sensor, the (positive) integral force feedback

P
u"g ¹ dt (1)

(refer to Fig. 1a for notations) also belongs to this class, because the power flow from
the control system is ¼"!¹uR "!g¹2. This control law was already applied to
the active damping of a truss structure in Ref. [8], and it is quite remarkable that it
also applies to nonlinear structures; all the states that are controllable and observable
are asymptotically stable for any value of g (infinite gain margin).

2.2. Experiment

The foregoing theoretical results have been confirmed experimentally with a labor-
atory scale cable structure similar to that represented schematically in Fig. 1a, where
the active tendon consisted of a piezoelectric actuator [5]. Fig. 1b shows the experi-
mental frequency response between a force applied to the structure and its acceler-
ation; also shown in the figure is the free response of the structure with and without
control. We see that the control system brings a substantial amount of damping to the
system, without destabilizing the cable; this behaviour is maintained at the parametric
resonance, when the natural frequency of the structure is twice that of the cable.

2.3. Decentralized control

The foregoing approach can readily be extended to the decentralized control of


a structure with several active cables; each active tendon works for itself, with a local
feedback following Eq. (1). Fig. 2 shows experimental results obtained with the test
structure of Fig. 6 [9]. It consists of a T shaped structure with appropriate bending
and torsion stiffness to simulate the dominant behaviour of a bridge. The relative
values of the torsion and the bending modes can be adjusted by changing the masses
of the system. Two active cables of 0.5 mm diameter act on the structure as indicated
Y. Achkire et al./J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 913–921 915

Fig. 1. Active damping of cable structures.

Fig. 2. Experimental frequency response functions between the shaker and one of the accelerometers for
the structure with two active cables of Fig. 6, with and without control.

on the figure; the active tendons consist of piezoelectric actuators collocated with
force sensors. The actuator displacement is amplified by a mechanism which allows
a tip displacement d of 100 lm. Once again, a substantial damping is added to the
structure, without threatening the stability of the cables.
At this point, it is important to point out that the concept of active tendon control
of cable structures would not require that all the cables be active. On the contrary, the
916 Y. Achkire et al./J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 913–921

control system would involve only a small set of cables judiciously selected. Next
section describes an approximate linear theory to predict the performance of the
control system and provides design guidelines to select the active cables.

2.4. Approximate linear theory

If we assume that the dynamics of the active cables can be neglected and that their
interaction with the structure is restricted to the tension in the cables, the governing
equation for the structure is
Mx( #Kx"!B¹, (2)
where M and K refer to the structure without the active cables (but with all the passive
ones), ¹ is the vector of the tension in the guy cables and B is the influence matrix
relating the local coordinate systems of the active cables to the global coordinates. If
we neglect the active cable dynamics, their tension is given by
¹"K (BTx!d), (3)
#
where K "diag(h ) is the diagonal matrix containing the stiffnesses of the active
# #
cables, BTx are the relative displacements of the extremities of the cables projected on
the chord lines and d contains the active displacements of the tendons. The fact that
the same matrix B appears in Eqs. (2) and (3) is a consequence of the displacement
actuator (d) and the force sensor (¹) being collocated. Combining Eqs. (2) and (3) and
the decentralized control law with the same gain for every channel
g
d" K~1¹, (4)
s #
we obtain the closed-loop equation (in Laplace variable)

C D
g
Ms2#(K#BK BT)! BK BT x"0. (5)
# s#g #
From this equation, it is readily observed that as gPR, the dynamics of the
closed-loop system converges to that of the structure where the active cables have
been removed.
If we assume that the mode shapes are not changed substantially by the control
system and if we denote by X the natural frequencies of the cable structure with the
i
active cables and by u those when the active cables have been removed, it can be
i
shown [6] that the closed-loop poles follow, for each mode, a root locus defined by
s(s2#X2)#g(s2#u2)"0. (6)
i i
Its general shape is represented in Fig. 3. Using a perturbation method, it can be
shown that the maximum achievable damping, obtained for g*"X , is
i
X !u
m.!9" i i. (7)
i 2u
i
Y. Achkire et al./J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 913–921 917

Fig. 3. Root locus of mode i when varying the gain g from 0 to infinity (only the upper half of the symmetric
diagram is shown).

Thus, the maximum modal damping is controlled by the relative spacing of the
natural frequencies with and without the active cables. This simple and meaningful
result provides a rationale for selecting the number, the size and the location of the
active cables in the early stage of the design process.
We would like to stress that although based on the assumption that the dynamics of
the active cables can be neglected, the foregoing approximate model is likely to be
fairly accurate, because only a few cables would normally be active in a cable-stayed
bridge. A comparison of the approximate model with the fully non-linear model of
a guyed truss was conducted in Ref. [6]; the agreement was found excellent.

3. Flutter control

On several occasions, suspended bridges have suffered damages, or even complete


destruction under wind excitation. The most famous example is the Takoma Narrows
bridge, which collapsed under torsion flutter at a wind speed of 42 mph, on 7 Novem-
ber 1940. Large bridges are more sensitive to flutter which, in most cases, is associated
with the aeroelastic damping coefficient in torsion becoming negative above a critical
velocity [10]. Aside from a careful aerodynamic design aiming at achieving proper
flutter derivatives, the traditional way of increasing the flutter velocity consists of
increasing the torsional stiffness. An alternative approach which may be promising for
future very large bridges consists of applying active control techniques; these can be
based on the use of active tendons, or of gyrostabilizers. In this study, we restrict our
discussion to the active tendon control; this approach was already considered analyti-
cally by Yang [11,12] with a noncollocated sensor placed on the bridge deck. As
mentioned earlier, noncollocated actuator/sensor pairs are prone to spillover instabil-
ity, while the proposed control law is guaranteed to extract energy from the system.

3.1. Unsteady aerodynamic forces

In order to extend the theory of control of cable structures to aeroelastic systems,


we need to obtain a Laplace representation of the unsteady aerodynamic forces; this is
918 Y. Achkire et al./J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 913–921

possible because the aerodynamic lift and moment are related to the downwash
velocity by convolution integrals. Referring to Fig. 4, the lift and moment about the
elastic axis can be expressed, in each cross-section, by [13]

CP A B P D
1 q h@@ q
¸(q)" oº24bp a@# / (q!q ) dq # a@@/ (q!q ) dq , (8)
2 b 1 0 0 2 0 0
~= ~=

CP A B P D
1 q h@@ q
M(q)" oº28b2p a@# / (q!q ) dq # a@@/ (q!q ) dq , (9)
2 b 3 0 0 4 0 0
~= ~=
where a is the local angle of attack and h is the vertical displacement of the elastic axis;
and “and” denote the first and second derivatives with respect to the reduced variable
q"ºt/b; / , / , / and / are the heredity functions, relating the lift and moment to
1 2 3 4
the angle of attack and its time derivative. The desired relations can be obtained by
Laplace transform of the foregoing equations. Following Lin [14], we shall use the
more compact form

G H
1 h(s)
¸(s)" oº24bp U (s)#a(s)U (s) , (10)
2 b 1 2

G H
1 h(s)
M(s)" oº28b2p U (s)#a(s)U (s) , (11)
2 b 3 4

where a(s) and h(s) denote the Laplace transforms of the angle of attack and the
vertical displacement, and U (s) are combinations of the Laplace transforms of / (q).
i i
For an airfoil, the four heredity functions involved in Eqs. (8) and (9) can be expressed
in terms of a single function called the ¼agner function [15]. For a bridge, they
depend on the deck cross-section and must be determined from wind tunnel tests.
Unfortunately, the large amount of experimental data available in the literature is not
based on the foregoing representation, they are based on the definition of the flutter
derivatives given by Scanlan and co-workers [10] which mixes the time and frequency
domains. The derivation of the Laplace transform U (s) from the flutter derivatives of
i
Scanlan is possible, but not straightforward.

3.2. Flutter control

The proposed approach to the flutter control problem is summarized in Fig. 4 for
a simplified system involving a typical cross-section; for lack of information about the
heredity functions, we have assumed that they are identical to those of an airfoil.
Unlike K in Eq. (2), K and K in Fig. 4 include the contribution of the active cables.
h a
The control law is the same as in the previous section (decentralized integral force
feedback). Fig. 5 shows the evolution of the closed-loop poles with the control gain
g and the wind velocity º; the dotted lines represent the evolution of the closed-loop
poles as a function of the control gain g in absence of aerodynamic forces [Eq. (6)],
while the full lines show the evolution of the poles when the wind velocity increases;
the curves are drawn for g"0 and for some positive value of g. This simplified
Y. Achkire et al./J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 913–921 919

Fig. 4. Typical cross-section and proposed approach to flutter control.

Fig. 5. Evolution of the poles of the system with the wind velocity, for g"0 and for some value of control
gain (only upper half of the symmetric diagram is shown).

example only aims at illustrating the methodology which can easily be extended to
more complicated bridge geometry and more realistic representations of the unsteady
aerodynamics using the so-called “strip” theory and projecting the governing equa-
tions in modal coordinates.
920 Y. Achkire et al./J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 913–921

3.3. Experiment

In order to verify the foregoing flutter control strategy, we have used the mock-up
of Fig. 6. To avoid the use of a wind tunnel, the torsion flutter is simulated by
a separate control system based on two accelerometers sensing the rotational acceler-
ation and a shaker, as indicated in Fig. 6. A state feedback with observer is used to

Fig. 6. Decentralized control of structures with several cables. Simulated flutter experiment.

Fig. 7. Root locus of the torsion mode as a function of the flutter gain g* (experimental results).
Y. Achkire et al./J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 913–921 921

relocate the torsion poles in the unstable region [9]. In this way, a scalar “flutter gain”
g* is used to move the torsion poles from their open-loop position to the unstable one,
simulating the wind velocity. Fig. 7 shows the effect of varying the flutter gain g* from
0 to 1, with and without active tendon control: the first curve (without control) crosses
the imaginary axis for g*"0.25. When the active tendon control is added, the starting
point is brought further left in the complex plane and the second curve enters the
instability region for g*"1.35. This demonstrates that the active tendon control
increases substantially the stability margin.

4. Conclusion

A control strategy with guaranteed stability has been proposed for the decentra-
lized control of cable structures. The control system uses an active tendon collocated
with a force sensor. The efficiency of the active damping strategy has been demon-
strated and simple design criteria have been developed. The theoretical results have
been confirmed by experiments on small laboratory models using piezoelectric
actuators. The applicability of the methodology to flutter control has been
demonstrated.

References
[1] Y. Fujino, T. Susumpow, An experimental study on active control of planar cable vibration by axial
support motion, Earthquake Eng. Struct. Dyn. 23 (1994) 1283—1297.
[2] Y. Fujino, P. Warnitchai, B.M. Pacheco, Active stiffness control of cable vibration, ASME, J. Appl.
Mech. 60 (1993) 948—953.
[3] P. Warnitchai, Y. Fujino, B.M. Pacheco, R. Agret, An experimental study on active tendon control of
cable-stayed bridges, Earthquake Eng. Struct. Dyn. 22 (2) (1993) 93—111.
[4] J.C. Chen, Response of large space structures with stiffness control, AIAA J. Spacecraft 21 (5) (1984)
463—467.
[5] Y. Achkire, A. Preumont, Active tendon control of cable-stayed bridges, Earthquake Eng. Struct.
Dyn. 25 (6) (1996) 585—597.
[6] A. Preumont, Y. Achkire, Active damping of structures with guy cables, AIAA J. Guidance Control
Dyn. 20 (2) (1997) 320—326.
[7] A. Preumont, Vibration Control of Active Structures: An Introduction, Kluwer Academic Publishers,
Dordreche, 1997.
[8] A. Preumont, J.P. Dufour, Ch. Malekian, Active damping by a local force feedback with piezoelectric
actuators, AIAA J. Guidance 15 (2) (1992) 390—395.
[9] Y. Achkire, Active tendon control of cable-stayed bridges, Ph.D. Thesis, ULB, Belgium, May 1997.
[10] R.H. Scanlan, J.J. Tomko, Airfoil and bridge deck flutter derivatives, ASCE J. Eng. Mech. Div. 97
(EM6) (1971) 1717—1737.
[11] J.N. Yang, F. Giannopoulos, Active control and stability of cable-stayed bridge, ASCE J. Eng. Mech.
Div. 105 (1979) 677—694.
[12] J.N. Yang, F. Giannopoulos, Active control of two-cable-stayed bridge, ASCE J. Eng. Mech. Div. 105
(EM5) (1979) 795—809.
[13] R.L. Bisplinghoff, H. Ashley, Principles of Aeroelasticity, Dover, New York, 1975.
[14] C.G. Bucher, Y.K. Lin, Stochastic stability of bridges considering coupled modes, ASCE J. Eng. Mech.
Div. 114 (12) (1988) 2055—2071.
[15] Y.C. Fung, An Introduction to the theory of Aeroelasticity, Dover, New York, 1969.

You might also like