Вы находитесь на странице: 1из 15

Politecnico di Milano

Facolt di Ingegneria Civile, Ambientale e Territoriale


Corso di Laurea Magistrale in Ingegneria Civile

Dynamics of Structures
instructor: Prof. F. Perotti

Wind forces on slender structures

by
Federico Perotti and Luca Martinelli

06/06/2014
Introduction
In the following notes the engineering approach to the computation of structural re-
sponse to dynamic actions due to fluid-structure interaction will be treated, with special
reference to wind forces. Attention is focused on the effects of wind turbulence; the
concepts and criteria here introduced, however, can contribute to the understanding of
more complex fluid-structure interaction effects, the so called aeroelastic phenomena,
which can be harmful for very sensitive structural systems.
The topic here discussed is essentially the simplified modelling of the forces acting
on a slender body which is free of oscillating due to the effect on an unsteady fluid flow.

1. Fluid dynamic actions on slender bodies: the Morison approach


We pursue the definition of actions acting normally to the body axis, thus neglecting
friction effects on the lateral surface; in the spirit of the Morison approach the actions
are seen as sectional resultants (forces per unit length) i.e. the resultants of the pressure
and shear effects acting on the section contour due to the interaction with the flow.
We first assume, for simplicity, that the flow velocity is parallel to a symmetry axis of
the section (figure 1); this is the case of circular sections, when the velocity fluctuation
orthogonal to the average flow is neglected. Under these hypotheses the body is subject
to a distribution of forces g(F)(z,t), parallel to the fluid velocity, which can be expressed
as the sum of two terms;
g ( F ) ( z , t ) = g I ( z , t ) + gT ( z , t ) (1)
The first term in equation (1) represents an inertial, acceleration driven, contribution,
which is espressed, in the classical Morison formulation (see [1] for a complete treat-
ment), in the following way:
g I ( z , t ) = A ( z ) w t ( z , t ) + A ( z ) C A w t ( z , t ) A ( z ) C Au( z , t ) (2)
where is the fluid density, A(z) the section area, wt(z,t) the flow velocity, CA an ex-
perimental coefficient, typical of the section, and u(z,t) the function expressing the dis-
placement of the body axis.
The first term in (2) is the so called buoyancy effect, due to fluid acceleration, analo-
gous to the well known hydrostatic effect due to the acceleration of gravity (Archimedes
force). To describe the other terms the concept of added mass must be introduced; this
represents a portion of the fluid surrounding the section (having mass per unit lengh
equal to CA A(z)) which is assumed to share the same motion u(z,t) as the body. In this
light, the second and third term in equation (2) represent, respectively, the buoyancy and
the inertia force acting on the added mass; in the framework of structural analysis,
buoyancy effects simply act as external dynamic forces, while the latter term can be
simply added to the structural mass matrix. Expression (2) can be thus re-written, by
summing buoyancy effects, in the more usual form
g I ( z , t ) = A ( z ) CI w t ( z , t ) A ( z ) C Au ( z , t ) (3)
where the inertia coefficient CI =1+CA appears; for the circular section CA=1 is gener-
ally assumed.

1
It must be considered that the inertial term appearing in (1) is usually negligible, due
to air low density, in wind forces, while is to be introduced in expressing hydrodynamic
effects due, for example, to ocean waves. Wind forces, on the other hand, are dominated
by the second term, the so called drag, due to the effect of the relative velocity be-
tween the fluid and the structure; different formulations of the drag force can be found
in the scientific literature (see again [1]), among which the most suitable for wind
analysis purposes is the following:
1
gT ( z , t ) = dCD | wt ( z , t ) u ( z , t ) | ( wt ( z , t ) u ( z , t ) ) (4)
2

wt (z, t) u(z, t)
d

x
Figure 1. Symmetric section under 1D fluid flow

in which CD is an experimental coefficient depending, in general, on the section shape,


on the angle of attack of the fluid velocity and the on Reynolds number. The latter is
equal to the ratio:
Wd
Re = (5)

where W is the fluid average velocity, d a characteristic section dimension and the ki-
nematic viscosity of the fluid, i.e. the ratio of the viscosity to the density. By expressing
lengths in meters we get:
Re = 67000 Wd in air
Re = 1x106 Wd in water
It can be derived that the Reynolds number represents the ratio of the inertia to the
viscous forces.
The preceding derivation of fluid-structure interaction forces is consistent to the hy-
pothesis of the development of a so called quasi steady regime, in which:
1) the drag force is uniquely defined in terms of the instantaneous position and ve-
locity of the body;
2) it is possible to define the unsteady fluid-dynamic forces on a moving body on
the basis of experimental coefficients measured in stationary test on a fixed and
rigid model.

2
The quasi steady regime hypothesis can be invoked when the dynamic behaviour of
the body is characterized by oscillations at frequencies far from the ones of vortex shed-
ding, this being a typical phenomenon related to the aerodynamic behavior of bluff bod-
ies. We recall that in bluff bodies, differently from the case of streamlined profiles (such
as wings), significant phenomena of flow separation from the body surface tend to occur
in a wide range of flow situations.
The drag force (4) is non linear in the body velocity; the latter is also coupled, in (4),
to the fluid motion, thus resulting in an state dependent excitation term. However, un-
der some hypotheses, which are deemed to be acceptable in a wide range of engineering
situations, the drag force expression can be linearized; to this aim, the fluid velocity at
the elevation z can be decomposed into an average value W(z) and a fluctuating compo-
nent w(z,t), i.e.:
wt(z,t)=W(z)+w(z,t) (6)
Lets now assume that both the fluctuating component of the fluid velocity and the
body velocity are small when compared to the average value W(z); in such situation the
drag force can be expressed, by neglecting small terms, in the following approximate
form:
1
g t ( z ,t ) = C ( z ) | w( z ) + w( z ,t ) u ( z ,t ) | (w( z ) + w( z ,t ) u ( z ,t )) =
2
1
= C ( z ) W 2 ( z ) + C ( z ) W ( z ) w ( z , t ) C ( z ) W ( z ) u ( z , t ) +
2 (7)
( ) ( )
+ O w2 + O u 2 + O ( wu )
The assumption here introduced is well suited for the case of wind forces; on the oth-
er hand, fluid velocities due to seawave kinematics are usual modeled as zero-mean
processes, unless a wave-current field is considered. Thus, in offshore engineering
analysis different linearization strategies are adopted.
It can be observed that the terms appearing in the approximate form (7) can be re-
garded as a static force, depending on the square of the average velocity, a dynamic
force, proportional to the wind velocity fluctuation and a viscous damping term, de-
pending as well by the average velocity.
Coming to a more general case, though specifically referring to wind actions, we will
now pursue the expression of interaction forces acting onto the section of a slender
body, starting from the knowledge of the aerodynamic sectional properties and of the
projection on the section plane of the instantaneous values of the components of the
wind velocity. These will be expressed in the wind reference frame defined by the di-

rection of the average velocity vector W , this being rotated of the angle with respect
to the body axes (u,v); accordingly, the zero-mean functions w1(t) (alongwind) and w2(t)
(crosswind) will denote the fluctuation components parallel and orthogonal, respec-
tively, to the average velocity, while v1 and v2 define the components onto the same
axes of the velocity of the section centroid.
Under these premises we can decompose the total drag force into the instantaneous
drag and lift sectional components, i.e. the interaction forces (per unit lenght) acting
along the direction of the relative velocity vector and orthogonally to it, in the following
form:

3
w1
r

w2
W

-v1 Vr

-v2

v2
v

x2,
u
G
x1,
v1

Figure 2. Slender body section in a 2D unsteady flow

1
gT ( z , t ) = d ( z ) CD ( r ) Vr2 ( z , t ) (8a)
2

1
g P ( z, t ) = d ( z ) CL ( r ) Vr2 ( z , t ) (8b)
2
where d(z) is a characteristic dimension of the section, C D and C L are experimental drag
and lift coefficients, while the relative velocity amplitude Vr and the instantaneous angle
of attack r are defined as follows (see figure 2)
2 2
Vr ( z , t ) = (W + w1 v1 ) + ( w2 v2 ) (9)

w2 v2
r ( z ,t ) = tg 1 (10)
W + w1 v1
The preceding expressions, though simple and meaningful, are not convenient in the
context of structural analysis, since forces are defined in a time-dependent reference
frame; to overcome this problem the projection of the drag and lift force onto the above
defined wind reference frame ( x1 ,x2 ) are first considered, i.e.:
1
g1 ( z , t ) = d ( z ) Vr2 ( z, t ) CD ( r ) cos r CL ( r ) sin r (11a)
2

1
g2 ( z, t ) = d ( z ) Vr2 ( z, t ) CD ( r ) sin r + CL ( r ) cos r (11b)
2
These forces are the components of the resultant of the local pressure and shear
stresses acting on the section contour; the line of action of this resultant will not contain,
in general, the section centroid, to which applied forces are referred in structural analy-

4
sis. The aerodynamic moment is thus introduced, with the same criteria and defini-
tions, as:
1
( z, t ) = d 2 ( z ) Vr2 ( z, t ) CM ( r ) (12)
2
Within the framework of the QST (Quasi Steady Theory), the aerodynamic coeffi-
cients appearing in the above formulation are defined, as functions of the angle of attack
, on the basis of the aerodynamic forces which are measured on a rigid body in a test
conducted in steady and laminar conditions, i.e.:
gT ( )
CD ( ) = (13a)
1
d ( z ) w2
2

gP ( )
CL ( ) = (13b)
1
d ( z ) w2
2

( )
C M ( ) = (13c)
1 2
d ( z )w 2
2
The aerodynamic actions (11) and (12) are non-linearly dependent on the body veloc-
ity components (aerodynamic non-linearity), which, in addition, are coupled to the flow
velocity fluctuations in (9) and (10). They are also dependent on the body configuration
when, as it will be shown in the following, the section rotation will be taken into ac-
count.
In wind engineering applications such equations can be introduced in the non-linear
analysis of complex dynamic phenomena, such as galloping, which can be simulated
according to the QST.
In this section the so called buffeting effects, i.e. the dynamic effects of the velocity
fluctuations due to turbulence, are of primary interest; within this context, linearization
of the aerodynamic forces has been proven as an analysis tool leading to close approxi-
mation of the structural dynamic response.
As simplified, approach to the linearization, we first assume that the variation of the
aerodynamic coefficients with the angle of attack either can be neglected or, as it often
happens, cannot be fully represented due to lack of data; by introducing the trigonomet-
ric relationships
W + w1 v1 w2 v2
cos r = sin r =
Vr Vr
the expressions of the aerodynamic forces take the following form:
1
g1 ( z , t ) = d ( z ) Vr CD ( r )(W + w1 v1 ) CL ( r )( w2 v2 ) (14a)
2

1
g2 ( z, t ) = d ( z ) Vr CD ( r )( w2 v2 ) + CL ( r )(W + w1 v1 ) (14b)
2

5
By assuming now that both the turbulence components and the body velocities are
small compared to the average flow velocity, the following approximations hold:
2
2 w2 v2
(W + w1 v1 ) Vr = (W + w1 v1 ) 1+ 2
W + 2Ww1 2Wv1
W + w1 v1

(W2 v2 ) Vr ( w2 v2 )W


Upon substitution of the preceding expressions into (14a,b) we obtain, with simple
manipulation, that the alongwind and crosswind forces, now referred to the average ve-
locity direction, can be seen as the superposition of the following three terms: a static
one (subscript s), i.e. time-independent, a known dynamic one (d), i.e. time varying,
and a term linearly dependent on structural velocity (D), which can be seen as an
equivalent viscous contribution to the structural system damping.
g k ( z , t ) = g k , s ( z ) + g k , d ( z , t ) + g k , D ( z , t ) ; k = 1, 2

1
g1, s ( z ) = d ( z ) CD ( r ) W 2 ( z ) (15a)
2

1
g 2, s ( z ) = d ( z ) CL ( r ) W 2 ( z ) (15b)
2

1
g1,d ( z , t ) = d ( z ) W ( z ) CD ( r ) w1 ( z , t ) CL ( r ) w2 ( z , t ) (16a)
2

1
g 2,d ( z , t ) = d ( z ) W ( z ) CD ( r ) w2 ( z , t ) + CL ( r ) w1 ( z , t ) (16b)
2

1
g1, D ( z , t ) = d ( z ) W ( z ) CD ( r ) v1 ( z , t ) CL ( r ) v2 ( z , t ) (17a)
2

1
g 2, D ( z , t ) = d ( z ) W ( z ) CD ( r ) v2 ( z , t ) + CL ( r ) v1 ( z , t ) (17b)
2
Applying the same criteria, the aerodynamic moment can be written in the following
linear form:
1
( z, t ) = d 2 ( z ) CM W 2 ( z ) + d 2 ( z ) CM W ( z ) w1 ( z, t )
2 (18)
d 2 ( z ) CM W ( z ) v1 ( z , t ) = s ( z ) + d ( z , t ) + D ( z , t )
Note that in the most general situation, both the average velocity and the fluctuating
terms show components aligned to the slender body axis z; these are responsible for
friction forces which are normally disregarded. To compute normal components of wind
velocity, i.e. the ones introduced in the above equations, the following practical proce-
dure can be adopted:
1. given the section considered a plane is defined, containing the tangent to the
body axis, as computed at the section level, and the wind velocity;

6
2. the tangent component, i.e. the one parallel to the axis, is defined as the velocity
component in the plane defined in 1;
3. the normal component is obtained as the difference between the total velocity
and the component defined in 2.

Note that if the out-of-plane (flexural) rotation of the body is taken into account in the
procedure, a follower effect, i.e. a further dependence from configuration, is intro-
duced.
When the torsional rotation of the section is considered, the two following effects are
usually recognized.
Even though the field of wind velocity fluctuations is assumed to be homogene-
ous in the section neighborhood, the relative velocity between the fluid and the
body varies, along the section contour, due to the rotational component; this ef-
fect is significant for sections which, as often happens in bridges, are elongated
in the direction of the average wind velocity. It will be introduced here accord-
ing to the following simplified criterion (figure 3) in which the vertical velocity
is augmented by a term depending on a characteristic length R and on the angu-
lar velocity, i.e.

w2 v2 + R
tan r = (19)
W + w1 v1

The relative angle attack under which the flow sees the section, and thus af-
fecting the value of the aerodynamic coefficients, must be obtained as the differ-
ence of the wind angle of attack (19) and the section rotation, i.e., again with
reference to Figure 3:

r (t ) = r ( t ) ( t ) (20)
x2,

y, v
v2

v1
x 1,

x, u

G
R
W
Vr

w1 r
-v2
w2

-v1
Figure 3. Elongated aerodynamic section

7
Under this premises a Taylor series expansion of expressions (11), (12) and (19), (20)
can be performed in the neighborhood of = 0, r = 0 ; this criterion is consistent to the
usual structural hypothesis of small rotations and to the assumption, which is obvi-
ously retained, that the velocity components due to wind turbulence and body motion
are small when compared to the average wind velocity. In doing this we shall now take
the variation of the aerodynamic coefficients into account; the coefficients and their de-
rivatives as computed for the reference angle will be denoted with a prime symbol:
1
2
(
g1,d ( z , t ) = d ( z ) W ( z ) CD w1 + CD' CL w2

) (21a)

1 1
2
( 2
) ( )
g1, D ( z , t ) = d ( z ) W ( z ) CD v1 + CD' CL v2 R CD' CL 

(21b)

1
(
g1, K = d ( z ) W 2 ( z ) CD' CL ( t )
2
) (21c)

1
2
(
g 2, d ( z , t ) = d ( z ) W ( z ) CL w1 + CD + CL' w2

) (22a)

1 1
2
( 2
) ( )
g 2, D ( z , t ) = d ( z ) W ( z ) CL v1 + CD + CL' v2 R CD + CL' 

(22b)

1
(
g 2, K = d ( z ) W 2 ( z ) CL' + CD ( t )
2
) (22c)

1 1
( z , t ) = S ( z ) + d 2 ( z )W ( z ) CM w1 + CM' w2 d 2 ( z ) W ( z )CM'
2 2
1 1
d 2 ( z ) W ( z ) CM v1 + CM' v2 CM' R (23)
2 2
= S ( z ) + d ( z, t ) + K ( z , t ) + D ( z , t )
In expressions (21) through (23) the derivatives of the aerodynamic coefficient now
appear to take account, in linearized form, of the variation of the coefficients with the
angle of attack. In addition these equations show how the effect of the section rotation
on the relative angle of attack is responsible for the introduction of new terms (21c),
(22c) depending on the section rotation itself, which can thus be regarded as aerody-
namic stiffness effects.
Note that aerodynamic stiffness terms are usually disregarded in the analysis of build-
ing response to turbulence effects (the so called buffeting analysis), while aerody-
namic damping can be significant, especially when very low values of structural damp-
ing are estimated.
Both terms play a crucial role in the occurrence of aeroelastic phenomena; the
across-wind aerodynamic damping (22b), for example, is responsible for the stability
of linearized oscillations in galloping phenomena, while aerodynamic damping and
stiffness terms affecting the crosswind and rotational motion are essential in flutter
analysis. In the latter case, however, the quasi steady assumption is no longer valid; lin-

8
earization of the loads is still possible but the results are significantly affected by the
frequency of the body oscillations, so that experimental testing of rigid models is no
longer totally meaningful and full aeroelastic testing is necessary.

2. Discrete formulation of the linearized aerodynamic forces


The above derived forces represent a continuous distribution acting along the slender
body; in structural analysis of large systems a discretized version of applied forces must
be pursued, either by a simple lumping process or by more refined finite-element based
techniques. In addition, the above derived linearized aerodynamic forces are expressed
in the wind reference system, i.e. in terms of alongwind and crosswind components re-
ferred to the average velocity vector; the motion of the section is described as well in
the same reference frame. Both sets of components, forces and section displace-
ments/velocities must be transformed, for computational purposes, into the structural
reference frame, this being the one to which the structural model is referred. The same
consideration does not apply to the wind fluctuation components which are more con-
veniently expressed in the wind reference frame, this being the one in which they have
been obtained by measurement or by artificial generation from a given stochastic model.
In this light well denote with q (ju ) (t ) , q (jv ) (t ) the lagrangian coordinates of the struc-
tural model expressing nodal displacements parallel to u ( z ,t ) , v( z ,t ) and with and j (t )
the rotation of the j-th.
Under these premises, the following manipulation must be applied to the aerodynamic
forces and moments (21), (22) and (23).
Integration of forces per unit length over a suitable height centered on the j-th
node in order to obtain the floor resultants; in the following well represent this
operation by the simple multiplication by a suitable length lj, such as the in-
terstory height in the case of a bulding.
Projection of the nodal aerodynamic forces onto the structural model axes.
Transformation of the nodal velocity components onto the ones in the structural
reference frame, according to the relationships

v1 = q (ju ) cos + q (jv ) sin ; v2 = q (ju ) sin + q (jv ) cos


The aerodynamic resultant obtained by the above sketched procedure can be written,
for node/floor j, in the following form, listing again static (Qs), dynamic (Qd), aerody-
namic damping (QD) and aerodynamic stiffness terms (Qk)

Qs(,uj) C D c CL s
(v) 1 2
Qs, j = Qs , j = d jW j l j CD s + CL c ; c = cos , s = sin (24a)
2
Qs(, j) d j CM

Qd( u, j) QD( u, )j QK( u,)j


(v ) (v) (v)
Qd , j = Qd , j = F w ; QD, j = QD , j = c a , j q j ; QK , j = QK , j = k a q j (24b)
w

Qd(, j) QD(, )j QK(, )j

9
(v)
x2

v, qj

lj
x1

z

u, qj(u)
j Node j

Figure 4. Discretization of the aerodynamic forces

where:
q (ju ) (t )
w1 ( z j , t )
Wj = W (z j ) ; w= ; q = q (jv ) (t )
w2 ( z j , t ) j (t )

1
( CD c C L s ) 2 ( C L C D ) c + ( C D + C L ) s
' '


1
Fw = d jW j l j ( C D s + CL c ) ( C L C D ) s + ( C D + CL ) c
' '
(25)
2
1
d j CM d j CM
2

0 0 ( CL CD' ) c ( CD + CL' ) s
1
= d j l jW j 0 0 ( CL C D ) s + ( CD + C L ) c
2 ' '
k a, j (26)
2
0 0 d j CM'

2CD c 2 + ( CL + CD' ) cs + 2CL s 2 + ( CD CL' ) cs (


R C L CD' c + )

+ ( C D C L' ) s 2 ( C L CD' ) c 2 (
+ R C D + C L' s )

2CL c + ( CD CL ) cs + 2CD s + ( CL + CD ) cs + (
R CL C D' s + )
2 ' 2 '
1
ca , j = d j l jW j (27)
2 + ( C L C D' ) s 2 + ( C D + CL' ) c 2 (
R C D + CL' c )

d ( 2CM c CM s ) d ( 2CM s + CM' c )
'
d R CM

Note that the contribution (27) to the aerodynamic damping matrix is non symmetric.
The aerodynamic stiffness is non symmetric as well: in addition, the diagonal coeffi-
cient corresponding to the section rotation has the sign of the derivative of the aerody-

10
namic moment coefficient. If the latter is negative, a phenomenon known as torsional
divergence can occur in static conditions.
If aerodynamic stiffness can be neglected and aerodynamic damping can be intro-
duced with different criteria, as it will be shown later, the dynamic response to turbu-
lence effects can be carried on by using general purpose structural analysis packages. To
perform time domain integration, time histories of the wind velocity fluctuation must be
available at the floor levels to form external dynamic forces QD,j (24b). These time his-
tories can be artificially generated starting for a time-space stochastic model of the ve-
locity field: a model of this type is specified, for example, in the italian regulations CNR
DT-207 [5].

3. Modal analysis and aerodynamic damping factors


For a practical discussion of the nature and effect of aerodynamic damping, we shall as-
sume that modal superposition analysis is performed for computing the dynamic re-
sponse; this amounts to introducing the coordinate transformation
q ( t ) = y ( t ) (28)
in which the so called modal matrix
= [1 ... n ] (29)

lists, by columns, the eigenvectors on the undamped system. These are obtained by
solving the following algebraic eigenvalue problem:
k 2j m j = 0 (30)

As it is well known, the transformation (28) makes the equations of motion uncou-
pled, under the condition that damping is of proportional type. In standard analysis,
when dynamic forces independent of the system motion are applied, the generic j-th
equation of the transformed system takes the following form:

2
Tj Q(t )
y j + 2 j j y j + y j =
 j (29)
Mj

where the eigenvalue 2j appears, representing the square of the j-th natural circular
frequency of the system, in addition to the modal mass M j = Tj m j and to the modal
damping factor j , which is kept constant, in standard situations, for all normal
modes.
When standard aerodynamic forces are considered, Q(t) is replaced by the sum of the
dynamic components (24b,25), depending on the wind velocity fluctuations due to tur-
bulence, and of the aerodynamic stiffness and damping force components (24b,26,27).
In formulas:

2
Tj [Q(t , w1 , w2 ) + Q k (t ) + Q D (t ) ]
y j + 2 j j y j + y j =
 j (30)
Mj

11
In the following, lift and moment and aerodynamic stiffness will be neglected for
simplicity. The case will be considered, very frequent in applications, in which the os-
cillations along two directions, say u,v, are uncoupled, due to structural symmetry, in
absence of aerodynamic damping. There exist, in such cases, the following two sets of
eigenvectors
(u ) = 1( u ) ... m(u ) ; ( v ) = 1( v ) ... m( v ) (31)
Each set contains vectors (modal shapes) having non zero components along one of
the two directions only; the system configuration can be thus espressed in the uncoupled
form:
m m
q( ) = k k( ) yk( q( ) = k k( ) yk(
u) v)
u u
(t ) ;
v v
(t ) (32)
1 1

The following trigonometric factors are also defined:


1 1
1 ( ) = cos 2 + sin 2 ; 2 = cos 2 + sin 2 ;
2 2

1
12 ( ) = sin cos
2
On the basis of the above hypotheses and definitions we can express the generalized
component of aerodynamic damping (defined according to (24b) and (27)) with respect
to the normal coordinate yr( ) as:
u

m
r(u )T Q(Du ) = j r(,uj)QD( u,)j =
1
(33)
(u ) (v) (v)
m m m
CD d j l jW jr , j 1 k r , j y k + 12 k k , j y k =
(u ) (u )
1 1 1
m m m m
= CD1 j k d j l jW jr , j k , j y
( ) ( ) ( )
 k + CD12 d j l jW jr(, j)k( , j) y k(
u u u u v v)
j k
1 1 1 1

Expression (33), and the analogous for the v direction, can be re-written in the more
compact form
m m
r( ) Q(D ) = CD1 k (kr ) y k( ) + CD12 k (kr ) y k( )
u T u uu u uv v
(34a)
1 1

m m
r( ) Q(D ) = CD 2 k (kr ) y k( ) + CD12 k (kr ) y k( )
vT v vv v uv u
(34b)
1 1

where the following coefficients have been introduced:


m m
(kr ) = j j r(,uj)k(u, j) (kr ) = j j r(,vj)k( v, j)
uu vv
; (35a)
1 1

m
(kr ) = j jr(, j)k( , j)
uv u v
with j = d j l j Wj (35b)
1

12
Expressions (34) deliver the aerodynamic damping forces for two normal modes act-
ing, respectively, in the u and v directions; their examination suggests the following
considerations.
1. The coefficient multiplying y r(u ,v ) in the k=r term of the first summation repre-
sents the direct aerodynamic damping factor.
2. The contribution of the other terms of the same summation depends on the val-
ues of the coefficients (35a) for k r . These are zero provided that the j coef-
ficients share the same spatial distribution as the inertia of the system; in this
case the aerodynamic damping mechanism can be denoted as quasi-
proportional; are presumably small in practical applications.
3. The second summation usually shows, at least, one significant terms; it repre-
sents the linear aerodynamic coupling between the normal modes in the u,v di-
rections. Due to this term the aerodynamic damping can be fully proportional
only in the 12 = 0 case, which means that the average flow is aligned to one of
the directions u,v.
Finally, lets develop the expression of the direct aerodynamic damping factor:
m
= CD1 (rr ) = CD1 d j l j W j r(, j)
2

Cr(
a ,u ) uu u
(36)
1

For a lumped mass system the modal inertia coefficient can be expressed in the following
form:
m
M r(u ) = j ( s ) j d 2j l j r(u, j)
2

in which j d 2j is the section area and ( s ) the equivalent density of the structure.
Upon substitution of the preceding expressions the modal damping factor reads::
2

( a ,u ) Cr( a ,u ) 1 1 CD j d j l j W j r(,uj)
r = = (37)
2r( u ) M r(u ) 2 2 f r(u )
2
( s ) j d 2j j l j r(,uj)
Assuming a homogeneous structure and a uniform discretization (constant dj, lj, j) and denot-
(u )
ing as Wr the weighted average (over the squares of the r components) of the mean velocity
values Wj we get:
Wr 1CD
r( a ,u ) = (u )
(38)
df r ( s ) 4
Similarly, for modes in direction v:
Wr 2C D
r( a ,v ) = (v)
(39)
df r ( s ) 4
The expression (38, 39), which are very close to the ones reported in Eurocode 1 [4]
show how the modal aerodynamic damping ratio is proportional to the average velocity
and inversely proportional to the natural frequency. The density ratio (fluid to structure)
usually makes the values (38-39) remarkably small; significant values are found for sys-
tems having a low natural frequency, such as suspended cables, slender (cable stayed or
suspended) bridges or slender tower systems (tall buildings, stacks, ).

13
References quoted in the text
1 S.K. Chakrabarti, Hydrodynamics of offshore structure, Computational Mechanics Publications,
Springer Verlag, 1987.
2 R.D. Blevins, Flow-induced vibration, Van Nostrand Reinhold Company, 1977.
3 G. Piccardo, A methodology for the study of coupled aeroelastic phenomena, Journal of Wind En-
gineering and Industrial Aerodynamics, 48 (1993), pp. 241-252.
4 CEN, Eurocode 1: Actions on structures General actions - Part 1-4; Wind actions, prEN 1991-1-4,
January 20015/05/2007.
5 CNR Consiglio Nazionale delle Ricerche, Istruzioni per la valutazione delle azioni e degli effetti del
vento sulle costruzioni, CNR-DT 207/2008, Jan 2008.

Other general references


C. Borri, S. Past, Lezioni di ingegneria del vento, Firenze University Press, 2006.
E. Simiu, R.H. Scanlan, Wind effects on structures. Fundamentals and applications to design, J. Wiley,
1978.
J.D. Holmes, Wind loading of structures, Taylor & Francis group, 2007.

14

Вам также может понравиться