Topic 4: WIND INDUCED VIBRATIONS

Instructor: Dr. Tim K.T. Tse ([email protected])

 Crosswind or Lift, Along-wind or
FL , y Drag, FD, x

Fx, x
α

Mean Wind u u
Fy , y

(a) Wind Axes (b) Body Axes

2
‒ The wind-induced dynamic forces and responses are
conventionally resolved into two orthogonal directions
a) parallel (along wind or drag) and perpendicular
(cross wind or lift) to the mean wind direction, or
b) parallel and perpendicular to a direction related to
geometry of the structure.
c) The torsional dynamic response of a tall building may
also occur when the alongwind and / or crosswind
dynamic forces do not coincide with the elastic
center of the building.

3
‒ The turbulence or gustiness in the natural wind causes
buffeting which directly produces fluctuation forces on a
structure in the flow
a) wind-induced force and the consequent response
are random in nature.
b) The magnitude of the fluctuation loading is a function
of the intensity of turbulence.
c) The turbulence scale, in relation to the size of the
structure, determine how well the fluctuations are
correlated over the structure.
- Relationship between the loads on a structure and the
velocity fluctuations.
- More precisely the correlation and spectrum of the
loads and the correlation and spectrum of velocity
fluctuations

4
‒ The effectiveness of a gust in terms of producing a load on
a large structure depends largely on the gust size in relation
to the size of the structure
a) In the case of high frequency component the
pressures or forces produced are well correlated or
organized only on quite small areas of the structure.
b) The total effect produced by these high frequency
components is small because in some areas, they will
tend to produce increased loads while simultaneously
at other parts of the structure there will be a decrease
in load.
c) the very low frequency components are felt over the
whole or at least large areas of the structure
simultaneously.

5
Turbulence in incoming flow:
• Incident wind flow is turbulent
• Wind velocity is already a
fluctuating vector: U = U(t)
• Fluctuations in wind speed and
wind direction
• “Turbulence buffeting”

Body-generated turbulence:
• Flow separation, wake,
vortex shedding 
fluctuations in wind
pressures

6
Turbulence buffeting Instantaneous
 Consider wind pressure at point wind vector
A on one face of a body (at an Mean wind
a q
angle q to the mean wind flow vector A
pressure at
direction) this point, p
 When wind vector changes
direction, the face is at a
different angle to the flow

At a time when wind vector flows at the mean wind speed and
mean wind direction, pressure at A is:
p(t )  p  C p (q )  12 U 2

7
Turbulence buffeting Instantaneous
 At any time when U = U(t), wind vector
pressure at A is: Mean wind
a q
vector A
p(t )  C p q (t )   12 U (t ) 2 pressure at
this point, p

U  u (t ) 
 q  a (t )    
q (t )    U (t )   v (t ) 
  (t )   w(t ) 
 

p(t )  C p q  a ,    12  (U  u ) 2  v 2  w2 

 C p C  1
p(t )   C p (q )  a  p     U 2  2U u  u 2  v 2  w2 
 q   2

8
Assumptions: Instantaneous
wind vector
1. Neglect higher-order terms;
a q
Mean wind
2. Ignore the change of Cp due vector A
to wind direction change. pressure at
this point, p

 C p C  1
p(t )   C p (q )  a  p     U 2  2U u  u 2  v 2  w2 
 q   2

p(t )  C p (q )  12  U 2  2U u(t ) 

9
Fluctuating wind pressure:
p(t )  C p (q )  12  U 2  2U u(t ) 
p(t )  p  p(t )
Mean pressure: p  C p (q )  12 U 2

Fluctuating part: p(t )  C p (q )  U u(t )

 r.m.s. pressure:  p  ( p) 2  C p (q )  U  u

40

35
Wind pressure (Pa)

30

25

20
Time

10
 Peak wind pressure:
pˆ  p  g p p
pˆ  C p (q )  12 U 2  g p C p (q )  U  u

 
pˆ  C p (q )  12  U 2  2 g pU  u  C p (q )  12  U  g p u 
2

Gust wind speed: Uˆ  U  g u

Assuming same peak

pˆ  C p (q )  12 Uˆ 2
factors (gp = g) 

“Quasi-steady equivalent-gust” approach:

• Maximum pressure is induced by maximum gust in the same
way as mean pressure is induced by mean wind speed.
11
 In general,
Good for bluff bodies made up
of walls joined at sharp corners
 Cp usually same sign on a
face

• Failed on bodies made up of

smooth curved walls  Cp
can change from +ve to –ve
within a small distance
• When (mean) Cp = 0 at a point,
mean and peak pressures are
both zero!

12
• For a loading surface (e.g., a curtain wall on a building), the
wind force on its external surface always fluctuates with time
because pressures at all points fluctuate with time
• At any time, the wind force is the integral effect of pressures
on the surface:  
F (t )   p(t )dA
A

 Mean force:

F  F (t )   p  dA
A

13
 Peak force:

Fˆ   pˆ dA
A

Reasons
• Turbulent wind force contains mixture of all shapes and sizes
of eddies.
• Maximum pressures at all points on a surface would occur at
the same time only if the area of the surface is within the size
of the eddies.
• Depends on the relative size of the loading surface and the
eddy sizes
14
Are pressures at 2 points acting together?
• Depend on the correlation between the
wind pressures (a loading effect) at
these 2 points.
• The 2 points are spatially separated on
the body surface by a distance x .

15
 Are p1 and p2 acting together?
• For a single eddy (wave) of size d:
• If x < d, pressures (water level) at
the 2 points rise and fall together
• If x > d, pressures (water level) at
the 2 points rise and fall
independently

Effect of eddy size in wind flow becomes

frequency of fluctuations in the pressure signals
40

35
Wind pressure (Pa)

30

25

20
Time

16
Correlation between p1 and p2 
• Correlation to be made between
these two pressure signals at
different frequencies  correlation
analysis in the frequency domain
• In signal processing  “coherence”
function:
S p1 p 2 ( n )
 (n ) 
2

S p1 p 1 ( n )  S p 2 p 2 ( n )

17
Expected coherence function

18
 Different wind speeds and
separations
 Use non-dimensional frequency:
St = nx/U

19
4.1.5 Gust response factor

Velocity Force Response

V
P Y

Gust Aero Aero force Mechanical Response

spectrum admittance spectrum admittance spectrum

σ2 (V) σ2 (P) σ2 (Y)

Elements of the statistical approach to gust loading

(Davenport, 1967)

20
Mean Square Value
 x2  lim  x 2 t dt
1 T /2
T  T T / 2

Root Mean Square

(r.m.s)
x t dt
1 T /2 2
 x  lim 
T  T T / 2

Variance
 x2  lim  xt    x 2 dt
1 T /2
T  T T / 2

x t dt  2 x lim  xt dt   x2

1 T /2 2 1 T /2
 lim 
T  T T / 2 T  T T / 2

 x2   x2   x2

Standard Deviation
 x   x2   x2

21
Probability Distribution Function
P( x)  Pr obxt   x   lim
1
 ti
T x T i

P    0 0  P x   1 P   1

Probability density function

Px  x   P x  dP x 
px   lim 
x  0 x dx

22
Normal or Gaussian Distribution Rayleigh Distribution
1  e  x 2
Px  
1 x x 2 2
2 
dx
2
x0
e Px   
 0 x0
px  
1  x2 2
e  xe x 2 2 x0
2 px   
 0 x0

23
  x  x 2 
px  
1
exp  
x 2  2 2
x 

24
Autocorrelation function PSD function

R f    lim f t  f t   dt S f     R f  e i d
1 T /2
T  T
T / 2 Fourier 
Transform S f    0 (nonnegative)
S f    S f    (even function)

Autocorrelation
Time Domain + Frequency Domain
Fourier Transform
xt    f  g t   d S x    G  S f  
t 2
0

G   
2 1
1       2x  
n
2 2
n
2

S f    S0 (white noise excitation)

25
4.1.5 The response of small, point-like structures
‒ consider a very small structure of area A placed in the
atmosphere boundary layer,
‒ the bulk of the energy of the turbulence of gusting is at
wavelengths much greater than a typical dimension of the
structure,
‒ regard the flow past the structure as quasi-steady

26
 Fluctuating drag forces on a bluff body
u  t   u  u  t 

which u is mean longitudinal wind speed and u t  is fluctuating

longitudinal wind speed.

Drag force: 1
FD  t   CD Au 2  t 
2
1
 CD  A u 2  2uu   t   u 2  t   CD is the drag coefficient
2

1 2
FD  t   CD  u A  CD  Auu   t 
2 FD is mean drag and
 FD  FD  t  FD' t  is fluctuating drag

27
Spectral
representation:
S D  n    CD A u  Su  n 
2

FD2
 4 2 Su  n  Su(n) is the spectrum of
u the longitudinal turbulence

Aerodynamic
n A
admittance   
function:  u 

FD2 2  n A 
S D  n   4 2    Su  n 
u  u 

28
Experimental and theoretical values of aerodynamic admittance for flat plates and
prism normal to the flow (Re=2x104) (Vickery, 1966 & Davenport, 1967)

29
Gust response: xˆ  x  g f  x Static component: FD
x
k
Spectrum of the displacement response:

1 1
Sx  n  SD  n 
k 2   2  2 n
2
n
1      4 S2  
  n0    n0 
1
 2 H  n  SD  n 
2

k
H(n) is the Mechanical Admittance
FD2 1 2  n A 
   Su  n 
2
 4 2 2   H n Function.
u k  u 

FD2 1  n A
  H  n  Su  n  dn
2
 4 2 2
2
x  2 
u k 0
 u 

30
4.1.5 Gust response factor

AR

AB

no Log n
Response spectrum of a SDF system to turbulence buffering

31
Variance of the fluctuating component of the displacement:

 x2  u2 1  n A 2 S n
4   2    
u
H n dn
x2 u 2 k2 0
 u  u2

 u2
4 2  AB  AR 
u

 n0 A  Su  n0   2
AR    2
  H  n  dn
 u  u
2 0


2 n0 A  Su  n0   n0
   
Resonant component:  u   2
u 4 S

  n A  S u ( n)
AB    2  
 dn
 u  u
0 2

Background excitation: B

32
  n0 A  Su  n0   2
AR    2
  H  n  dn
 u  u
2 0

 n A  Su  n0   n0
  2  0 
  u 4 S
2
 u

S be a size reduction factor E be a gust energy factor

 n0 A   n0 Su  n0 
S   
2
 E
 u  4  u2

 x2  u2 
SE 
1
SE x u  SE 
AR     2

 s 
4 B
s 2  
 s 
x2 u2  B
x u 

33
Expected peak value of x: xmax  xˆ
 t 
 x  xmax
 x  g fx

gf is an average peak factor: 1

u  SE  2
xˆ  x  g f x 2  B  
u  s 
 1

  
 x 1  2 g f u  B  
2
SE
 
u  s  
xˆ  
G
x
1
u  SE  2
 1  2g f  
 S 
B G is a gust response factor
u 

34
4.1.6 Aerodynamic damping
− Fluctuation forces induced by turbulence buffeting exist
irrespective of the movements of the structure. However,
aerodynamic damping forces are induced only by the
movement of the structure
− they are appreciable in comparison with the forces due to
structural damping and can markedly influence the
behavior of a structure in wind

35
4.1.6 Aerodynamic damping
− Assuming the structure is a linear system and the amplitude
of vibration is small, the drag force acting in the x-direction
is
1 xሶ
In which UR (t) = V – ,
FD t = CD ∙ ρUR2 t ∙ A (4.30) the relative wind speed
2 and A is the area normal
to the wind
1 2 2
1 2
FD t = CD ∙ ρV t A − CD ρVxሶ A + CD ρxሶ A (4.31)
2 2

a) The first term is the mean drag force and the second
term is an aerodynamic damping force.
b) The negative sign of the damping forces means that it
opposes the motion and energy is extracted from the
structure
c) The aerodynamic damping ξa is positive and adds to
any structural damping.

36
4.1.6 Aerodynamic damping
− For vertical structures, assuming a linear first mode, and if
the mass distribution, the cross-sectional shape and CD are
constant, the expression for aerodynamic damping is
3ρVn CD
ξa = (4.32)
4πρs 3 + γ no d

− For typical values of ρ = 1.2kg/m3, ρs = 200kg/m3, CD = 1.2

and
− d = 30m, Vh = 50m/s, γ = 0.25,
1
ξa = 10−3 (4.33)
no

− For a typical tall building where no = 0.2Hz, ξa is about 0.5%

of critical. This is a significant addition to the structural
damping which is typical 1 % to 2% of critical

37