SOME IMPORTANT ISSUES RELATED TO

WIND SENSITIVE STRUCTURES

Dr. Somnath Ghosh
Professor of Civil Engineering
Jadavpur University
Kolkata 700 032
India

For centuries together, the action of wind forces on structures has been
classified as an act of a supernatural deity. Although it is fundamentally
correct, yet, only within the last few decades, serious attempts have been
made to measure,

Wind Speed

Wind energy

Wind force estimation on a Flexible or Rigid structures - based on
statistical considerations.

The probability that a particular wind force may not be exceeded
is stated, and this probability can be translated into a probable
lifetime for the structure. The choice of a life-duration is then consciously made.






Wind has got two qualitative aspects


Gentle breeze blowing over a considerable interval of time represents
the effect of time-average component of wind velocity.

There might be a sudden draught of air, with varying speed,
flow pattern, direction of flow, extent of flow etc.


) ( ) ( t V V t V
Z Z Z
' + =



**Time average component is measurable, predictable and
hence deterministic.

**Time varying components neither measurable in that sense, nor it is
predictable for any time interval, in other words non-deterministic or
random. Such an aerodynamic force has no recognizable predominant
frequency, but varies in a random manner. Effect of such a force on
structures is called as Random frequency excitation.

To determine the excitation force on a structure, two statistical
characteristics are needed

*Mean square energy spectrum, giving the energy content at each
frequency band-width

*Spectral coherence between two points on the structure.






The random force varies in time and amplitude in an entirely random
manner about a mean level.


It can be analyzed mathematically into forces at different frequencies,
with a mean energy level at each frequency.









3 / 4
10
) 1 (
4 ) (
x
x
kV n S
V
+
= '
10
V
nL
x =
) ( ) , ( ) , ( n S ij n R ij n S
V V
' A = A '
) (
) , (
) )) ( ( )) ( ((
) , (
) , (
5 . 0
n S
ij n S
n S n S
ij n S
ij n R
V
V
j V i V
V
'
A '
=
' '
A '
= A
) ( ) , ( ) ( ) ) , ( ( ( ) , (
2
n S A n C n S dA ij n R A n S
V
A
V V
' = ' A = '
}}
) / * 10 1 (
1
)) * 3 /( ) * 8 ( 1 (
1
) 1 (
1
) , (
2
2
h m h m
V b V h
A n C
+ + +
=
o
}

= e e e
t
d t j H t h ) exp( ) (
2
1
) (
}

= e e e d t j t h H ) exp( ) ( ) (
) / 2 ( ) ) / ( 1 (
1
) (
2
m m
i
H
e e e e
e
+
=
) ( ) ( ) (
2
e e e
F Y
S H S =
) ( ) ( t F F t F ' + =
bh PV C F
h d
2
5 . 0 =
Plot of energy against frequency gives the response spectrum,
and the total area represents the total energy in the random force variation.
This is added statistically to the mean non-variant wind energy.
In fact, energy level at each frequency does not remain constant, but varies
in the same random manner, so that the energy level
is a mean value.
This facts are usually considered in the theoretical analyses.
ALONG WIND ANALYSIS

*A spectral density function normally considered which gives the spatial distribution of the frequency
of the excitation force, generated by the dynamic component confronting a structure.













ALONG WIND RESPONSE CURVE FOR TIP VELOCITY
UNIFORM CHIMNEY AS PER TIME DOMAIN ANALYSIS
-2
-1
0
1
2
0 0.2 0.4 0.6 0.8 1 1.2 1.4
TIME(sec)
V
E
L
O
C
I
T
Y



(
m
e
t
r
e
/
s
e
c
)
*A fully developed flow continuum normally assumed to prevail,
encompassing the vicinity of the structure in order to obtain the true
picture.
*A coherence function is also considered to take care of the spectral
coherence of the dynamic wind force.
*The response spectrum is obtained by introducing a system transfer
function.
*Finally, the standard deviation of the dynamic components of wind
velocity, force and displacement are obtained by integrating the
corresponding spectral densities, taking help of numerical methods
and then gust factors are obtained.
*Once the random excitation spectrum is known, the response of a structure
is simple calculated, if its characteristics such as mass, stiffness, damping
etc. are known. All types of vibrations are affected by damping. Damping is
broadly used to denote either the dissipation of energy in, and the
consequent decay of, oscillation of all types, or the extent of the dissipation
and decay. The energy losses arise from frictional, or analogous, which are
unavoidable in any system, or from the radiation of energy to space or to
another system. Damping may be defined as the inherent force that causes
the gradual dying out of mechanically excited natural vibrations within a
structural member and reduces the efficiency of transfer of dynamic forces
through a structure. From the energy point of view, it is the strain energy
lost within a structure as it is caused to deflect, usually being significant only
under dynamic loading conditions.
A ready-made, user-friendly software for estimating Along-wind stochastic
response of structures need to be developed for the practicing engineers,
helping them in understanding, the behaviour of the structure more clearly.

Results of the analysis being more transparent and more in the realm of
reality, will help in obtaining progressively safe, efficient and economic
structures. Some attempt has been made to popularize exact analysis
among practicing engineers. Analysis need to be made a wide velocity range
which will help in understanding those band-width which are more prone to
cause random excitation and extent of frequency sensitivity.

91 cases (uniform and tapered chimney) have been studied.

Parameters considered in the investigation are Gust factor, Height to Diameter
ratio (H/D), Basic Wind Velocity (V
a
) and Tapering angle.











GUST FACTOR vs H/D
2
2.2
2.4
2.6
2.8
3
3.2
10 15 20 25 30 35 40 45
H/D
G
U
S
T

F
A
C
T
O
R
(
G
F
)
Vb=33m/s
Vb=39m/s
Vb=44m/s
Vb=47m/s
GUST FACTOR vs BASIC WIND
SPEED
2
2.2
2.4
2.6
2.8
3
3.2
33 38 43 48 53 58 63
BASIC WIND SPEED
G
U
S
T

F
A
C
T
O
R
H/D=20
H/D=30
H/D=35
H/D=40













GUST FACTOR vs DIAMETER
2
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
3
3 5 7 9 11 13
D(m)
G
U
S
T

F
A
C
T
O
R
(
G
F
)
Vb=33m/s
Vb=39m/s
Vb=33m/s
Vb=33m/s
d(dyn)/H vs H/D
0
20
40
60
80
100
120
10 15 20 25 30 35 40
H/D
d
(
d
y
n
)
/
H

Dt=4.0 and
Db=15.0
Dt=4.0 and
Db=15.0
Dt=4.0 and
Db=15.0
BASE MOMENT vs H/Dav
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
10 15 20 25 30 35
H/Dav
B
A
S
E

M
O
M
E
N
T
(
M
B
)


(
x
1
0

k
N
-
m
)
Vb=47m/s
Vb=44m/s
Vb=39m/s
Vb=33m/s
For tapered chimneys average diameter concept as well as C.G. level
diameter concept are considered, but the results are not significantly
different.


The present result show similar trend as obtained by many previous
researchers.

Though in many cases it is matching but in several cases it is not
tallying with codal values. Therefore, the codal method is not a
general one.
However, more parametric and field studies are essential to arrive at
a definite conclusion.



CHIMNEY
MARKED



D
t




D
b




D
av




H



H/
D



t



f

ALONG WIND TIP
DISPLACEMENT
BY METHOD

I

II



m

m

m

m



m

Hz

m

m

CU1

10

10

10

150

35

0.5

2.39

0.38

0.332

CU2

10

10

10

200

20

0.5

1.30

1.33

1.2

CU3

10

10

10

300

30

0.5

0.832

7.37

8.3

CT1

7.5

12.5

10

150

15

0.5

1.88

0.29

0.27

CT2

7.5

12.5

10

200

20

0.5

1.06

1.02

0.96

CT3

7.5

12.5

10

300

30

0.5

0.495

6.57

6.88

CT4

4.0

15.0

10

150

15

0.5

1.37

0.51

0.25


I According to 1S-875 Part III- 1987
II According to Proposed present method

Across Wind Response

Vickery and Basu (16,17) seems to be the most acceptable world-wide, as it
has been found to yield estimates of the response that have the least scatter,
when compared to the results from wind tunnel test and field measurements (9).

Two forms of the Vickery Basu model have been suggested for codal application

*Simplified Model Requiring more effort than the latter, as it involves
modal analysis.
*Equivalent Static Model - Two separate formulas have been recommended
for chimneys with little or no taper, and for highly tapered chimneys.
A lesser response is predicted in the case of the latter.

}

+ u
=
h
r e a a r r z r e n
a bv L p
r
m d K dz
h
m S
l d C g
Y
taper no or little with towers For a
0
2 / 1
,
2 2
, ,
2
2
2 / 1 3
)] / )( ) (
1
[( 8
)] 4 2 ( [

: ' ' ) (
, t
t
}

=
h
r e a a r r z r e n
f r zc a r c z L p
r
m d K dz m S
t l d C g
Y
taper high with towers For b
0
2 / 1
,
2 2
, ,
2
2
2 / 1 4
, ,
)] / )( ) ( [( 8
)] 2 /( [ ) (

: ' ' ) (
, t
t
onent law power
z
d
D h
D D
f unction taper a t
z height at diameter d
e r the in critical is shedding which at height z
t coef f icien damping c aerodynami K
Menon Hsieh and Kareem ly respective
and r f or ratios damping
z height at tower the of height unit per mass m
dz dz m
e r the in height unit per mass equivalent m
d h ratio aspect
diameters of number in length n correlatio l
third one top the over averaged diameter ef f ective d
m kg air of density
with e r the in z height at shape e
of value peak a has it velocity wind and width
band spectral on depends that f actor ion amplif icat an
number Strouhal S
t coef f icien lif t rms C
f actor peak a g
r c
r zc
b
b t
f
r c r zc
th
r c
a
r
z
h h
r z r z Z
th
r e
a
r h
th
r z
bv
n
L
p
exp
/
/ 1
' '
mod ' '
) 1994 ; 1986 (
3 2 , 1 043 . 0 , 021 . 0 , 015 . 0
) ( / ) (
mod
/
0 . 1
' '
/ 23 . 1
1 mod mod
5 . 2 ;
0 . 4
,
,
, ,
,
0 0
2
,
2
,
,
3
, ,
=
+

= =
=
=
=
= = =
=
=
=
= =
~ =
=
= =
= =
= u
=
=
~ =
} }
o
o
,

A modified version of the Vickery-Basu Simplified Model has been incorporated
in IS:4998 -1992.

Modified versions of the Vickery-Basu Equivalent Static Model had earlier been
incorporated in the ACI 307 (1988) , NBCC (1980), Pinfold (1984).

There are differences, however, in the various codal recommendations with
regard to the values of the various aerodynamic parameters, the structural
damping ratio, as well as the definition of tapering. The concept of taper should
ideally relate to the slope in the elevation of the tower, i.e. it should not only involve
the Taper ratio( D
t
/D
b
), but also the Slenderness ratio (h/D
b
) of the tower.
IS:4998-1992 takes this into consideration.



It is instructive to identity the conditions whereby the expected maximum
moments under Across-wind analysis exceed those which are obtained
from Along-wind analysis. Such a study was reported by Vickery (18)
who showed the strong influence of the tower geometry (in terms of the
aspect ratio ) on the response.
Typical across-wind moment profiles
0
50
100
150
200
250
0 50 100 150 200 250
M z,ac (MNm)
Z
(
m
e
t
r
e
s
)
mode 2
mode 1


LET US DEFINE q
mo
= Codal Base moment using Codal method
Base moment using Vickery Basu model

Comparison of q
mo



D
t
/D
b


h/D
b


ACI 307

Pinfold

IS 4998

0.5

7

1.392

1.177

2.823

0.6

7
8

1.066
1.118

1.091
1.166

2.022
2.098

0.7

7
8

0.838
0.977

1.183
1.345

1.503
1.733

0.8

7
8
9

1.548
0.992
1.087

2.282
1.444
1.544

0.518
1.664
1.717

0.9

7
8
9

3888
1.024
1.017


1.472
1.552


1.614
1.656

0.10

7
8
9
10

3.404
0.098
1.566
1.701


1.594
7.576
1.864


1.675
1.608
1.647

It has been observed that considerable uncertainty is involved in the prediction of the
Across- wind response, when compared with
actual field measurements (9,16).

This has been attributed more to the uncertainties in the specification of the various
design parameters than to the error in the model itself.

The uncertainty in Across-wind analysis should be logically accounted for,not by
ignoring the analysis altogether , but by prescribing an appropriate Wind load
factor
w.
However, the results of comprehensive reliability studies reported (8)
indicate that the load factor for the Across-wind loading condition should be higher,
by approximately 15%, than that specified for the Along wind loading condition. As
a matter of fact, considerable disparities exist among all the codal specifications for
the various load and resistance factors. In particular, the values of
w
are found to
vary from 1.4 to 1.85. Hence the disparities among the codes relate not only to the
analytical procedures to estimate the expected maximum wind moments,
but also to the specified safety factors.

There is an evident need, worldwide, to achieve uniformity among international
standards, and to adopt optimal load and resistance factors based on comprehensive
reliability studies .

CONCLUDING REMARKS

*In case of Along wind analysis the present results show similar trend as
obtained by many
previous researchers.

*In many cases it is matching but in several cases it is not tallying. Therefore the
codal methods are
not a general one. However, more parametric and field studies are essential to
arrive at a
definite conclusion.

*Generally, for towers which are relatively stocky and cylindrical in shape are
most susceptible to
high Across-wind loading, if the tower is highly tapered. In towers that are
slender but
cylindrical, there are possibilities of the second mode (Across-wind) condition
becoming
critical- particularly with respect to moments.




*The provisions for Across-wind loading in ACI 307, NBCC/Pinfold and
IS 4998 are found to yield results that are generally conservative, when
compared to the original Vickery-Basu model, on which these simplified
methods are based.

*It is revealed from sensitivity analyses that the most sensitive basic
variables influencing Across-wind response are the damping ratio, lift
coefficient and strouhal number.

*The probability distribution of Across-wind base moment, generated by
Monte Carlo simulation, indicates that the overall uncertainty in the
estimated moment is very high.

The international codes incorporated recommendations to estimate
Across-wind moments in reinforced concrete chimneys during the course
of the recent past decade or so.

*Although there has been considerable research in this area in the past,
the problem of
accurately predicting the Across-wind response is yet to be resolved
fully. It is essential to

( i) to identify the conditions whereby Across-wind response governs
the design

(ii) to assess the disparities in the prevailing codal estimate under the
above conditions

(iii) to examine the influence of uncertainties in the predicted response.