, Journc, l o f Industrial Aerodynamics, I (1975/76) 239--247 239

© Elsevier Scientific Publishing Company, Amsterdam -- Printed in The Netherlands

MEASUREMENTS OF WIND-INDUCED VIBRATIONS ON A CONCRETE

CHIMNEY

F.P. MULLER and H. NIESER

Institut ffir Beton und Stahlbeton, University o f Karlsruhe, Karlsruhe (W. Germany)
(Received February 13, 1975)

Summary

In the years 1971--1973, measurements of wind velocity and structure vibration were
made on a concrete chimney of 180-m height in South Germany. All data were recorded
continuously in an analogue form. From the results, the statistical properties of excitation
and response, as well as the relationship between these two quantities, were determined.
The report contains a short description of the problems that generally occur with mea-
surements of this type and shows methods of solving them.
Some of the results are given as mean values and functions of correlation and spectral
density. They show reasonable agreement with the assumptions already known concerning
the profile of mean wind velocity and gust spectra. The data of wind velocity at different
levels were analysed to determine coherence and phase of cross spectral functions. The
power spectra of structure response were calculated theoretically from the data of wind
velocity and then compared with the experimentally derived spectra. The result of this
study is that theory and experiment are in good agreement if cross correlation of wind
velocity is taken into account.

1. I n t r o d u c t i o n

In the last two years, measurements of wind velocity and structural vibra-
t i o n s w e r e c o n d u c t e d o n a c o n c r e t e c h i m n e y o f 1 8 0 - m h e i g h t in S o u t h e r n
G e r m a n y . T h e b u i l d i n g is l o c a t e d in a c o m p a r a t i v e l y f l a t r e g i o n o f t h e u p p e r
R h i n e V a l l e y w e s t o f t h e c i t y o f K a r l s r u h e . W i t h i n a r a d i u s o f s o m e 4 0 0 m it
is s u r r o u n d e d b y r e f i n e r y s t r u c t u r e s u p t o 3 0 m in h e i g h t . T h e p r e v a i l i n g
winds come from the south-west.
T h e c h i m n e y d e s i g n is o f c i r c u l a r - r i n g c r o s s - s e c t i o n , t h e d i a m e t e r a n d w a l l
thickness of which taper continuously from the base to the top. A sketch of
t h e s t r u c t u r e is s h o w n in Fig. 1.
T h e n a t u r a l f r e q u e n c i e s o f t h e c h i m n e y a r e as f o l l o w s :

1st mode: no = 0 . 2 6 s -1
2nd mode: nl = 0.99 s -l
3rd mode: n2 = 2 . 3 4 s -~
240

MASS STIFFN[SS
B~m] s[cm] DISTRIBUTION
~+ 180_rnWD,WV,A L,,75 17,0 9,Z~[ Mp/m] 6L, [ m~]

WO=WINODIR[CTION y+130m WV 5,9g 17,0 10,9 12,6

A=A-4
CCELERATION~o
WV•WIND VELOCITY

S~SIR~SS
LINING-
,z_+80 m _ - 7,Z,2 25,0- ~'~I
/.._ 17,L, / ~ 36,2

Q j+35m WV 8,82 60,O 38,7 131.6

s
,__ -+0 $ .... 9,8~_ 60,_0_ j 1&3,6 187,9

Fig. 1. Dimensional sketch o£ the chimney and arrangement o£ the measuring points.

Measurements of wind velocity were made at three different levels at 35 m,

130 m and 180 m above ground. These data were recorded continuously in
analogue form as well as the acceleration at the top and strain at the base of
the building (Fig. 1). 24 runs varying in length from 10 to 15 min were ana-
lysed by computer to estimate the mean value, variance, power spectral den-
sity and cross-correlation spectral density of each quantity measured.

Analysis of records

The instruments used for the measurements of wind velocity were three-
cup anemometers arranged circumferentially around the chimney in sets of
4 at each of the 3 levels. Because of the inertia of cup anemometers and the
influence of the building on the airflow, the recorded data do not represent
the true windspeed. Therefore a correction had to be applied to the data.
The influence of the instrument's inertia was to a large extent eliminated
by prior calibration in a wind tunnel at fluctuating velocities.
In this way, a relationship was derived between the true velocity, v, on the
one hand, and the recorded velocity, v*, and its differential dv*/dt on the
other hand [ 1 ]

v = f (v*, ~-TdV*]

The corrected data were compared with that of a hot-wire anemometer. The
comparison shows that for frequencies up to 1.5 s -1 the two were in good
agreement.
Wind tunnel studies were also conducted to determine the influence of the
flow around the building on the wind velocity which demonstrated that the
 241

flow can be described b y a potential equation in a definite zone in the

approaching airstream (Fig. 2).
At each level of the chimney, the instrument located in the most favourable
position with regard to deviation from potential flow was recorded. Thus, the
free-stream velocity could be calculated from the recorded data b y means of
the potential equation.

j~-~- ~ ----~.MEASUREDDATA>THEOREIICALDATA
........

V LUUl,, ,,oo 'L,/',

/

Fig. 2. Difference in percent between theoretica| atld measured data of wind velocity
around a cylinder.

Generally, accelerometers, as arranged at the t o p of the chimney, are not

suitable to respond to slowly varying movements. The sensitivity of the in-
struments employed permitted the investigators to record all vibrations with
frequencies above 0.6 times the fundamental frequency of the building and
with amplitudes greater than 1 mm.
For computation of the displacements of the chimney t o p from measured
acceleration data, double integration yields large errors because, comparatively,
this method is very sensitive. Therefore, a mechanical model was chosen to
determine the displacement in digital form (Fig. 3) [1]. It should be noted
that the data calculated in this manner are high-pass filtered. Therefore, the
vibrations can only be judged in the range of the natural frequencies of the
building.

Experimental results

From inspection of the dynamic components of displacement in and nor-

mal to the wind direction (Fig. 4), it can be seen that the vibrations normal
242

I I
~ ~ - ~ -6 It-~l
+rt) ,t) r it: = ITI.e .sinu !t-z', d~"
+x(t) o

~V

co: CIRCULARFREQUENCYOF[HI IF~'=&,5 AND £ ~ 2,Gw,WE OBTAIN

SPRANG-MASS SYSTEM
r(tj ~ 0,985xit).
£: CIRCULARFREQUENCYOF THE
EXCITATIONFUNCTION
ASSUMING x { t } r(t),
4: LOGARITHMrEDAMPINGDECREMENT
THE ERRORWILL BE :~ 1,5°/o.
Fig. 3. Mechanical model for determining the deflections x ( t ) from acceleration data ~(t).

[cm]
12
AMPLTuo
NTHEWNOORE
N /
10 - -- - AMPLITUDE NORMAL TO THE W I N D /

O- @

L,

2
(180)
0 I I I I I I [ [ [ I I I I J I I I
& 8 12 16 20 2/, 28 32 36 ~m/sec]
Fig. 4. Oscillation amplitude versus wind velocity.

to th e wind reach a m a x i m u m value at relatively slow velocities. This effect is

due to resonant vor t e x shedding excitation. A calculation of the Strouhal
n u m b e r yields S ~ 0.19.
The lift response is of the same order of magnitude as the dynamic drag
response, however statical displacements must be added to the dynamic
response in the wind direction. Thus, the loading in the drag direction deter-
mines the stability of the structure in this case.
Measurements of wind velocity at different levels confirm the pow er law
o f the mean wind velocities at least at the measuring points (Fig. 5). Neverthe-
less, th e power law e x p o n e n t was found to have a value of 0.35 which was
greater than the values generally assumed for such a t y p e of surface. The
results o f the spectral density calculations which are shown by s m o o t h curves
on Fig. 6, are in good agreement with the assumptions of Davenport [2]. The
 243

z =180 m

+~

z =130 m

SYM•L
L •
V(180)
RUN[m/sec]
11~ 2o~7
oi
\O

l o 115 71,3
L + 118 19,0
+ [72 37,3
l ° 1 ~3 3],~ z ko,3~1
~124 31,1

z = 35m
\\o
o

Vlz)
0 0,5 1,1] V(18Oi
Fig. 5. Profile of the mean wind velocity,

n.SvvInl
T 3
6v 1 ~

I / / ~.,.-~ \ \ 1 22 130 2B,2

/ 4" / "~ \\\ 3 23 130 29,7

k / .~_k\\ ~ ~3 Js 18,1
/ /,/ \~\ ~ x mo +i,1
/ /- \\\\\ + +L, ++ ++,+

O'2F / ~k~'~ THEORETI[ALRELATIONSHIP

L / \\\~\ AOCORDN
I GTODAVENPORI

11 5

~ ' ~ + )
+ :,,oo+,,o+
n

+'+++,I+
10-1 100 101 10? 103
Fig. 6. Power spectra of horizontal wind velocity.

u p p e r limiting f r e q u e n c y is d e t e r m i n e d b y the limited f r e q u e n c y response o f

the cup-anemometers.
As can be seen b y cross-correlation f u n c t i o n s the gusts reach various levels
o f the building with d i f f e r e n t t i m e lags. This is s h o w n b y a shifting o f t h e
244

peaks relative to the origin r = 0 (Fig. 7). This difference of phases produces
a favourable effect cn the strain distribution of the building and may be
taken into account at the design stage. The recognition of this phenomenon

,/RI,(O)R~,,(O~ RUN22
~ ~ SYMBOLz[m] z'[m] ! Vm[m/sec]
~,5 @ + 130 180 r 30,25
I\ [01 : ~30 351~3,00
+~.7 0,4L:'~ I®1 o 1130i3sl 25,05

/e\/;2 o,1
"" 0
f
f ..,

~' -L,O -30 -20 -10 10 20 30 L,O 50 +r[sec]

Fig. 7. Cross-correlation functions of wind velocity.

is expressed in the cross-spectrum, generally stated as a coherence function

and a phase relationship. The results of wind velocity measurements are in
good agreement with the mathematical functions given in Figs. 8 and 9. The
functional phase relationship, generally not specified in previous publications,
may be described clearly by a linear function as shown in Fig. 9.

C~oh RUN 22
,~ SYMBOLz[ml z'[m; Vm[m/sec][
1,or+__t ~_~ + 13o •
__ 1130 _20,25 4
130 35 23,00/
0,8 ~,~ o 180 35--,--25,05
[ ntz-z'l I

0,/-.* +++o~ o o ~/m= ,~[V{z)*V(z':;]

0,2 o • o
[ n o -e+ o+ ++ oo • • o o • o
~ o • • o
0 l • I"° °t _ n ]z-z']
~m
0,1 0,2 0,3 0,/., 0,5 0,6 0,7 O,B 0,9 1,0
Fig. 8. Coherence-function for two components of wind velocity.
 245

RUN 22
SYMBOLI z [ml z'[mil Vm [m,,'sed]

2 + '130 180! 30,25

18035 25,05

I I I I I I I I " I Z-Z' n
-0,~ -0,3 -0,2 -0,1 ~, 0,1 0,2 0,3 0,~ )~

-11 :z-z'l~ I

~ u," _~7!z; -3 o

Fig. 9. Relationship of phase between two components of wind velocity.

Comparison with theoretically determined data

The cross-spectrum Spp, of the wind forces was calculated from the experi-
mentally derived cross-spectrum of the wind velocity, taking into account the
relation between wind forces and wind velocity, according to McNown and
Wolf [3]

1 V(t) 2 Ao dV(t)
p(t) = ~pC D - - -2- + pC M
d dt

Thus, the power spectrum of the displacements at the top of the chimney
could be calculated from the fundamental relation as follows [ 4]
h h

0 0

It should be noted that for the purpose of calculation a logarithmic damping

decrement 0 = 0.045 was assumed.
Power spectra of the displacements were computed with different values
of the virtual mass coefficient cM. It may be seen that the influence of c M is
exerted only for large values of the non-dimensional frequency ratio h n / V
(h = total height of the building, V = mean wind velocity). In this frequency
range the spectral densities are extremely small. Consequently, the value of
CM may be neglected for further computations.
The power spectra of the displacements, theoretically determined from the
data of wind velocity measurements, were compared with the experimentally
derived spectra. Theoretical and experimental results agreed in so far as only
246

vibrations of the fundamental mode are significant and that all higher modes
are negligible. Hence, it may be assumed that the shape of the structural
deflection at any time was identical with the shape of the first mode. Ttmre-
fore, a well-defined relationship exists between the displacements at the top
and the bending moments at the base as well as between the spectral densities
of these two quantities. Thus, the power spectra of the displacements could
be computed from that of the base bending moments. This assumption can-
not be automatically transferred to other systems. It holds true in the case of
buildings, the section and stiffness of which are relatively constant over the
entire height, or at least not changing abruptly.
The power spectra of the displacements at the top of the chimney deter-
mined by different methods for a specific measurement are given in Fig. 10.

n.S. [crnZ] ' : z,OI,8

lO - -

| i IFROMEASUREMEO
: NTFS
~ ~,.-~,
101.' . . . . . ri. . ~ ~.:_ ~ ' I ' !~ A
~ . ~ ~ ~'|~ FC
RC
OEM
LER
EO
A
IIN
A SA
UH
IR
EG
IEH
O
MIE
FBID
OnTrF
N sT
SRESSATTHEBASE
100.-- , "~.~ THEORETC IALDLY
ETERMN IED
10-1 / , [ } ;~ ~"~ /" FUNCTI
/ IIONI

- ' "!]il .....

,~/~" ]ii~ Ji ~ RUN24
' ../L 1o-026 ~]]! l[ I Vi(IBO] -31,1 [rnlsec]
10"210-3 2 10- Z 5 7 10" 2 n [l/sec ]
Fig. 10. Power s )ectra of displacements at a height of 180 m.

The results of the other measurements show the same tendency. The differ-
ences between Curve (2) and Curve (3) in the low-frequency range are due to
the high-pass filtering action of the time-dependent acceleration function.
Over the entire frequency range, the theoretically derived curve lies above the
experimentally determined curves. Among others, the reason for this variation
may be the assumption of fully-correlated wind forces in the horizontal direc-
tion.
The probability distribution of the response shows a fairly good agreement
with the normal distribution (Fig. 11). This confirms that the response of a
structure to wind may be treated as a stationary Gaussian process, defined by
the values of the mean and root-mean-square responses. The largest value
recorded was about 3.5 times the standard deviation, ax.
 247

o
p{×l ox - -0,5

o V~180)
0,~°o SYMBOL RUN [m,"secj I

NORMALDISTRIBUTION..... ~ o~f+
.L o

°
•
o~'.
&
"+ "
"
*
+
22
2"
J
32,3
33,&
31J

Fig. 11. Probability distribution of the displacements at the top.

By means o f stochastic calculations considering the relationship and param-

eters derived f r om measurement data, gust loading factors were determined as
a f u n c t i o n of the natural frequencies, of the structural height and of mean
wind velocity {Fig. 12). The results are compared with t hat of a deterministic
m e t h o d [5], which are shown by the br oke n line on Fig, 12. There are great
differences, particularly in the low-frequency range. To confirm these values,
further e x p e r i m e n t a t i o n and measurements are necessary.

) 1 ~c0nst
---4-
maxx h

__ ~KS!HLAICH IDETERMINIS~ICMEIHODI
/,,0 ,~,~_~~ ,/h=60m

! i

. . . . . . [-\ h-l!Oml + . _

0 i i [ , , ; ' , l . . . .
0,5 1,0 1,5 no [I/sec]
Fig. 12. Oust loading factors for V(10) = 20 [m/s].
248

References

1 H. Nieser, Schwingungsberechnung turmartiger Bauwerke bei Belastung durch biSigen

Wind, Dissertation, Universit~it Karlsruhe, 1974.
2 A.G. Davenport, The spectrum of horizontal gustiness near the ground in high winds,
Q. J. R. Meteorol. Soc., 87 (1961) 194--211.
3 J.S. McNown and L.W. Wolf, Resistance to unsteady flow: I. Analysis of tests with flat
plate, Engineering Research Institute, Univ. of Michigan 2446--I.P., Internal report to
Sandia Corp., June 1956.
4 I.D. Robson, An Introduction to Random Vibration, Edinburgh University Press, 1964.
5 J. Schlaich, Beitrag zur Frage der Wirkung yon Windsti~ssen auf Bauwerke, Bauingenieur,
41 (1966) 102--106.