Вы находитесь на странице: 1из 23

ARTICLE IN PRESS

Journal of Wind Engineering


and Industrial Aerodynamics 91 (2003) 1363–1385

Parametric experimental study of wire rope spring


tuned mass dampers
Rafik R. Gerges*, Barry J. Vickery
Alan G. Davenport Wind Engineering Group, Department of Civil and Environmental Engineering,
Faculty of Engineering Science, The University of Western Ontario, London, Ont., Canada N6A 5B9

Abstract

An extensive experimental study of wire rope spring tuned mass dampers (TMDs) is
presented. A single-degree-of-freedom system with a pendulum-type TMD was tested on the
shaker table. The pendulum-type TMD employed wire rope spring(s) to provide both the
elastic and the damping forces. The test program included two mass ratios, single- and double-
sided spring arrangements, two wire rope diameters, three spring attachment positions and
four excitation levels. The system was tested under a Gaussian random white noise base
acceleration. Time histories of the excitation, the primary system displacement and the
auxiliary system rotation are presented and, from which, frequency response functions were
obtained. Based on areas under the mechanical admittance functions, TMD response
parameters were evaluated. The dynamic properties of the wire rope springs were determined
through a free vibration test as amplitude-dependent effective stiffness and equivalent viscous
damping ratio. Further, amplitude-dependent relationships for tuning and response
parameters were established and were compared to the experimentally obtained response
parameters. Moreover, the frequency response functions and the response parameters were
compared to those of equivalent optimized linear TMDs.
r 2003 Published by Elsevier Ltd.

Keywords: Wire rope spring; Non-linear tuned mass damper; Random vibrations; Shaker table testing;
Equivalent viscous damping

1. Introduction

Tall light structures such as chimneys and slender towers may suffer from
excessive deflections under wind actions. An effective technique to reduce the

*Corresponding author. Tel.: 519-661-3338; fax: 519-661-3339.


E-mail address: rrgerges@uwo.ca (R.R. Gerges).

0167-6105/$ - see front matter r 2003 Published by Elsevier Ltd.


doi:10.1016/j.jweia.2003.09.038
ARTICLE IN PRESS
1364 R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385

Nomenclature

ag acceleration of gravity
Dc coil diameter
dr wire rope diameter
fn particular vibration mode natural cyclic frequency
fr frequency ratio
g forcing ratio
Hx complex frequency response function for primary system displacement
Hy complex frequency response function for auxiliary system rotation
h distance between spring/damper (or wire rope spring) attachment point
and pivot point
Ia auxiliary
pffiffiffiffiffiffiffi system mass moment of inertia about the pivot point
i 1
K particular vibration mode generalized stiffness
kmin minimum effective stiffness of wire rope spring
kay effective rotational stiffness of auxiliary system
ka effective translational stiffness of auxiliary system
kwr effective stiffness of wire rope spring
M particular vibration mode generalized mass
ma auxiliary mass
R ratio of RMS relative displacement to primary structure RMS displace-
ment
ra ratio of mass moment of inertia about pivot to that about center of mass
Sx spectrum of primary system displacement
Sx. g spectrum of base acceleration
Sy spectrum of auxiliary system rotation
Xr relative/representative displacement amplitude
Xrmin relative/representative displacement amplitude corresponding to mini-
mum wire rope spring stiffness
x displacement of the primary system
xr relative displacement between auxiliary and primary systems
x. g base acceleration excitation
z distance between auxiliary mass center and pivot point
b particular vibration mode damping ratio (% of critical)
ba damping ratio of auxiliary system (% of critical)
be equivalent damping ratio of structure equipped with a TMD
bwr equivalent damping ratio of wire rope spring (% of crictical)
m mass ratio
y rotation
o circular frequency
oa natural circular frequency of auxiliary system
on particular vibration mode natural circular frequency
ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1365

Operators
B
root-mean-square (RMS) of quantity

Superscripts
opt parameters corresponding to maximum be

Abbreviations
AISI American Iron and Steel Institute
IWRC independent wire rope core
RMS root-mean-square
SDOF single degree of freedom
TDOF two degree of freedom
TMD tuned mass damper
WSC wire strand core

dynamic response is adding a tuned mass damper (TMD). The TMD consists of an
auxiliary mass attached to the structure through a spring and a damper. Wire rope
springs, which act as both spring and damper, are attractive from the point of view
of initial cost and maintenance. The use of wire rope springs in TMDs was first
introduced by Hirsch and was marked by KABE in Germany.
Amplitude-dependent effective stiffness and equivalent viscous damping can
approximately model the non-linear hysteretic behavior of the wire rope springs [1].
It is well known that there are optimum frequency and optimum damping for the
TMD, at which, the structure response is minimized. Having amplitude-dependent
stiffness and damping causes the performance of the TMD to vary with the
excitation level. Since it is desired that the TMD operates efficiently for a range of
excitations, one need to show that this achievable with the wire rope springs.
The effectiveness of a wire rope spring TMD in controlling the resonant response
of structures was shown by experimental results of an aeroelastic model excited by
the across-wind [2]. However, due to scaling limitations, only one large mass ratio,
about 0.14, and a 7  7 wire rope structure were considered. Since the performance
of a light TMD is more sensitive to tuning than a heavy system, smaller mass ratios
need to be tested. Also, wire rope springs comparable in dimensions and wire rope
structure to the full-scale systems should be tested.
In this study, a two-degree-of-freedom (TDOF) system was designed and tested on
the shaker table at the Boundary Layer Wind Tunnel Laboratory. The primary
structure was modeled with a single translational degree of freedom while the
auxiliary system had a single rotational degree of freedom. The wire rope springs
were used to provide both the stiffness and the damping. The parametric study of the
auxiliary system included two mass ratios, single- and double-sided spring
arrangements, two wire rope diameters, three spring attachment positions and four
base acceleration excitation levels. The system was subjected to a random base
acceleration and the primary system and the auxiliary system responses were
ARTICLE IN PRESS
1366 R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385

recorded. In the analysis phase, the frequency response functions were obtained and
compared with that of an equivalent optimized linear system [3]. Also, the damping
ratio equivalent to the TMD effect and the ratio of the root-mean-square (RMS)
displacement of the auxiliary system to that of the primary system were compared to
corresponding values for the optimum linear TMD.

2. Structure and TMD modeling

A mixed TDOF system was employed to model the primary structure and the
auxiliary system. Photographs of the system are presented in Fig. 1. The tall interior
steel frame represents the primary system and is a translational degree of freedom.
The auxiliary system is a single rotational degree of freedom represented by a
pendulum attached to the tall frame. The exterior short frame is rigidly attached to
the shaker table to transfer the base excitation to the interior frame.
The primary system stiffness was provided by 22 linear springs (11 on each side)
placed in between the exterior and interior frames, in addition to the bending
stiffness of four hangers. The interior frame was hung from the exterior frame with
four steel strips, one at each corner. The steel strips had their larger dimension
positioned in the direction perpendicular to the motion to minimize the lateral
motion. The primary system mass, M; is roughly equal to the generalized mass, for
the fundamental vibration mode, of a 20 m high circular steel stack having an outside
diameter of 1 m and a 4 mm wall thickness. The primary system has about the same
natural frequency as the steel stack. Table 1 lists the dynamics properties of the
primary system.
A pendulum was connected to the primary system through a rotational bearing.
The wire rope springs were placed between the pendulum and reaction beams rigidly
connected to the interior tall frame. Two arrangements for springs were considered,
either placing one spring on one side (see Fig. 1(a)) or placing two springs, one on
each side (see Fig. 1(b)). All the springs were provided with visco-elastic stops to
prevent excessive deformations. Table 2 lists the mass ratios, m; where m ¼ ma =M
and ra ratios where ra ¼ Ia =ma z2 ; Ia is the mass moment of inertia of the auxiliary
system about the pivot point, ma is the auxiliary mass and z is the distance between
the pivot and the auxiliary system center of mass as shown in Fig. 2(a).

2.1. Wire rope springs

Six wire rope springs were used in this experimental study. A schematic drawing of
a wire rope spring is shown in Fig. 2(b). Two wire rope diameters, dr ; were used. The
coil diameter, Dc ; was kept constant while the number of coils, nc ; were varied to
obtain equal stiffness values for springs #1 and 3. Springs #1, 2a and 2b are identical
except that nc for springs #2a and 2b is half of that of spring #1 and therefore,
springs #2a and 2b each possesses half of the stiffness value of spring #1. The same
applies for springs #3, 4a and 4b corresponding to springs #1, 2a and 2b,
respectively. The dimensions and the properties of the wire rope springs are listed in
ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1367

Fig. 1. Photos of the test setup. (a) TDOF system with single-sided wire rope spring arrangement and
(b) close up on double-sided wire rope spring arrangement.

Table 1
Primary system dynamic properties

M (kg) fn (Hz) b ratio

331.0 3.220 0.0012


ARTICLE IN PRESS
1368 R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385

Table 2
Auxiliary system properties

m ratio ra ratio h (mm)

0.1156–0.1168 1.099–1.080 866.8, 930.3, 993.8


0.0341–0.0354 1.519–1.417 279.4, 304.8, 333.4

x(t)
β

h
θ z
M
+ ma, Ia
K

..
x g (t)
(a)

Dcd+ dr

Roll

Tension-Compression

(Dc + dr) / 2
dr Roll

(b)
Fig. 2. Modeling of structure with a wire rope spring TMD. (a) SDOF system equipped with wire rope
spring TMD and subjected to base acceleration and (b) close-up on a wire rope spring.
ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1369

Table 3
Dimensions of wire rope springs

Spring # nc Dc (mm) dr (mm) Dc =dr ratio Wire rope structure

1 2 150.0 12.70 12.20 6  19 IWRC


2a 1 152.0 12.70 11.93 6  19 IWRC
2b 1 150.0 12.70 11.81 6  19 IWRC
3 10 150.0 7.938 18.90 7  19 WSC
4a 5 149.8 7.938 18.87 7  19 WSC
4b 5 150.5 7.938 18.96 7  19 WSC

Table 3. All wire ropes used are AISI (American Iron and Steel Institute) regular
right lay stainless steel type 316.
The dynamic characteristics of the springs were determined from free vibration
tests. For each representative amplitude, Xr ; the equivalent viscous damping ratio,
bwr ; was determined from the logarithmic decrement of the velocity between the two
points of zero potential energy. The effective stiffness, kwr ; was determined from the
observed period of vibration. In other words, at a given Xr ; bwr results in the same
amount of energy dissipated per cycle as the actual damping mechanism and kwr
yields the same natural frequency as the non-linear spring. A detailed description of
the free vibration tests and the analysis procedures can be found elsewhere [1]. For
each spring, kwr values were normalized by kmin ; where kmin is the minimum value for
kwr : Also, Xr values were normalized by the value corresponding to kmin ; which is
Xrmin : For all springs, normalized kwr values have similar variation with normalized
Xr while bwr values are larger for springs with dr ¼ 7:938 mm as shown in Fig. 3. kwr
decreases rapidly with increasing Xr up to about 0.25Xrmin ; after which, it varies
slightly with Xr : bwr was found to be proportional to 1=Xr +constant, which
represents a combined friction and rate-independent linear damping mechanism. The
springs were originally designed to have equal kmin values (adding kmin for springs
#2a and 2b and for springs #4a and 4b). From testing, it was found that the four kmin
values had a coefficient of variation of about 4%.

2.2. Optimum linear TMD

Fig. 4(a) shows a mixed (translational and rotational) TDOF system. The non-
dimensional complex frequency response functions for the primary system
displacement, Hx ðigÞ; and the auxiliary system rotation, Hy ðigÞ; were derived, for
small rotations, in terms of the non-dimensional parameters ra ; fr ; b; ba ; m and g [3]
and are given in Table 4. fr is the frequency ratio between the auxiliary system and
the primary system (oa =on ), b is the damping ratio of the primary system, ba is the
damping ratio of the auxiliary system and g is the frequency normalized by the
natural frequency of the primary system (o=on ).
Vickery and Davenport [4] introduced the concept of the equivalent viscous
damping and provided design charts for several mass ratios. By equating the
variance of the response of a single-degree-of-freedom (SDOF) system equipped with
ARTICLE IN PRESS
1370 R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385

2.0
spring # 1 spring # 2a
spring # 2b spring # 3
spring # 4a spring # 4b
1.8 fit

1.6
k wr / k min

1.4

1.2

1.0

0.8
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Xr / Xrmin

40

35

30

25
β wr (% critical)

20

15

10

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
X r / X rmin

Fig. 3. Effective stiffness and equivalent viscous damping ratio of wire rope springs.

a TMD to that of a SDOF system with a viscous damping, one can obtain the
viscous damping ratio equivalent to the TMD effect, be (see Fig. 4(b)) as
p
be ¼ R N : ð1Þ
4 0 jHx ðigÞj2 dg
ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1371

x(t)
β

βa h
θ z
M
ka +
K ma, Ia
..
x g (t)
(a)
x(t)
βe

K
..
x g (t)
(b)
Fig. 4. Modeling of structure with a linear TMD. (a) Mixed TDOF system subjected to base acceleration
and (b) equivalent SDOF system subject to base acceleration.

Further, the ratio of the primary system RMS displacement to the auxiliary system
RMS displacement at z is given by
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
RN
*
yz jHy ðigÞj2 dg
R ¼ ¼ R0N : ð2Þ
x* 0 jHx ðigÞj dg
2

General expressions for be and R for any fr and ba are given in Table 4.
Expressions for the optimum frequency ratio, fropt ; and the optimum auxiliary
system damping ratio, bopt a ; that minimize the primary system displacement were
derived as functions of m and ra [3] where
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
1 þ m 1  ð3=2ra Þ
fr D
opt
ð3Þ
1þm
and
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  ffi
m þ m2 1  ð5=4ra Þ
a D
bopt : ð4Þ
4ra þ 2mð4ra  3Þ þ 2m2 ð2ra  3Þ
Substituting for fropt and bopt a in the expressions for be and R; given in Table 4,
results in
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 m
be D
opt
   þ 0:75b ð5Þ
4 ra þ m ra  54 ð1 þ mÞ3
1372
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385
Table 4
Response parameters under base acceleration

g2 ðra þ m  ra mÞ þ igð2fr ra ba  2fr ra ba mÞ þ ðfr2 ra  fr2 ra mÞ

ARTICLE IN PRESS
Hx ðigÞ ¼
g4 ðra  m  ra mÞ  ig3 ð2ra b þ 2fr ra ba þ 2fr ra ba mÞ  g2 ðra þ fr2 ra þ fr2 ra m þ 4fr ra bba Þ þ igð2fr2 ra b þ 2fr ra ba Þ þ ðfr2 ra Þ

igð2bÞ þ ð1Þ
Hy ðigÞ ¼
g4 ðra  m  ra mÞ  ig3 ð2ra b þ 2fr ra ba þ 2fr ra ba mÞ  g2 ðra þ fr2 ra þ fr2 ra m þ 4fr ra bba Þ þ igð2fr2 ra b þ 2fr ra ba Þ þ ðfr2 ra Þ

ra ð1 þ mÞ2 ðfr4 ra bba Þ þ ð1 þ mÞð2fr2 ra bba Þð2fr bba þ 2b2a  1Þ þ ðfr mÞðfr2 b2 þ 2fr bba þ b2a Þ þ ðra bba Þð4fr2 b2 þ 4fr bba þ 1Þ
be ¼ 4
ð1 þ mÞ ðfr4 r2a ba Þ þ ð1 þ mÞ3 ð2fr2 r2a ba Þð2fr bba þ 2b2a  1Þ þ mð1 þ mÞ2 ðfr2 ra Þðfr b þ 3ba Þ þ ð1 þ mÞ2 ðr2a ba Þð4fr2 b2 þ 4fr bba þ 1Þ  mð1 þ mÞð2ra ba Þ þ m2 ðfr b þ ba Þ


sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1 þ mÞ2 ðfr3 ra ba Þ þ ð1 þ mÞð4fr2 ra ba Þðfr b þ ba Þ þ ðfr mÞðfr b þ ba Þ þ ðra bÞð4fr2 b2 þ 4fr bba þ 1Þ
fr ra ð1 þ mÞ ðfr ra ba Þ þ ð1 þ mÞ ð2fr2 r2a ba Þð2fr bba þ 2b2a  1Þ þ mð1 þ mÞ2 ðfr2 ra Þðfr b þ 3ba Þ þ ð1 þ mÞ2 ðr2a ba Þð4fr2 b2 þ 4fr bba þ 1Þ  mð1 þ mÞð2ra ba Þ þ m2 ðfr b þ ba Þ
3 4 4 2 3
ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1373

Table 5
Auxiliary system and response optimum parameters

fropt bopt
a bopt
e
Ropt

0.8751–0.8799 0.1527–0.1575 0.06794–0.07016 2.297–2.376


0.9651–0.9669 0.07430–0.07723 0.03570–0.03709 3.218–3.360

and
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2½ð2ra þ mð2ra  1Þ ð1 þ mÞ2
Ropt D : ð6Þ
m½4ra þ mð4ra  5Þ ½2ra þ mð2ra  3Þ
For the range of m and ra values used in this study, values for fropt and bopt
a are listed
in Table 5 together with the corresponding bopte and R opt
values.

2.3. Amplitude-dependent tuning and response parameters

The amplitude-dependent wire rope spring effective stiffness, kwr ; and equivalent
viscous damping ratio, bwr ; were fitted with mathematical expressions. The auxiliary
system rotational stiffness, kay ; is the summation of the wire rope spring rotational
stiffness and the pendulum rotational stiffness as
kay ðXr Þ ¼ kwr ðXr Þh2 þ ma ag z; ð7Þ
where ag is the acceleration of gravity. The frequency ratio between the auxiliary
system and the primary systems is
sffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffi
kay ðXr Þ K
fr ðXr Þ ¼ = : ð8Þ
Ia M
The auxiliary system damping ratio, ba ; can be determined based on dissipated
energy equivalence as
kwr ðXr Þh2
ba ðXr Þ ¼ bwr ðXr Þ : ð9Þ
kay ðXr Þ
Expressions for be and R for any fr and ba ; given in Table 4, are used to describe be
and R as amplitude dependent.
Values for fr ; ba ; be and R; each normalized by the corresponding optimum value,
are shown in Figs. 5 and 6 for two mass ratios and three spring attachment positions.
The representative stiffness ratio is kopt /kmin where kopt is the optimum spring
stiffness based on fropt and is given by
mz
kopt ¼ 2 ððfropt Þ2 ra Kz  Mag Þ: ð10Þ
h
If the representative stiffness ratio is unity then fr is equal to fropt at single amplitude,
which is Xrmin : For stiffness ratio greater than unity, fr is equal to fropt at two
amplitudes. However, as the stiffness ratio increases the separation between the two
ARTICLE IN PRESS
1374 R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385

2.5
rep. stiffness ratio = 1.046
rep. stiffness ratio = 1.185
2.0 rep. stiffness ratio = 1.360

1.5 fr
all positions βa
f ropt
Ratio

β opt
a

1.0

0.5

0.0
0.0 0.5 1.0 1.5 2.0

Xr /X rmin

1.50
R
R opt
1.25

1.00
Ratio

0.75

0.50

βe
opt
0.25 βe

0.00
0.0 0.5 1.0 1.5 2.0
Xr /X rmin

Fig. 5. Normalized amplitude-dependent parameters vs. normalized relative displacement amplitude for
spring #1 and m ¼ 0:1160:

amplitudes increases and the larger amplitude may exceed the operating range of the
wire rope spring. It can be noticed from Figs. 5 and 6 that the variation of fr with Xr
is much less than the variation of kwr with Xr : This is due to the fact that the total
stiffness has the pendulum rotational stiffness as a constant component added to the
amplitude-dependent component kwr : Further, the auxiliary system frequency is the
square root of the total stiffness divided by a constant mass.
ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1375

2.5
rep. stiffness ratio = 1.150
rep. stiffness ratio = 1.365
2.0 rep. stiffness ratio = 1.615

all positions βa
1.5 β opt
a
Ratio

1.0

0.5 fr
f ropt

0.0
0.0 0.5 1.0 1.5 2.0
Xr /X rmin

1.50

1.25
βe R
βopt
e
R opt
1.00
Ratio

0.75

0.50

0.25

0.00
0.0 0.5 1.0 1.5 2.0
Xr /X rmin

Fig. 6. Normalized amplitude-dependent parameters vs. normalized relative displacement amplitude for
spring #1 and m ¼ 0:0345:

For m ¼ 0:1160 ‘‘heavy case’’ (see Fig. 5), bopt


a approximately synchronizes with
resulting in peaks in be approaching bopt
fropt e : However, as Xr increases, ba becomes
considerably smaller than bopt
a and therefore be values decrease and R ratios increase.
For m ¼ 0:0345 ‘‘light case’’ (see Fig. 5), ba approaches its optimum value at a
normalized Xr about unity and thus ba is considerably larger than bopt
a for the lower
ARTICLE IN PRESS
1376 R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385

half of the operating amplitude range (Xr =Xrmin p1) and slightly smaller for the
upper half (Xr =Xrmin > 1). This resulted in smaller normalized R ratios compared to
the heavy case. A double peak trend for be is observed. The first set of peaks took
place at small amplitudes corresponding to near fropt but ba much larger than bopt a :
The second set of peaks occurred at large Xr values (only the one for 1.150
representative stiffness ratio can be seen in Fig. 6) corresponding to near fropt and ba
slightly smaller than bopt
a :

3. Experimental study

3.1. Instrumentation

The shaker table motion was measured by two accelerometers attached to two
corners of the table and their outputs were almost identical. Two laser transducers
were attached to two sides of the exterior short frame with targets placed on the
interior tall frame to measure the displacement of the primary system relative to the
shaker table. Comparison between the outputs of the two transducers showed
negligible torsional motion and the displacement was taken as the average of the
two. Another laser transducer was inserted to measure the relative displacement
between the pendulum and the tall interior frame (see Fig. 1).

3.2. Base acceleration

A numerically generated time history of the base acceleration was fed to the shaker
table. The record was based on a limited band white noise spectrum with a Gaussian
probability distribution. The lower and upper frequency limits were 1.5 and 4.5 Hz,
respectively. The total time of the record was an hour with a time step of 0.02 s.

3.3. Test procedures

For each mass ratio, four different arrangements were considered. Spring #1
attached to one side of the pendulum, springs #2a and 2b attached to both sides,
spring #3 treated as spring #1 and springs #4a and 4b treated as springs #2a and 2b.
For each arrangement, three different attachment positions were considered (see
Table 2). For each position, four different base acceleration multiplier values, for the
same time history, of 0.4, 0.6, 0.8 and 1.0 were applied. Some other amplitudes that
were applied in special cases are 0.88 and 1.08. In order to check the reliability of the
results, a number runs were repeated.
The time histories of the base acceleration, primary system displacement and the
auxiliary system relative displacement were recorded. An analogue low pass filter,
with a cut off limit of 10.0 Hz, was used to suppress the noise from the recorded
signals. The sampling rate was chosen to be 50 Hz. Some results could not be
obtained due to large primary system response that produced force on the shaker
ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1377

exceeding the permissible limit. For springs #2a and 2b, the wire rope was damaged
due to excessive deformations after several runs.

4. Results and discussion

4.1. Time histories

Thirty-second records for the base acceleration, the primary system displacement
and the auxiliary system rotation are shown in Fig. 7. The auxiliary system stick-slip
behavior is evident and is identified by red circles.

4.2. Frequency response functions

The frequency response functions for both the primary system displacement and
the auxiliary system rotation were obtained, from experiments, as
sffiffiffiffiffiffiffiffiffiffiffiffiffi
2 Sx ðgÞ
jHx ðigÞj ¼ on ð11Þ
Sx. g ðgÞ
and
sffiffiffiffiffiffiffiffiffiffiffiffiffi
Sy ðgÞ
jHy ðigÞj ¼ o2n z ; ð12Þ
Sx. g ðgÞ

2
Base accel.
(m/s 2)

-2

12
Primary system
disp. (mm)

-12

36
rotation (radx103)
Auxiliary system

-36
30 35 40 45 50 55 60
Time (s)

Fig. 7. Time history for excitation and responses.


ARTICLE IN PRESS
1378 R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385

where Sx ðgÞ; Sy ðgÞ and Sx. g ðgÞ are the spectra of primary system displacement,
auxiliary system rotation and base acceleration excitation, respectively, and on ¼
2pfn : The response functions are shown in Fig. 8 and are dependent on the excitation
level, which is a typical feature of a non-linear system.
The optimum frequency response functions for an equivalent linear TMD were
obtained by substituting with the optimum values for fr and ba in the expressions for
Hx ðigÞ and Hy ðigÞ; given in Table 4, and are plotted in Fig. 8 for comparison.

10
µ = 0.1160 optimum linear
8 ra = 1.095 0.4 multiplier
rep. stiff. ratio = 1.046
Hx (ig)

6 0.6 multiplier
0.8 multiplier
4
1.0 multiplier
2

25

20
Hθ (ig)

15

10

25
µ = 0.0345
20 ra = 1.500
rep. stiff. ratio = 1.365
Hx(ig)

15

10

40

32
Hθ (ig)

24

16

0
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4

Fig. 8. Frequency response functions for spring #1.


ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1379

For the heavy case, the dependency of the frequency response functions on
the excitation level is low once a certain level of excitation is exceeded
(multiplier=0.40 as shown in Fig. 8). This can be explained, as the level of the
excitation increases, Xr becomes in the range where the frequency ratio, fr ; and
the auxiliary system damping ratio, ba have less variation with amplitude, about
their optimum values, and hence the frequency response functions approach those of
an optimum linear system.
The dependency on the excitation level is more pronounced for the light case. This
can be attributed to the high sensitivity to frequency tuning with small mass ratios
and the considerable variation of ba with Xr for the range Xr =Xrmin o1 (see Fig. 6).

4.3. Equivalent viscous damping ratio, be

A measure for the efficiency of a non-linear TMD is the ratio of the equivalent
viscous damping provided by such a system, be ; to that of an optimized linear
system, bopt
e ; which has the same auxiliary mass parameters, m and ra : be is given by
Eq. (1) and the trapezoidal role was employed to evaluate the integral numerically.
Values for the normalized equivalent viscous damping ratio, be /bopt e ; are listed in
Table 6 and are plotted in Fig. 8 for spring #1.
Table 6 includes be /bopte values for four spring arrangements, two mass ra-
tios, three attachment positions and multiple excitation levels. Some of the values
slightly exceeded unity for large excitations (0.8 and 1.00 multipliers), which may be
due to the damping provided by the visco-elastic stops. By comparing the re-
sults of the repeated runs (marked with  in Table 6) excellent reproducibility is
observed.
For all the cases listed in Table 6, the heavy auxiliary mass resulted in be /bopt e
ratios larger than those of the light mass. Single- and double-sided spring
arrangements resulted in comparable results as can be observed by comparing
spring #1 with springs #2a and 2b and spring #3 with springs 4a and 4b. However,
by comparing spring #1 with spring #3 and springs #2a and 2b with springs #4a and
4b, one can show that springs with smaller dr ; which possess the same Dc and kwr
produce considerably smaller damping for low excitation levels. This is attributed to
larger bwr associated with smaller dr (see Fig. 3).
Contours of be /bopt
e for spring #1 are plotted in Fig. 9 versus the representative
stiffness ratio and the excitation multiplier. For small excitations, the damping is
higher for larger stiffness ratios, at which, fr is closer to optimum (see Figs. 5 and 6).
As the excitation increases, the damping also increases since fr has less variation with
Xr about its optimum value. The auxiliary system damping also plays a role in
determining the efficiency of the TMD. As ba approaches its optimum value, be /bopt e
increases. Relatively lower be /bopt
e for the light mass case under small excitations, at
which fr is optimum, is due to considerably large ba (see Fig. 6).
Since the TDOF system was excited randomly, the response included many
amplitudes, and therefore, be cannot be based on single relative displacement
amplitude but rather averaged over a range of amplitudes [2,5]. Analytical
Xr =Xrmin  be /bopt
e relationships (see Figs. 5 and 6) can only be compared with
1380
Table 6
Response parameters under different arrangements and excitations

R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385
h (mm) Spring #1 Springs #2a and 2b

ra ratio m ratio k opt


=k min Excitation be =bopt
e R=R opt ra ratio m ratio kopt =kmin Excitation be =bopt
e R=Ropt
multiplier multiplier

993.8 1.095 0.1160 1.046 0.40 0.7708 0.6649 1.084 0.1168 0.945 0.40 0.6053 0.5479
0.60 1.068 1.119 0.60 0.8446 0.8708

ARTICLE IN PRESS
0.80 1.062 1.211 0.80 0.9450 1.119
1.00 1.122 1.260 1.00 1.046 1.247
930.3 1.097 0.1160 1.185 0.40 0.9562 0.8390 1.087 0.1168 1.071 0.40 0.9205 0.8325
0.60 1.053 1.104 0.60 1.039 1.102
0.80 1.095 1.234 0.80 — —
1.00 — — 1.00 — —
866.8 1.099 0.1160 1.360 0.40 1.030 0.9786
0.60 0.9342 1.107
0.80 0.9782 1.217
0.80 0.9706 1.219
333.4 1.479 0.0345 1.150 0.40 — —
0.60 0.3872 0.3089
0.80 0.5965 0.5038
1.00 0.7907 0.6716
304.8 1.500 0.0345 1.365 0.40 — —
0.60 0.5114 0.3868
0.80 0.7334 0.5584
1.00 0.8420 0.6405
279.4 1.519 0.0345 1.615 0.40 — —
0.60 0.5791 0.4377
0.80 0.6934 0.5791
1.00 0.6503 0.6447
1.00 0.6757 0.6572
Spring #3 Springs #4a and 4b

ra ratio m ratio kopt =kmin Excitation be =bopt


e R=Ropt ra ratio m ratio kopt =kmin Excitation be =bopt
e R=Ropt
multiplier multiplier

R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385
993.8 1.080 0.1156 1.020 0.60 0.5513 0.4826 1.084 0.1168 0.9990 0.40 0.3521 0.2941
0.80 0.8082 0.7113 0.60 0.4302 0.3864
1.00 0.9642 0.8986 0.80 0.6520 0.5924
1.08 1.003 0.9543 1.00 0.8726 0.8296
930.3 1.140 0.1156 1.130 0.40 0.4868 0.4151 1.087 0.1168 1.130 0.40 0.4810 0.4111
0.60 0.8107 0.6791 0.60 0.7002 0.5950
0.80 1.012 0.9002 0.80 0.9181 0.8109

ARTICLE IN PRESS
1.00 1.103 1.053 1.00 1.048 0.9970
866.8 1.084 0.1156 1.324 0.40 0.6099 0.5079 1.089 0.1168 1.297 0.40 0.7755 0.6324
0.60 0.9555 0.8077 0.60 0.9544 0.8046
0.80 1.028 0.9824 0.80 1.059 0.9741
1.00 1.066 1.101 1.00 1.078 1.096
1.00 1.062 1.105
333.4 1.417 0.0341 1.123 0.40 — — 1.442 0.0354 1.101 0.40 — —
0.60 — — 0.60 — —
0.88 0.2794 0.2272 0.88 0.2625 0.2191
0.88 0.2526 0.2058 1.00 0.3059 0.2693
304.8 1.440 0.0341 1.330 0.40 — — 1.466 0.0354 1.305 0.40 — —
0.60 — — 0.60 0.2148 0.1658
0.80 — — 0.80 0.3374 0.2652
1.00 0.4565 0.3696 1.00 0.4612 0.3602
279.4 1.461 0.0341 1.573 0.40 — — 1.489 0.0354 1.540 0.40 — —
0.60 — 0.60 — —
0.80 0.4603 0.3474 0.80 0.4089 0.3100
1.00 0.5766 0.4401 1.00 0.5660 0.4273
1.00 0.5739 0.4343 1.00 — —

1381
ARTICLE IN PRESS
1382 R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385

0.8 1
1.1

8
1.0

9
1.0

0.9
0.75

5
1.0
0.7

6
1.08

0.9
Excitation multiplier

0.65

9
1.0

0.9
0.6 1.05
1.05
1.02 0.
96
0.55 0
.99 1.02
0.9 0.
0.5 0.9 6 99
3 0.99

0
0.45 0. .9
0.8 87 0.9
0.8 4 0.9 6 0.9
1 3 9
0.4
1.05 1.1 1.15 1.2 1.25 1.3 1.35
Representative Stiffness Ratio

0.69
0.81
75
0.

0.95 0.7 0.78


5
78

2
0.7
0.

0.9
0.7
0.6 2
Excitation multiplier

0.6 9
0.85 6 0.7
5
0.6
3
0.8 0.6 0.72
0.69
0.69
0.6
0.75 6
0.5 0.6 0.66
7 3
0.7 0.5 0.6
1 0.63
0.5
4
0.6
0.65 0.4 0.57
0.4 8
0.4 5 0.5
2 1 0.54 0.57
0.6
1.15 1.2 1.25 1.3 1.35 1.4 1.45 1.5 1.55 1.6

Representative Stiffness Ratio

Fig. 9. Normalized equivalent viscous damping ratio contours for spring #1 (upper for m ¼ 0:1160 and
lower for m ¼ 0:0345).

experimentally established contours (see Fig. 9) in a qualitative sense. Calculated be


reflected some of the trends observed experimentally. For example, the normalized
damping increases with the increase of the representative stiffness ratio for low
excitation levels but decreases for large excitations. Also, higher be /bopt
e for the larger
mass ratio.
ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1383

4.4. Auxiliary to primary RMS displacement ratio, R

Another measure for the efficiency of a non-linear TMD is the ratio of the RMS
displacement ratio achieved by such a system, R; to that of a similar optimized linear
0.8

1.2
1.2
0.75 1.2

0.7

1.15 1.15 1.15


Excitation multiplier

0.65

0.6 1.1 1.1 1.1


1.05
0.55 1 1.05
0. 95 1 1.05
0.5 0.9
0.95
0.8 1
5
0.45 0.8 0.9
0.7 0.95 1
5
0.7 0.8
5
0.4
1.05 1.1 1.15 1.2 1.25 1.3
Representative stiffness ratio

1
0.66
0.63
0. 63 0. 63
0.95

0.6
0.9 0. 6
0.6
0. 57
Excitation multiplier

0.85
0.54 0.57

0.8 0. 57
0.54
0.51
0.54
0.75
0.48 0.51
0.42 0.51
0.7 0.45 0.48
0.39
0.45 0.48
0.42
0.65 0.36
0.39 0.45
0.33 0. 42
0.6
1.15 1.2 1.25 1.3 1.35 1.4 1.45 1.5 1.55 1.6
Representative stiffness ratio

Fig. 10. Normalized RMS displacement ratio contours for spring #1 (upper for m ¼ 0:1160 and lower for
m ¼ 0:0345).
ARTICLE IN PRESS
1384 R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385

system, Ropt : R is given by Eq. (2) and was integrated numerically by applying the
trapezoidal role. Values for the normalized RMS displacement ratio, R=Ropt ; are
listed in Table 6 and are plotted in Fig. 10 for spring #1.
For all the cases listed in Table 6, the heavy mass case resulted in considerably
larger R=Ropt values in comparison with those of the light mass case. Single- and
double-sided spring arrangements resulted in similar results. However, springs with
smaller dr maintaining similar Dc and kwr produce smaller R values. This is due to
larger bwr characterizing smaller dr springs (see Fig. 3).
Contours of R=Ropt for spring #1 are plotted in Fig. 10 versus the representative
stiffness ratio and the excitation multiplier. For small excitation levels, R values are
larger for large stiffness ratios since the TMD frequency is reasonably tuned (see
Figs. 5 and 6). As the excitation increases, the R values become less dependent on the
stiffness ratio. For large excitation levels, the R values show almost no-dependency
on the stiffness ratio. The auxiliary system damping also plays a role in determining
the R values. Considerably small R values for the light mass case under small
excitations are partially due to large ba values (see Fig. 6).
Calculated R=Ropt ratio (see Figs. 5 and 6) follows some of the trends observed
experimentally (see Fig. 10). For example, the normalized R values are considerably
less dependent on the amplitude for large Xr =Xrmin values. Also, higher R=Ropt for
the larger mass ratio.

5. Conclusions

The wire rope spring is a feasible economical alternative for replacing both the
spring and the damper typically needed in a TMD system. The damping provided by
the non-linear system approaches that of an optimum linear TMD as the mass ratio
increases. Also, the sensitivity of the added damping to the excitation level decreases
with the increase of the mass ratio. The auxiliary system relative displacement is
smaller than that of an equivalent optimum linear system for small mass ratios,
about 0.035, but slightly larger for large mass ratios, about 0.11.
Single- and double-sided spring arrangements produce similar results. Systems
with smaller wire rope diameter but similar coil diameter and stiffness provide more
damping to the auxiliary system. This resulted in reducing both the added damping
and the auxiliary system relative displacement. For possible extremely high level of
excitation, adding a visco-elastic stop is beneficial since it protects the spring from
excessive deformations, attenuates the impact force on the structure and forms a
secondary energy dissipation mechanism.

Acknowledgements

The Natural Sciences and Engineering Research Council of Canada and Rowan,
Williams, Davis and Irwin Inc. have supported the initial stage of this work
through the Industrial Postgraduate Scholarships program. The Ontario Ministry of
ARTICLE IN PRESS
R.R. Gerges, B.J. Vickery / J. Wind Eng. Ind. Aerodyn. 91 (2003) 1363–1385 1385

Education continued the support through the Ontario Graduate scholarships


program.

References

[1] R.R. Gerges, B.J. Vickery, J.K. Galsworthy, Experimental investigation of the dynamic characteristics
of wire rope coils, Proceedings of the 29th Annual Conference of the Canadian Society for Civil
Engineering, Victoria, BC, Canada, 2001.
[2] R.R. Gerges, B.J. Vickery, Wind tunnel study of the across-wind response of a slender tower with a
nonlinear tuned mass damper, J. Wind Eng. Ind. Aerodyn., accepted for publication.
[3] R.R. Gerges, B.J. Vickery, Optimum design of pendulum-type tuned mass dampers, J. Eng. Mech.
ASCE, submitted for publication.
[4] B.J. Vickery, A.G. Davenport, An investigation of the behaviour in wind of the proposed centrepoint
tower in Sydney, Australia, Engineering Science Research Report No. BLWT-1-70, The University of
Western Ontario, Faculty of Engineering Science, The Boundary Layer Wind Tunnel Laboratory,
London, Ont., Canada, 1970.
[5] B.J. Vickery, J.K. Galsworthy, R.R. Gerges, The behavior of simple nonlinear tuned mass dampers,
Proceedings of the Sixth World Congress of the Council on Tall Buildings and Urban Habitat,
Melbourne, Australia, 2001.

Вам также может понравиться