Seventh International Congress on Advances in Civil Engineering, October11-13, 2006
Yildiz TechnicalUniversity, Istanbul, Turkey
Tuned-Mass Systems for the Seismic Retrofit of Buildings
Peter Nawrotzki
GERB Vibration Control Systems, Berlin/Essen, Germany
Abstract
Passive seismic control strategies are based on the reduction of energy, which affects a
structure in case of earthquake events. Some well known approaches make use of
frictional, plastic or other energy dissipating behaviour of special devices. The
following presentation reflects some special ideas for the increase damping in order to
improve the seismic performance of buildings. For this purpose additional-mass systems
are proposed and their performance is investigated theoretically as well as on the
shaking table. Usually these systems are considered as not suitable for seismic
applications, but this thesis is no more valid as a general rule, if certain design
approaches are kept. Tuned-Mass Control Systems (TMCS) can be used to control the
displacements, accelerations and internal stress variables of a structure in case of
earthquakes. The safety against collapse and defined states of serviceability of the
structures can be achieved. This system can also be used for the seismic retrofit of
existing buildings as the inside of the structure is usually not objective to modification.
Hence, the usual operation inside the building may go on during the upgrade activities.
Keywords: Earthquake Protection, Passive Control, Damping, Tuned-Mass Systems,
Seismic Retrofit.
Introduction
A well accepted strategy in utilizing seismic control systems is based on the increase of
structural damping. As a first idea damping devices can be installed solely. Then, they
have the task to damp the relative motion between two structures, two parts of the same
structure, or the structure and the ‘rigid’ vicinity. The damping effects may be obtained
by friction, plastic deformation or viscose behaviour inside the device. The entire
improvement of the seismic performance becomes obvious by different national and
international standards. Some well known curves are compared, and Fig. 1 provides an
idea of possible control effects. Usually 5 % of critical damping can be assumed for
buildings, and an increase of the damping ratio causes a reduction of the stress or
acceleration response as indicated by the correction factor ξ. As an example the increase
from 5 to 20 % of critical damping would cause a reduction of the induced seismic
responses by about 50 % according to the Japanese provisions (see Fig. 1).
1,4
Eurocode 8
1,3
Uniform Building Code 97
1,2
Taiwan Building Code
Correction Factor ξ
1,1 Architectural Institute Japan
1,0 IEEE Std 693-1997
0,9
0,8
0,7
0,6
0,5
0,4
0 5 10 15 20 25 30
Damping in %
Figure 1 Seismic control effects depending on structural damping.
Tuned-Mass Damper Systems (TMD) are widely used for the reduction of vibration
caused by wind and traffic like pedestrians or railway trains. Typical structures like
slender bridges, stacks, high and slender buildings possess low levels of damping and
may therefore undergo unacceptable vibration. TMDs cause control effects which are
similar to the increase of damping. Depending on the mass ratio, the tuning frequency
and the damping capability the amplitude reduction can be very significant and achieve
values of about 10 to 20 % of the figures without TMD. The reduction effects in these
applications are higher that in case of seismic events because the governing vibration is
similar to stationary motions and the TMD gets better adjusted to the motion.
Nevertheless significant reduction effects can also be observed for seismic excitation.
The ideas of the improvement of seismic performance according to Fig. 1 can be
confirmed by theoretical and practical investigations. In order to distinguish between
ordinary Tuned-Mass Systems and those for seismic applications the expression Tuned-
Mass Control Systems (TMCS) is used. The layout of such systems is slightly different
from that for a usual TMD system. Here, the mass and tuning ratio as well as the
damping are chosen according to different criteria.
A typical situation for structures is shown in Fig. 2. Here, a multi-storey building is
equipped with a tuned-mass system on the rooftop. The additional mass consists of
reinforced concrete and rests on helical steel spring devices with integrated dampers.
Figure 2 Typical tuned-mass system at the top of a building.
2
Numerical Investigations
Numerical simulations of buildings under earthquake with tuned-mass systems have
frequently been performed. In many cases a special building model is taken and the
additional mass is connected with the building elastically; sometimes the mass ratio is
varied. Then, different recorded earthquakes are run and the responses of the structure
with and without tuned mass are compared. The obtained results are usually not
showing a unique picture. It can be concluded from this procedure that the tuned mass
improves the response behaviour for most of the investigated cases, but there are also
models under seismic excitation without significant improvement. In all of the latter
cases without significant difference the structural response without TMCS turned out
not to be dangerous for the building. The reasons are the induced internal forces and
acceleration responses which are at a low level without any need for further reduction.
In these cases the governing natural frequencies are not excited. The described steps for
the layout of a tuned-mass system do not reflect the required procedure for real projects!
For real projects there is a building with columns, beams, frames, walls, floors, and
other important members. The structure consists of certain materials, possesses certain
dimensions and there is a certain mass or mass distribution, stiffness, ductility and many
other mechanical parameters. On the other hand there is the seismic risk which can be
described with statistical parameters. The most suitable representation for engineering
purposes can be seen in a site specific response spectrum. Here, for instance, we can
directly see whether the building is in the dangerous frequency range and furthermore
we can derive artificial base-excitation functions which correspond to the project site.
Also recorded seismic events can be taken for the layout of the tuned-mass of a real
structure, but in these cases the acceleration-time histories have to be scaled according
to the site specific response spectrum.
In Fig. 3 two examples for the numerical modeling of buildings with TMCS are given.
On the left hand side the first mode of an RC structure is shown with a combined lateral
force resisting system. The tuned-mass is modeled as a single mass which is elastically
connected to the center of the rooftop. A braced steel frame structure is shown on the
right hand side of Fig. 3. There are four RC blocks arranged in a manner that the dead
load for a single center column is increased by about 6 % of the TMCS dead load only.
Figure 3 FE structures with TMCS on the roof - RC and steel frame building.
3
0,20
Without TMCS
0,15 With TMCS
Hor. Top Displacement [m]
0,10
0,05
0,00
-0,05
-0,10
-0,15
-0,20
0 5 10 15 20
Time [s]
Figure 4 Typical time-history of displacement responses with / without TMCS.
Fig. 4 shows the performance of the structural response improved by a TMCS. The
original response is given as a red curve and the induced peak responses are reduced by
about 40 % by only activating the tuned mass (blue curve). It becomes obvious that the
mass has a significant influence already at an early phase of the ground motion. The
TMCS causes an increase of damping for the structure and this can also be seen in the
displacement-time history starting at about 10 s. The residual motion after the strong
motion phase is significantly damped out as the amplitude becomes nearly zero after
about 15 s. The unprotected structure has still amplitudes of about ±100 mm in this time
domain. Having a look at Fig. 1 we can conclude an increase of the damping ratio from
5 to 15 % when the AIJ regulation is taken as a basis. Assuming the UBC 97 we can
even derive a damping ratio of about 25 % in case of the activated TMCS.
The TMCS significantly reduces the top storey displacements, inter-storey drifts,
response accelerations and consequently induced internal stress responses due to
earthquakes. The corresponding effects of the performance with TMCS can also be
shown in modified phase diagrams (Fig. 5). On the left hand side the original response
of an RC structure is given assuming a damping ratio of 5 %. By activating a mass of
less than 2 % of the total mass of the building, the response curves change significantly.
On the right hand side of Fig. 5 the improvement can be described in terms of
displacement and acceleration values. Of course, the tuning frequency plays a role in the
improvement of the seismic performance as well as the choice of the critical damping.
Optimum values can usually be found in the range of 5 to 20 % also depending on the
target motion values. The damping broadens the working frequency band of the TMCS.
Acceleration [m/s²]
Acceleration [m/s²]
12 12
8 8
4 4
0 0
-4 -4
-8 -8
-12 -12
-100 -75 -50 -25 0 25 50 75 100 -100 -75 -50 -25 0 25 50 75 100
Displacement [mm] Displacement [mm]
Figure 5 Structural behaviour without (left) and with TMCS (right) in the phase plane.
4
Experimental Investigations
In parallel to numerical investigations it is always important to compare the results with
experimental data. For this purpose shaking table tests have been performed at IZIIS,
Skopje, Macedonia. The geometry of the tested steel frame model is given in Fig. 6. The
columns as well as the beams are made of steel hollow profile appropriately welded at
the joints. The structure in the central span has a special bracing substructure, which is
used for modeling the stiffness and damping needed for the test. In the orthogonal
direction the frame model has only one span of 1.5 m, but by adding a bracing the
structural stiffness in that direction is increased several times. The height of each floor
is 0.75 m, while all three spans in the longitudinal direction amounts to 1.5m each. The
total mass of the tested structure is 19.0 t, and the mass of the TMCS is 260 kg which
corresponds to about 1.3 % of the entire mass. The total mass of 19.0 t has been
obtained by adding steel blocks on each floor without any influence on the stiffness.
SPRINGS TMCS
& DASHPOTS
LEVEL 05
LEVEL 04
LEVEL 03
REFERENT BEAM
LEVEL 02
DEAD LOAD
LEVEL 01
SG SG
SG SG SG
LEVEL 00
BIAXIAL SHAKING TABLE
Figure 6 Tested model on the shaking table with position of the TMCS.
The performance of the frame with and without TMCS has been studied by simulating
ten different earthquakes records. Eight of them are recorded during the real earthquake
motion, and two are representing artificial earthquake time histories. Among them are
Northridge, Kobe, Mexico City, Vrancea (Romania), Izmit (Turkey). The effectiveness
of the TMCS has been estimated based on the difference of time responses of the tested
model in terms of accelerations, displacements and strain measurements for
identification of axial loads and bending moments. At first, the model was tested with
unlocked TMCS, and all selected earthquakes were simulated for minimum three
different intensities. The same testing has been repeated with locked TMCS and
response-time histories were recorded for the same measuring points as for the case
with the unlocked TMCS.
From the large number of recorded time histories responses of relative displacements at
fifth level (top of the model) and bending strains recorded at the bottom of the middle
column of the first floor are presented. In Fig. 7 a comparison of time responses of the
mentioned quantities for unlocked (black curve) and locked TMCS (red curve) are
presented for the Turkish earthquake (under reduced intensity). For the locked case the
peak relative top displacement is recorded to be about 14 mm (Fig. 7, left), while the
activated tuned-mass system causes an improvement of performance by nearly 40 %.
5
15 600
Level 5 Middle
Relative Displacement (mm) 10 400
Bending strain 1E-6
5 200
0 0
-5 -200
-10 -400
-15 -600
0 2.5 5 7.5 10 12.5 15 0 2.5 5 7.5 10 12.5 15
Time (s) Time (s)
Figure 7 Recorded responses with activated (black) and locked TMCS (red).
Qualitatively these results correspond very well to the numerical investigation; the
corresponding records are shown in Fig. 4. Here, the same base-excitation function has
been used for a another building and under different seismic intensity. Having a closer
look at the bending strain response of the middle column (Fig. 7, right hand side) the
reduction of peak amplitudes with activated TMCS is even higher than 40 %. It
becomes obvious from Fig. 7 that the efficiency of the TMCS becomes significantly
larger after 10 seconds of ground motion in this case. The dynamics of the tuned mass
becomes better adjusted to the motion of the structure, and thus, the damping effect
becomes even more evident.
More insight into the mechanical behaviour of the structure with activated TMCS can
be derived from Fig. 8. Here the axial force response of the bottom of the middle
column is shown with locked (red) and activated tuned mass (black curve) under IEEE
excitation. It is well known that this artificial earthquake causes the high amplification
of responses for structural frequencies between 1.1 and 8 Hz. As the steel frame
possesses first frequencies at about 1.9 Hz resonance effects can be expected. The axial
force response can well be controlled by the tuned mass as shown in figure 8, left hand
side. The peak values are reduced from more than 10 kN to less than 6 kN. The
response is also expressed in terms of the corresponding FFT spectrum on the right
hand side of Fig. 8. A structural frequency of 1.9 - 2 Hz becomes visible for both cases,
and hence, the additional mass does not cause a significant change of the frequency. The
enormous reduction of the spectral amplitude can be understood by an effect which
corresponds to the increase of the critical damping ratio.
Figure 8 Recorded response of axial force of column and corresponding FFT spectrum.
6
Conclusion
Theoretical and experimental investigations have shown that tuned-mass systems can
well be applied for the control of seismically induced responses. The quantity of the
additional mass has to be chosen according to the target control efficiency; usual values
for these purposes can be found in the range of 1.5 - 4 % of the total building mass.
With an increase of the tuned mass the control effect becomes higher.
Tuned-mass systems are especially suitable for the reduction of seismic drift ratios.
Furthermore the damage of important structural members is reduced and thus, the loss
of stiffness by cracks, plastic zones, etc., becomes significantly less. Fig. 9, left hand
side, shows the typical nonlinear structural behaviour without TMCS in case of an
earthquake. The linear load deflection curve exceeds the limit force for the concerned
structural member which is given with a value of 2 units in this typical example. The
seismic energy which is represented by the remaining triangle (marked in Fig. 9, left)
cannot be taken by the member and hence, it causes plastic deformation. The nonlinear
deformation (energy) is increased by the corresponding portion in the rectangular
section which is also marked in Fig. 9, left side. The resulting total deflection amounts
to a value of 7.25 units where only 2 units are related to linear elastic behaviour.
On the right hand side of Fig. 9 the same energetic considerations are made for the
structure with activated TMCS. The peak forces are reduced by 40 % and hence, the
damage causing energy becomes considerably smaller. The corresponding triangle is
also marked in the sketch. The resulting overall deflection is this case corresponds to
3.25 units where again 2 units are related to the linear elastic behaviour of this member.
The damage causing energy can be reduced from 10.5 to 2.5 units only by activating the
TMCS. Consequently, a remarkable decrease of the structural frequencies by seismic
effects can be avoided with a proper layout of the tuned-mass system.
Fig. 9 Nonlinear seismic effects in structures without (left) and with TMCS (right).
A slight drop of the frequencies in case of seismically induced damage would not yet
cause the malfunction of the TMCS because
a) the tuning of the additional-mass system is usually sub-critical and hence, the
structural frequencies would even approach the optimal frequencies of the TMCS,
b) usually a high percentage of critical damping is used for the relative motion between
structure and additional mass which spreads the working frequency range of the TMCS.
For the seismic upgrade of existing buildings it is also possible to take measurements in
regard to the most important structural frequencies. In this case damage sustained
7
during the past periods as well as soil-structure interaction effects are already included
in the basic layout parameters for the TMCS. Furthermore non-sufficient rigidity or
significant loss of stiffness of the fundamental structure would also become evident by
measurement.
Properly designed tuned-mass control systems can be characterized as follows:
a) They reduce seismically induced responses in terms of displacements, accelerations,
internal stresses and strains as well as subsoil demands.
b) They increase the structural safety. The collapse of a building becomes less probable
and hence, human life is protected.
c) They improve the serviceability of structures. Damage and corresponding repair cost
in case of seismic events are reduced significantly.
d) In comparison to conventional strengthening methods, the building can usually be
under operation during the installation of the TMCS (if no additional measures are
required).
e) Regarding the overall procedure and required material for the installation of a tuned-
mass system this strategy can be classified as 'cost effective‘.
References
Nawrotzki P. (2002) Artificial Increase of Elasticity and Damping for Seismically
Excited Structures. Proc. 12th European Conference on Earthquake Engineering,
London, Paper No. 58.
Nawrotzki P., Jurukovski D. and Rakicevic Z. (2005) Shaking Table Testing of a Steel
Frame Structure with and without Tuned-Mass Control System. Proc. Eurodyn 2005,
Paris.
Villaverde R. (2002) Roof Isolation System Implemented with Steel Oval Elements:
Exploratory Study. Proc. 3rd World Conference on Structural Control, Como, Italy.
Nawrotzki P. and Chouw N. (2004) Effectiveness of Tuned-Mass Dampers in
Reducing the Response of Soil-Structure Systems to Near-Source Earthquakes. Proc.
The 11th International Conference on Soil Dynamics & Earthquake Engineering,
Berkeley, USA.
