See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/263219968

Damping in Tall Buildings – Uncertainties and Solutions

Article  in  IABSE Congress Report · January 2008

DOI: 10.2749/222137908796225618

CITATIONS READS

11 5,721

2 authors:

Rob J. Smith Michael Willford

Arup Arup
9 PUBLICATIONS   324 CITATIONS    53 PUBLICATIONS   1,743 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Dynamic soil-structure interaction View project

Damping model for nonlinear dynamic analysis View project

All content following this page was uploaded by Rob J. Smith on 13 October 2017.

The user has requested enhancement of the downloaded file.

Damping in tall buildings – uncertainties and solutions Smith and Willford, 2008
Submitted for IABSE Chicago 2008 conference Arup, London, UK

Damping in tall buildings – uncertainties and solutions

Rob SMITH Michael WILLFORD

Associate Director
Arup Arup
London, UK and
London, UK San Francisco, USA
rob-j.smith@arup.com michael.willford@arup.com

Rob Smith, graduated from Michael Willford has worked in

Oxford University and has the field of structural dynamics
worked as a structural engineer for over 30 years. He is an Arup
for 13 years. He leads the tall Fellow and Global Leader of
buildings team within Arup’s Arup’s Advanced Technology
Advanced Technology Practice. and Research Practice

Summary
In the design of tall buildings, the dynamic response in wind is a dominant consideration. The
damping of the building has a major influence on the dynamic response, and the structural engineer
has to make an informed decision as to an appropriate value to assume for design.
This paper reviews available measured data on the damping of tall buildings and compares against
values in common usage. It is found that damping ratios commonly assumed in design are
inappropriate and unconservative.
The paper describes the consequences of over-estimation of damping at the design stage, and
illustrates the benefits of incorporating robust supplementary damping systems to reduce dynamic
response, to reduce uncertainty and to improve economy in design.
Keywords: damping; wind; seismic; dampers; high-rise buildings; risk; occupant comfort.

1. Introduction
The predicted dynamic response of tall buildings in wind is particularly sensitive to assumptions
about the intrinsic damping made by the structural engineer. Current practice uses rules of thumb
based on limited measurements, most of these made on relatively low rise (70-100m) buildings by
spectral methods at low levels of response.
This paper reviews good quality measurements, recently published in a variety of journals, of the
intrinsic damping in buildings up to 450m in height. These are compared to values in common
practice for use in the design of buildings. It is evident that current practice overestimates the
damping in the many buildings of 200m+ now being constructed worldwide, and the need for more
reliable damping predictions is imperative.
The consequences of overestimation of damping during design include higher lateral accelerations
(potential occupant comfort problems), increased lateral deflections (potential damage to cladding
and fit-out) and higher forces in structural members (reduced factor of safety and potential
structural damage and fatigue in extreme winds).

2. Sources of intrinsic damping

The intrinsic damping of a building arises from a number of energy dissipating mechanisms within
the building fabric and its foundation. The principal mechanisms are discussed below.

2.1 Structural material damping

The damping characteristics of primary structural materials can be assessed in the laboratory. They
vary somewhat, but typical values for concrete material are 0.5% of critical [3] and 0.3% for steel.

Page 1 of 8 Under peer review

Damping in tall buildings – uncertainties and solutions Smith and Willford, 2008
Submitted for IABSE Chicago 2008 conference Arup, London, UK

2.2 Structural connections

Additional damping can be generated in certain types of structural connections, where slip or local
yielding arise. However, for monolithic concrete or steel structures with welded or slip-critical
bolted connections little damping will be generated in connections at service load levels.

2.3 Foundations
Foundations provide two generic sources of dissipation. There is the intrinsic damping of the soil
material, and also radiation damping – the propagation of energy away from the building in body
and surface waves in the soil. The main mode of wind-induced foundation deformation in many tall
buildings is low frequency rocking, which is associated with very low radiation damping.

2.4 Non-structural components (cladding etc.)

Non-structural elements such as cladding, partition walls and mechanical service ducts provide
potential sources of damping. Since many of these are not rigidly connected to the primary structure,
then movement within the building will cause some energy dissipation as the relative movement
between elements generates frictional forces. Previous work by others [4] shows clearly that the
presence of internal partitions affects the level of damping measured in a building. In tall buildings,
the primary structure dominates the elastic stiffness and the contribution of the non-structural
components in relatively small. This may contribute to the observed lower overall damping for
taller buildings.

2.5 Aerodynamic damping

The vibration of a building in the direction of incident wind flow results in a degree of aerodynamic
damping, which is calculable but generally small. Vibrations across the wind flow generally do not
generate damping, and in some cases result in enhanced excitation, or ‘negative’ aerodynamic
damping.

2.6 Hysteresis of yielding components

When structural components are deformed cyclically beyond their elastic limit energy is dissipated
in plastic strain. Repeated cyclic plastic strain is damaging, leading to low-cycle fatigue, and so this
form of damping cannot be relied upon for in the normal ‘design’ range. It is only acceptable in
rare earthquakes, which are of relatively short duration and in which damage is tolerable, and not
during severe wind loading which occurs over several hours and where fatigue could result.

3. Measurements of damping
3.1 Measurement techniques
The only basis for the prediction of damping in a tall building is the evidence of previous
measurements. Unfortunately, damping is difficult to measure with precision, and some historically
much-used techniques are unreliable, almost always resulting in over-estimation of the actual
damping. It is now well established that the various ‘random decrement’ methods are more
representative than the previously used spectral methods.
3.2 Variation of damping with height
Figure 1 shows a collation of high-quality data from a number of recent sources. The primary
source is Satake [7], but also included are a number of other recent measurements, for buildings up
to 450m This collation was first presented in [8]
The data shows a lower ‘floor’ of damping, at about 0.4% of critical at all building heights greater
than 50m. For buildings lower than 100m, whilst damping is generally below 3% of critical,
considerably higher damping is sometimes measured. However, for buildings above 250m it is
exceptional for measured values to significantly exceed 1% of critical. There is therefore a clear
trend that damping reduces with increasing building height. The authors believe that this decrease in
damping is because as the primary structural members become much larger in very tall buildings
the contribution of non-structural elements to energy dissipation becomes relatively less significant.

Page 2 of 8 Under peer review

Damping in tall buildings – uncertainties and solutions Smith and Willford, 2008
Submitted for IABSE Chicago 2008 conference Arup, London, UK

Also plotted on figure 1 are a number of measurements from tall chimneys. Chimneys, by their
nature of construction, are dominated by their primary structure, with internal flues contributing
little to damping and stiffness. British Standard 4076 [2] shows this variation in the treatment of
damping – flues typically increase damping by 15-20%. It can be seen that damping ratios for tall
buildings are similar to those of chimneys.

3.3 Variation of damping with material type

Previous studies have indicated a variation of damping with structural material type. Whilst this
seems to be the case for relatively short buildings, for high-rise there appears to be less difference.

Damping ratio vs building height

8
7
Damping ratio (%)

Buildings, Steel ,measured

6 Buildings, SRC / unknown, measured

5 Buildings, RC, measured
4
Chimneys, RC ,measured
3
2
1
0
0 100 200 300 400 500
Height [m]
Figure 1 - Measured damping vs building height
3.4 Variation of damping with
amplitude
Damping ratio vs Acceleration There is a body of evidence which
2 indicates that damping in buildings
Damping ratio [%]

1.5 increases with amplitude of motion. For

instance [5] and [6] indicate an increase
1 in damping at low levels of amplitude up
0.5
to a high level plateau. Tamura [9]
introduces the concept of critical tip drift
0 ratio, arguing that damping will stay
0 5 10 15 constant or even decrease at higher
Acceleration [milli-g] amplitudes. Figure 2 shows three
examples of the variation of damping
New Bank of China Hong Kong (SRC) {Jeary et al., 1998}
with amplitude. The horizontal axis has
200m-high office bldg {Okada et al.,, 1993, cited by Tamura}
been converted to acceleration to show
100m-high steel bldg {Jeary, 1998, cited by Tamura}
that at clearly perceptible levels of
motion the damping can be as low as
Figure 2 - Variation of damping with amplitude 0.5%.
There is a lack of recorded data for high amplitude response (i.e. towards strength design motions)

Page 3 of 8 Under peer review

Damping in tall buildings – uncertainties and solutions Smith and Willford, 2008
Submitted for IABSE Chicago 2008 conference Arup, London, UK

on tall buildings. However, the data referenced above indicates that peak damping may be reached
at relatively low amplitude. It should be noted that damping ratio reducing with amplitude is a
normal characteristic of friction-damped systems.

4. Selection of intrinsic damping in design

4.1 What damping is assumed
Figure 3 shows a dataset of damping values assumed in the design of tall buildings compared to the
measured values. The data comes from numerous sources, all publicly available. As can be seen, in
most cases, the damping assumed in design is significantly higher than measurements for tall
buildings, more particularly for those taller than 200m.

Assumed damping vs building height

8
Buildings, Steel, measured
7
Damping ratio (%)

Buildings, SRC / unknown,

measured
6 Buildings, RC, measured

5 Chimneys, RC, measured

Lower trendline
4 Design assumption

3
2
1
0
0 100 200 300 400 500
Height [m]

Figure 3 - Comparison between measured and assumed damping

It is common to assume different levels of damping for serviceability and ultimate wind design.
Where this is done, both values are shown. However, as shown in section 3.4, peak damping ratio
can occur at relatively low amplitude, so the authors believe that this practice may not be fully
justified.

4.2 Variation of design wind load with damping

4.2.1 Wind overturning moment

The dynamic response of a tall building has a significant effect on the lateral forces for which it has
to be designed. The forces are of two types:
• the mean and fluctuating applied aerodynamic loads and
• the inertia forces associated with the dynamic sway of the building
The building inertia forces (the ‘dynamic’ or ‘resonant’ response) need to be combined with the
fluctuating aerodynamic forces (the ‘gust’ or ‘background’ wind forces). When spectral methods
are used the peak dynamic (resonant) forces and peak gust forces are combined thus:

Page 4 of 8 Under peer review

Damping in tall buildings – uncertainties and solutions Smith and Willford, 2008
Submitted for IABSE Chicago 2008 conference Arup, London, UK

Total _ response = Mean ± Gust 2 + Dynamic 2

Assuming that the aerodynamic loading on the building is effectively a broadband random process
(in the frequency range in which the natural period of the building lies) the dynamic response is
inversely proportional to the square root of the damping coefficient of the building.
A typical example is shown in figure 4, which shows the sensitivity of the design overturning
moment of a 400m building to the assumed intrinsic damping. The data is based upon analysis of
high frequency force balance wind tunnel
test measurements.
Overturning Moment Mxx
In this example the measured value of
30000
damping was 0.8%. Had the designer
25000 assumed 2% damping, the design loading
would have been
(0.8/2) = 0.63 of the
20000
value derived from the measured
Moment (MNm)

15000
damping. This is clearly a significant
t = 8.75 sec difference. Given that measured damping
10000 in buildings can be as low as 0.3% and
assumed as high as 3%, there is clearly
5000
cause for caution.
0 If the damping were very high (>10%),
0

10

12

Damping ratio (%) the dynamic response would become

very small and the overturning moment
would approach the value of the applied
Figure 4 - typical variation of overturning moment with aerodynamic loading.
building damping

4.2.2 Perceived motions

The lateral motions felt by occupants are dominated by the resonant response of the building.
Whilst the acceptability of these motions is a complex subject, not covered in this paper, it is
commonly assumed that the acceleration is inversely proportional to the square root of the damping.
Acceleration ∝ 1 / ξ

As with the overturning moment, the consequences of overestimation of damping leads to higher
accelerations in the building than predicted at design stage.

4.3 Variation of seismic load with damping

The majority of damping provided by a structure during severe earthquakes comes from energy
dissipation in yielding elements. In addition, for long period structures responding to short duration
events, the response is not highly sensitive to damping. Thus, the effect of “elastic damping” on
seismic response is much less significant than for wind induced response. However, when
considering service level earthquakes (e.g. 50 year return period), it is advisable to assume the same
levels of damping as for wind loading rather than the traditional ‘seismic’ damping level of 5% of
critical which is inappropriate for tall buildings.

4.4 Other design issues

4.4.1 Load factors

Load factors commonly used for wind design generally vary between 1.2 and 1.6. The purpose of
load factors is often argued about amongst engineers, but for wind it can be seen as an allowance
for extreme wind events. Typically the nominal wind load is based upon a 50 to 100 year event.
For ‘normal’ buildings within the remit of design codes the load factor typically brings this up to
about 1000 year return period. The authors’ view is that the following additional considerations
should be taken into account when selecting load factors for wind on tall buildings:

Page 5 of 8 Under peer review

Damping in tall buildings – uncertainties and solutions Smith and Willford, 2008
Submitted for IABSE Chicago 2008 conference Arup, London, UK

• Uncertainty in wind speed data at the level of the top of the building.
• Uncertainty in wind tunnel model testing
• Uncertainty in dynamic response.
Wind loading codes are written with smaller buildings in mind where these points are of low
relevance. The uncertainty in design wind speed at the top of a tall building is significant. Almost
all wind climate estimates are based upon measurements made at 10m above ground level.
Extrapolation to heights in excess of 200m is therefore inevitably prone to some uncertainty.
As was demonstrated in section 4.2, it is possible that a traditional overestimation of damping alone
may lead to design loads that are only 63% of their “true” value. This uncertainty effectively “uses
up” the whole of a 1.6 load factor.

5. Supplementary damping
5.1 Overview
Supplementary damping systems are a means of providing additional energy dissipation using
engineered systems. They have a distinct advantage in that the damping can be predicted before
construction. Even if the damping systems are not operating at optimum performance, the relative
error in dynamic response is likely to be less, as is shown in table 1.
Case Assumed Actual Assumed Actual Effect on
intrinsic intrinsic supplementary supplementary dynamic
damping (%) damping (%) damping (%) damping (%) response
Undamped 1.5 0.5 0 0
(1.5/0.5)=1.72
Damped 1.5 0.5 5 4.0
(6.5/4.5)=1.20
Table 1 - Consequence of error in damping assumptions
When adopting supplemental damping in design, it is important to consider the redundancy and
robustness of the damping system. The engineer should only rely upon a supplementary damping
for strength design there is confidence in the absolute reliability of the system.
The next section describes two forms of building supplementary damping system – tuned mass or
liquid dampers and fluid-viscous dampers.

5.2 Tuned mass/liquid dampers

Tuned mass or liquid dampers are inertial devices in which the motion of a large mass is used to
counteract the sway of the building. Whilst a tuned mass/liquid damper system can often be
designed to provide the equivalent of 2% to 4% of critical damping for the building, these devices
have a number of disadvantages;
• they are large, heavy, and take up valuable space at the top of the building
• they have to be ‘tuned’ closely to the measured natural frequency of the building mode of
concern - if there are several modes of concern then several sets of differently tuned devices
are required
There is reluctance to rely upon tuned mass/liquid devices to reduce the strength design forces on a
building because:
• the natural period of the building might change with time and with response amplitude such
that the device becomes de-tuned and less effective
• It is more economical in terms of cost and space to provide one large damper unit. There is
therefore a question of dependability. If the unit or one component of it were to fail, there
would be no backup
• It is not practical to design these devices to provide control under severe seismic events.
For these reasons tuned devices have only been used to reduce occupant perception of motion and

Page 6 of 8 Under peer review

 Damping in tall buildings – uncertainties and solutions Smith and Willford, 2008
Submitted for IABSE Chicago 2008 conference Arup, London, UK

not strength design loads.

However, TMDs are often favoured by structural engineers since they can be incorporated into the
design at a late stage, even after construction, and can be designed separately from the building
itself. Active Mass Dampers are effectively powered tuned mass dampers, and can achieve greater
control with a limited mass. However, their reliance on external power sources reduces their
reliability in extreme events.

5.3 Viscous or visco-elastic dampers

Viscous or visco-elastic dampers are ‘resistance’ devices attached between two points on a structure,
and operate on the relative motion between those two points. They are best located to connect two
points having significant relative displacement in the vibration mode of concern. These devices do
not require frequency tuning. A number of generic types are available, and they have been used to
reduce seismic damage in bridges and buildings.
Further information on different damper types is provided in
[8], which also describes an example of their application, the
damped outrigger system. This concept is an effective
method of providing significant additional damping (typically
5-10% of critical) by using relative vertical movement
between outriggers and perimeter columns to actuate fluid
viscous dampers. The system has a number of advantages
over mass dampers because:
• Viscous dampers are not sensitive to exact tuning.
• They can provide greater levels of damping
• They take up less space, and are placed within the
structure
• They offer greater reliability and redundancy, so can
be used to reduce design wind forces. The implication
of this is that the main structure can be reduced in size
(and hence cost), thus providing a net cost benefit.
Viscous damping systems therefore have the potential to
significantly alter the economics of tall buildings, making
more slender buildings economically feasible.
Although, viscous dampers can be introduced at a late stage in
Figure 5 - The Arup Damped design (even retrofit), early incorporation into the design is
Outrigger System (patent pending) the best way to exploit their potential.

6. Other means of reducing dynamic response

6.1 Stiffening the building
An increase in the natural frequency of a structure will often result in reduced dynamic response.
This is usually much less economic than incorporating supplementary damping.

6.2 Aerodynamic shaping

The external shape of a building has a major effect on the wind forces and the potential level of
dynamic response. Adopting a non-uniform prismatic shape up the height of a building is generally
beneficial. For success, this approach generally requires extensive interaction between Architect,
Engineer and Wind Tunnel Laboratory at an early stage of a project. This was studied extensively
in the design of the Burj Dubai [1].

Page 7 of 8 Under peer review

 Damping in tall buildings – uncertainties and solutions Smith and Willford, 2008
Submitted for IABSE Chicago 2008 conference Arup, London, UK

7. Conclusions and recommendations

7.1 Conclusions
• There is a clear trend for the intrinsic damping of tall buildings to reduce with height.
• For buildings taller than 250m, almost all reliable measurements indicate damping below
1% of critical.
• Structural designers and wind tunnel testing laboratories frequently assume higher values of
damping for design than the measurements support. This is unconservative.
• There is little evidence to suggest that damping will increase between typical serviceability
amplitudes (1-10 year return period wind) and the design wind load.
• There is little difference in the measured damping of steel and concrete frame buildings
above 250m in height.
• Viscous dampers and similar components are an effective way of reducing both the dynamic
response and the risk of unconservative structural design.

7.2 Recommendations
Structural engineers and wind loading specialists should either:
• Assume damping ratios consistent with measurements – typically 0.5% - for buildings
above 250m in height.
• Incorporate supplementary damping systems, preferably robust systems which can improve
the economy and performance of tall buildings.

8. References
[1] BAKER et al, “Burj Dubai : Engineering the worlds tallest building”, The structural design of
tall and special buildings, Vol 16, 2007, pp 361 – 375
[2] BRITISH STANDARDS INSTITUTE, British Standard 4076, Specification for Steel
Chimneys, London, BSI, 1989
[3] COLE D, SPOONER D, “The damping Capacity of Concrete,” International conference on
the structure of concrete and its behaviour under load, London, Sep 1965, pp 217-225
[4] ENGINEERING SCIENCES DATA UNIT, Damping of Structures, Part 1, Tall Buildings,,
London, ESDU, data Item 83009
[5] FANG, J.Q., LI, Q.S., JEARY, A.P., LIU, D.K., “Damping of Tall Buildings: Its Evaluation
and Probabilistic Characteristics”, Structural Design of Tall Buildings, Vol 8, 1999, pp. 145-
153.
[6] LI QS, YANG K, ZHANG N, WONG CK, JEARY AP. “Field measurement of amplitude-
dependent damping in a 79-storey tall building and its effects on the structural dynamic
responses”. The Structural design of tall buildings, Vol 11, 2002, pp 129-153
[7] SATAKE N, SUDA K, ARAKAWA T, SASAKI A, TAMURA Y.. “Damping evaluation Using
Full-Scale Data of Building in Japan” ASCE Journal of Structural Engineering, Vol 129, 2003:
pp470–477
[8] SMITH R, WILLFORD M, “The damped outrigger concept for tall buildings”, The Structural
Design of Tall and Special Buildings, Vol 16, 2007, pp 501-517
[9] TAMURA Y., “Amplitude Dependency of Damping in Building and Estimation Techniques”
Keynote lecture, 12th Australasian Wind Engineering Conference, Queenstown, New Zealand,
Jan 31–Feb 2 2006.

Data sources for figure 1 and 2 are listed in Smith and Willford [8].

Page 8 of 8 Under peer review

View publication stats