Professional Documents
Culture Documents
Vdocument - in Design of Footbridges
Vdocument - in Design of Footbridges
RFS2-CT-2007-00033
Design of Footbridges
Background Document
Footbridge_Background_EN02.doc – 12.11.2008
Table of Contents
1 Introduction ....................................................................................... 5
2 Definitions ......................................................................................... 6
3 Design procedure................................................................................ 6
4 Design steps ...................................................................................... 6
4.1 Step 1: Evaluation of natural frequencies ...................................... 6
4.2 Step 2: Check of critical range of natural frequencies ...................... 8
4.3 Step 3: Assessment of Design Situation ........................................ 8
4.3.1 Step 3a: Assessment of traffic classes ......................................... 9
4.3.2 Step 3b: Assessment of comfort classes ...................................... 9
4.4 Step 4: Assessment of structural damping....................................10
4.4.1 Damping model...................................................................... 10
4.4.2 Damping ratios for service loads ............................................... 11
4.4.3 Damping ratios for large vibrations ........................................... 12
4.5 Step 5: Evaluation of acceleration ...............................................12
4.5.1 Harmonic load model .............................................................. 13
4.5.2 Response Spectra Method for pedestrian streams........................ 19
4.6 Step 6: Check of lock-in with bridge vibration ...............................20
4.7 Step 7: Check of comfort level....................................................21
5 Evaluation of dynamic properties of footbridges......................................22
5.1 Introduction .............................................................................22
5.2 Response measurements ...........................................................22
5.2.1 Measurements of ambient response for identification of critical
natural frequencies ........................................................................... 22
5.2.2 Raw measurement of damping ratios associated with critical natural
frequencies ...................................................................................... 23
5.2.3 Measurement of the response induced by one pedestrian ............. 23
5.2.4 Measurement of the response induced by a group of pedestrians... 23
5.2.5 Measurement of the response induced by a continuous flow of
pedestrians ...................................................................................... 24
5.3 Identification tests ....................................................................24
5.3.1 Forced vibration tests.............................................................. 24
5.3.2 Ambient vibration tests ........................................................... 27
5.3.3 Free vibration tests................................................................. 27
5.4 Instrumentation........................................................................28
5.4.1 Response devices ................................................................... 28
1
2
L length [m]
m number of half waves [-]
m* modal mass [kg]
M mass [kg]
n number of the pedestrians on the loaded surface S [P]
(n = S × d )
n’ equivalent number of pedestrians on a loaded surface S [P/m²]
p(t ) distributed surface load [kN/m²]
P mov moving load [kN]
S area of the loaded surface [m²]
δ logarithmic decrement for damping [-]
μ mass distribution per unit length [kg/m]
μD bridge deck mass per unit length [kg/m]
μP pedestrian mass per unit length [kg/m]
ρ influence factor for additional pedestrian mass [-]
Φ( x ) mode shape [-]
ψ reduction coefficient to account for the probability of a [-]
footfall frequency in the range the natural frequency for
3
4
1 Introduction
In recent years, there has been a growing trend towards the construction of
lightweight footbridges. Due to its reduced mass of such structures, the dynamic
forces can cause larger amplitudes of the vibration. The more slender structures
become, the more attention must be paid to vibration phenomena.
The increase of vibration problems in modern footbridges shows that footbridges
should no longer be designed for static loads only. But fulfilling the natural
frequency requirements that are given in many codes ([1], [2], [3], [4]) restricts
footbridge design: very slender, lightweight structures, such as stress ribbon
bridges and suspension bridges may not satisfy these requirements. Moreover
not only natural frequencies but also damping properties, bridge mass and
pedestrian loading altogether determine the dynamic response. Design tools
should consider all of these factors. Provided that the vibration behaviour due to
expected pedestrian traffic is checked with dynamic calculations and satisfies the
required comfort, any type of footbridge can be designed and constructed. If the
vibration behaviour does not satisfy some comfort criteria, changes in the design
or damping devices could be considered.
These lightweight footbridges have a decrease in mass, which reduces the mass
inertia and which lowers natural frequencies, resulting in a greater risk of
resonance. Resonance occurs if the frequency of the bridge coincides with the
frequency of the excitation, e.g. the step frequency of pedestrians. Pedestrian
induced excitation is an important source of vibration of footbridges. Pedestrian
loading is by nature unsteady, transient and waddling in a small range of
excitation frequency. It is therefore obvious that dynamic responses play a
fundamental role in the design of vibration susceptible structures. Vibrations of
footbridges may lead to serviceability problems, as effects on the comfort and
emotional reactions of pedestrians might occur. Collapse or even damage due to
human induced dynamic forces have occurred very rarely.
Vibrations of footbridges may occur in vertical and horizontal directions, even
torsion of the bridge deck is possible. Dynamic actions of cyclists are negligible
compared to the actions caused by walking and running individuals.
In recent years some footbridges were excited laterally by dense pedestrian
streams in which pedestrians interacted with the bridge vibration. A self-excited
large response causes discomfort. Footbridges should be designed in such a way
that this pedestrian-bridge-interaction phenomenon, also called ‘lock-in’, does
not arise.
Another dynamic loading on footbridges is intentional excitation by people that
are jumping on the spot, bouncing, swaying body horizontally, shaking stay
cables etc. in resonance to produce large vibrations. In that case, the comfort is
certainly not fulfilled but the structure must not collapse.
Hence, in modern footbridge design, the assessment of human-induced
vibrations needs to be considered by the designer to ensure that
• vibrations due to pedestrian traffic are acceptable for the users,
• the lock-in phenomenon does not arise,
• the footbridge does not collapse when subjected to intentional excitation.
5
2 Definitions
No further background information.
3 Design procedure
It is recommended to consider dynamic actions and the vibration behaviour of
the structure in an early design stage, even when damping and some foundation
properties are unknown and have to be estimated. Hence, the calculated
vibration behaviour gives an indication of the real behaviour only. If the response
is in the critical range, provisions for damping devices should be made in the
early design stage. Damping and accelerations caused by several dynamic loads
should then be measured after finishing the construction. Based on the real
dynamic properties it should be decided whether the damping devices are
necessary or not.
4 Design steps
4.1 Step 1: Determination of natural frequencies
Although hand formulas and simplified methods can be used in a preliminary
evaluation of natural frequencies, whenever these are close to a critical range
from the point of view of pedestrian excitation, a more precise numerical model
should be used. In modern bridge design the use of finite element software is
widely spread in all stages of design, even during the conceptual one.
Consequently, it is suggested to use a FE-Model of the bridge not only to
calculate the stress and deformation of the footbridge but also to determine its
natural frequencies. Hence, preliminary dynamic calculations can easily be
performed without additional means.
A first approach is to keep the model as simple as possible and to model the
bridge with beam elements, cable elements, spring or truss elements in a three
dimensional FE model. The latter should always allow for vertical, horizontal, and
torsional mode shapes. A rough overview over the natural frequencies and the
corresponding mode shapes is obtained and problems regarding the dynamic
behaviour can be identified. The more complex the static system and the higher
6
the mode shape order, the more finite elements are required. A more refined
model may take advantage of various types of finite elements such as plate,
shell, beam, cable or truss elements. To get reliable results for natural
frequencies, it is absolutely necessary that bearing conditions, foundation
stiffness, stiffness and mass distribution are modelled in a realistic way. All dead
load, superimposed dead load and pre-stressing of cables have to be considered
for the calculation of natural frequencies. The superimposed dead load of the
bridge caused by furniture, barriers, pavement and railings is considered as
additional masses as exactly as possible. A lumped mass approach, in which
rotational masses are neglected, is in many cases sufficient. For the modelling of
abutments and foundations, dynamic soil stiffness should be used. Otherwise the
obtained results will be very conservative or very inaccurate.
In any case it is recommended to determine first and foremost the natural
frequencies of a built footbridge by experimental investigation in addition to
computer calculations before the final configuration of the damping units are
determined.
The modal mass for each mode shape should be available, when verification of
comfort is done with the SDOF-method (cf. section 4.5.1.2).
The investigation of dynamic characteristics for selected footbridges shows
clearly that, especially for lightweight structures, the additional mass due to
pedestrians has a great influence on the natural frequencies of the system. For
individuals and group loading this effect is usually negligible, but if pedestrian
streams have to be taken into account, this influence may cause a significant
decrease in natural frequency. This depends on the ratio between mass
distribution of the deck and pedestrian mass distribution. The decrease in
frequencies is higher for the footbridges having less dead load.
The natural frequencies might fall to a more or to a less critical frequency range
(cf. section 4.2) for pedestrian induced dynamic excitation. With additional dead
load or live load, the natural frequencies of the footbridge could decrease and
shift into the critical frequency range or leave it. Furthermore, it has to be noted
that the given limit values of critical frequency ranges should not be taken as
sharp values but rather as soft values.
In some case the obtained increase of modal mass can be even greater than
50 % of the modal mass of the bridge.
The influence of the static pedestrian mass can be estimated easily: the modal
mass m* including the additional static pedestrian mass is calculated according
to eq. 4-1.
∫
m * = μ D ρ ( Φ (x))2 dx
LD
Eq. 4-1
where
μ D [kg/m] is the bridge deck mass per unit length
D + μ P
μ
ρ = is the influence factor for additional pedestrian mass
μ D
7
This is within the accuracy of the whole model compared to the natural
frequencies that will be measured in reality. Therefore, it is recommended to
neglect the influence of an increased modal mass lower than 5 % on the natural
frequency.
8
Eurocode principles for reliability [5] state some design situations out of which
the ones listed below could be relevant for footbridges subjected to pedestrian
loading. They can be associated with the frequency of exceeding a certain limit
state like a comfort criteria in question:
• Persistent design situations, which refer to the conditions of permanent use
• Transient design situations, which refer to temporary conditions
• Accidental design situations, which refer to exceptional conditions.
There are design situations which might occur once in the lifetime of a footbridge
like the inauguration of the bridge. But on the other hand there might be a
design situation where few commuters will pass daily.
Realistic assumptions of the different design situations should be taken into
account by using defined traffic classes (cf. section 4.3.1) for the verification of
pedestrian comfort. As aforesaid, the inauguration of the footbridge for example
would govern the design in almost every case though it happens only once in the
lifetime of a bridge. It must therefore be decided which comfort criteria are to be
chosen for the footbridge design (cf. section 4.3.2) for an extreme and rare
situation such as the inauguration or for the everyday density of pedestrians on
the structure.
9
10
4
Measured, steel
3,5 Measured, timber
Measured, stress-ribbon
3 Measured, concrete
2,5
)
%
( 2
ξ
1,5
1
0,5
0
0 2 4 6 8 10
Frequency (Hz)
Figure 4-2: Measured damping ratios under service loads: variation with natural
frequency
11
4 Measured, steel
3.5 Measured, timber
3 Measured, stress-ribbon
Measured, conc.
2.5
2
Η
1.5
1
0.5
0
0 20 40 60 80 100 120 140 160 180
Span (m)
Figure 4-3: Measured damping ratios under service loads: variation with span
12
Introduction
If a harmonic load (F 0 sin (2 π f 0 t )) is applied to a damped SDOF system, the
response of the system can then be given in the form which will be used
throughout the procedure for the assessment of an equivalent number n’ of
pedestrians using modal analysis:
F 0 4 π 2 M
x (t ) = sin(2 π f 0 t − φ ) Eq. 4-6
(f − f ) + 4 ξ
2 2 2
0
2
f 2 f 0 2
Modal analysis
Let a beam be modelled as a system with N degrees of freedom (cf. Figure 4-4)
and let a loading be represented as point loads on each of the (loaded) nodes.
When a solution to describe the dynamic behaviour of a system is sought by
modal analysis, displacements of the nodes are found in the form of
superposition of displacements belonging to different representative modes:
r
y (t ) = ∑ x (t )Φ , r ≤ N Eq. 4-7
i i
i = 1
where:
y (t ) is the vector of the movements of concentrated masses,
Φi are the vectors of modal displacements taken into consideration,
x i( t ) are the responses of the system for each mode i taken into
consideration.
F
m1 m j mN
φij
Φi
13
If all the loads share the same frequency, f 0 ≠ f i , the response of the system for
one mode only (e.g. mode i , with modal displacements φij , cf. Figure 4-4) is:
Φi T F 0 4 π 2 m* i
x i (t ) = sin(2 π f 0 t − φi ) Eq. 4-8
(f i 2 − f 0 2 )2 + 4 ξ i 2 f i 2 f 0 2
with: Φi T = {φi1 , φi2 ,… φij ,… φiN } vector of modal displacements,
F 0 vector of load amplitudes (F 0T = {F 1, F 2,… F j ,… F N}),
N
∑
m* i = m j φij 2
j =1
modal mass,
Q j
φij
φi ,max
Φi ( x ) L /n
L
jL n
where ∫
α nij = n Φi ( x ) dx .
( j −1 )L n
14
where the phase shift for mode i and a point load at node j is:
⎛ 2 ξ i f i f 0 j cosψ j + (f i 2 − f 0 2 j )sinψ j ⎞
φij = arctan ⎜
⎜ (f i 2 − f 0 2 j )cosψ j − 2 ξ i f i f 0 j sinψ j ⎟ .
⎟
⎝ ⎠
If the assumption that all the loads share the same amplitude but are not
necessarily in phase ( Q = Q sinψ ) is adopted, the response becomes:
j j
n
(α φi,max 4 π m* i L )sin (2 π f 0 j t − φij )
2
y i,max (t ) = Q ∑ j =1
nij
with the phase shift for mode i and a point load at node j :
⎛ 2 ξ λ (1 + μ u j )cosψ j + ( λi 2 − (1 + μ u j )2 )sinψ j ⎞
φij = arctan ⎜⎜ 2 i i ⎟
⎟=
⎝ ( λi − (1 + μ u j ) )cosψ j − 2 ξ i λi (1 + μ u j )sinψ j ⎠
2
⎛ 2 ξ λ (1 + μ u j ) ⎞
= arctan ⎜⎜ 2 i i ⎟ + ψ ,
⎝ ( λi − (1 + μ u j ) ) ⎠
2
⎟ j
15
and, finally:
) Q z i
&& = (2 π Eq. 4-10
2
Z i
Note: For λi = 1, μ = 0 and ψ j = 0 (deterministic resonant loading case):
Q ⎡ n
α Nij φi,max 4 π 2 m* i L ⎛ π ⎞⎤
&&
Z i
= (2 π )2 f s2 1
f s2 1
× max
t
⎢
⎣
∑
j =1 2 ξ i
sin⎜ 2 π f sj t −
⎝
⎟⎥ =
2 ⎠⎦
144 4 4 4 4 4 4 4
244 4 4 4 4 4 4 4
3 Eq. 4-11
z i '
= (2 π )2 Q z i '
n'
Equivalent stream with n’ ≤ n pedestrians (eq. 4-11): && = (2 π ) Q z '
Z i,eq
2
i
n
} n } n’
Q j Q j n’/n
Φi ( x ) Φi ( x )
The random feature in equation 4-13 is z i . The mean value E ( z i ) and the standard
deviation σ ( z i ) can all be assessed by simulations for different values of
intervening parameters:
⎛ n (α Nij φi,max 4 π 2 m* i L)2 (1 + μ u j )4 ⎞
⎜ × ⎟⎟
∑
⎜
z i = max j =1 ( λi − 1 − 2 μ u j − μ u j ) + 4 ξ
2 2 2 2
i λi (1 + 2 μ u j + μ u j ) ⎟
2 2 2 2
Eq. 4-14
t ⎜
⎜ × sin(2 π f t − φ ) ⎟
⎝ sj ij ⎠
Results
16
30 0
25 0
s
n
o 20 0
i
t
a
s
i
l 15 0
a
é
R10 0
50
2 6
1 9 2
6 8 3 4
0 9
7 5 4 1 7 2 4
0 , 1 , 3 , 4 , 6 , 7 , 9 , , 1 2 6
, ,
1 1 1 1 4
k eq
Figure 4-7: An example of the obtained histograms
With such a value of k eq, the equivalent number of pedestrians, n’ can be
obtained. Expressions for this equivalent number have been derived by
regression as a function of the damping ratio and the total number of pedestrians
on the footbridge.
4.5.1.3 SDOF-method
As an example of application of the SDOF – method, a simple supported beam is
considered. This beam has a distributed mass μ [kg/m], which is the cross
section times the specific weight, a stiffness k and a length L. The uniform load
p( x ) sin(ω t ) is distributed over the total length.
17
The mode shapes Φ( x ) of the bending modes are assumed to be represented by
a half sine function Φ( x ) = sin(m × x
/L × π) whereas m is the number of half
waves.
p( x )⋅sin(ωt )
0.
μ 00
Figure 4-8: Simple beam with harmonic mode shape Φ( x ), m=1
The generalised mass m* and the generalised load p* sin(ω t ) are calculated as
follows:
∫ ⋅ (Φ(x)) dx
m * = μ Eq. 4-15
2
LD
∫
p* sin(ω t ) = p(x) Φ(x) dx ⋅ sin(ω t )
LD
Eq. 4-16
Expressions for the generalised mass m* and the generalised load p* sin(ω t )
are systematised in Table 4-2 for a simple supported beam. The generalised load
for a single load P mov sin(ω t ), moving across the simple beam is also given in
this table. This excitation is limited by the tuning time which is defined as the
time for the moving load to cross one belly of the mode shape.
Table 4-2: Generalised (modal) mass and generalised load
m=1:
1 2
p(x) L
2
P mov
⎛ x ⎞ μ L L /v
φ( x ) = sin ⎜ π ⎟ 2 π π
⎝ L ⎠
m=2:
1 1
p(x) L
2
P mov
⎛ 2 x ⎞ μ L L /(2v )
φ( x ) = sin⎜ π ⎟ 2 π π
⎝ L ⎠
m=3:
1 2
p(x) L
2
P mov
⎛ 3 x ⎞ μ L L /(3v )
φ( x ) = sin⎜ π ⎟ 2 3 π π
⎝ L ⎠
with:
P mov [kN]: moving load L [m]: length
p( x ) [kN/m]: distributed load m [-]: number of half waves
18
μ [kg/m]: mass distribution per length v [m/s]: velocity of moving load
The 2nd mode shape of a single span beam has two half waves (m = 2). When
loading the entire length and when half of the uniformly distributed load is acting
against the displacements of one belly and the other half is acting within the
sense of displacements, then the generalized load will result in a value of p* = 0.
The generalised load in the given table is based on the assumption that each
belly of the mode shape is loaded, which results in larger oscillations. In doing
so, the load is always acting in the sense of displacements of the bellies and the
generalised load p* for all mode shapes is the same as for the first bending
mode (m = 1). It must be noted that this approach may differ from other
recommendations. According to some approaches [32] the loaded surface
depends on the shape of the normal mode under consideration, according to
others [9] the whole ‘loadable’ surface should be considered.
Both factors were derived from Monte Carlo simulations which are based on
numerical time step simulations of various pedestrian streams on various bridges
geometries.
The standard deviation of acceleration is obtained as a result of application of
stochastic loads to a determined system. These loads have been defined
considering bridges with spans in the range of 20 m to 200 m and a varying
width of 3 m and 5 m, loaded with four different stream densities (0,2 P/m²,
0,5 P/m², 1,0 P/m² and 1,5 P/m²). For each bridge type and stream density
5 000 different pedestrian streams have been simulated in time step calculations
where each pedestrian has the following properties, taken randomly from the
specific statistical distribution:
• Persons' weight (mean = 74,4 kg; standard deviation = 13 kg),
• Step frequency (mean value and standard deviation depend on stream
density),
• Factor for lateral footfall forces (mean = 0,0378, standard
deviation = 0,0144),
• Start position (randomly) and
• Moment of first step (randomly).
The peak factor k a,d is used to determine the characteristic response of the
system. In serviceability limit states the characteristic value is the 95th percentile
k a,95%. This factor is also a result of Monte Carlo simulations.
19
Another result of the simulations where the first 4 vertical and the first two
horizontal and torsional modes have been considered is the risk of lateral lock-in.
To identify this risk a trigger amplitude of horizontal acceleration of 0,1 m/s² has
been defined. The following frequency range is relevant for horizontal lock-in:
f i
0,8 ≤ ≤ 1,2 Hz ,
f s,m / 2
where: f i
is the horizontal lateral natural frequency and
f s,m is the mean value of step frequency.
Natural frequencies to be considered should coincide with mean step frequencies
of pedestrian streams.
20
Lateral movement of
the centre of gravity
lateral deck
displacement
Time
lateral deck
velocity
performed work
(pos. work = raising, + + + +
neg. work = reducing) - - - -
Time
Figure 4-9: Schematic description of synchronous walking
Experiments on a test rig within the project SYNPEX [13] indicate that a single
person walking with a step frequency f i ± 0,2 Hz tends to synchronise with deck
vibration. Faster walking persons are nearly not affected by the vibration of the
deck, as the contact time of the feet is short and the walking speed high. They
seem to be less instable than those walking with slow and normal speed.
The lock-in trigger amplitude is expressed in terms of acceleration. Further
frequency dependence could exist but has not been detected in measurements.
Tests in France [6] on a test rig and on the Passerelle Solferino indicate that a
trigger amplitude of 0,1 to 0,15 m/s2 exist when the lock-in phenomenon begins:
a − = 0,1 to 0,15 m/s
lock in
2
Eq. 4-17
On a different perspective, the research centred in the Millennium footbridge
[16] has led to an interpretation of lock-in as a phenomenon associated with the
generation of a negative damping dependent on the number of pedestrians on
the bridge. The triggering number of pedestrians for lock-in, that is the number
of pedestrians N L that could lead to a vanishing of the overall damping producing
a sudden amplified response, has been defined as a function of the structural
damping ratio ξ , of the modal mass m*, of the natural frequency f , and of a
constant k as
8 πξ m* f
N L = Eq. 4-18
k
On the basis of the Millennium footbridge tests, Dallard et al. [16] derived the
constant k to be approximately equal to 300 Ns/m over the range 0,5-1,0 Hz.
Recent experiments on two footbridges in Coimbra and Guarda, Portugal [17]
have shown the adequacy of the Millennium formula to describe the triggering for
lock-in. Amplitudes of acceleration of the order of 0,15-0,2 m/s2 have been
observed in correspondence, suggesting that the two approaches may be related.
21
5.2 Response measurements
22
23
members should be retained, and the group response should be the highest of
the peak responses recorded.
5.3 Identification tests
The identification of modal parameters, i.e., natural frequencies, vibration modes
and damping coefficients can be based on forced, free or on ambient vibration
tests.
24
The functions H ij (f ) are intrinsic of the system and form the basis for
application of a System Identification algorithm (in the frequency domain)
to extract natural frequencies f k, vibration modes φk and associated
damping coefficients ξ k , while γ 2 (f ) provides a measure of the correlation
between the measured input and response signals.
Considering a viscous damping model and response measurements
expressed in terms of accelerations, the frequency response functions
H ij (f ) relate to the modal components of mode k , (φi )k and φ j k , at
sections Ri and R j , respectively, through
− f (φ ) (φ )
2
H (f) = i k j k
Eq. 5-4
(f − f 2 ) + i (2 ξ k f k f)
ij 2
k
25
26
27
5.4 Instrumentation
28
29
6.2 Modification of mass
No further background information.
6.3 Modification of frequency
Possible strategies for modification of structural frequency comprehend, for
example, the replacement of a reinforced concrete deck slab formed by non-
continuous panels by a continuous slab, or the inclusion of the handrail as a
structural element, participating to the overall deck stiffness.
Other more complex measures can be of interest, like the addition of a stabilising
cable system. For vertical vibrations, alternatives are the increase of depth of
steel box girders, the increase of the thickness of the lower flange of composite
girders, or the increase of depth of truss girders. For lateral vibrations, the most
efficient measure is to increase the deck width. In cable structures, the
positioning of the cables laterally to the deck increases the lateral stiffness. In
cable-stayed bridges, a better tensional behaviour can be attained by anchoring
of the cables at the central plane of the bridge on an A-shape pylon, rather than
anchoring them at parallel independent pylons.
6.4.1 Introduction
No further background information.
30
e
h t d e h t
s f t n e
t h
e
d i o g
n a
n n n n t n r
e c
l
l n e
r l i g
f n e f g o u o
o e / m o i
i o o i m m o i
n l
l i
e r e b s o t t t p m a v
g t u )
m n e a c e p e e
l c e r a .
d
i n q z )
u a m p d r e p m t p t
e m e h f
r o e y e
r i r
b i y a s f a s
f v e e
B r H
C f ( N p
s ( T P v d y m
T d s i y
E d s o b R
f .
o f e
o o d
i t m
s t s o n m
f x a s m e 3
u
o o o r s
a m ,
8
u s b s m n e
n r e a d i o c d
i
t e d 0 t a n
l
n p i 0 m e a p u
s
s 6 , l
i r s o
o
c r m b
n f d a i i r
n a i d a
a , e , o
d e r v
e e l r
p d d
r l s e m
a
J e i i y t e a
d p a u n r d o
, y g u t o r s g e r .
r f
i
e a x
t q T e f
i t g m
s 4 s o l l g . n o l a d
l e m
g n 3 - b n i
d
i a 1 e l a
r d i r a r a c 9
r u ,
r 3 p l e e h e t % e e ]
B 9 +
s 5 b e t n s d
r n 7 d t d 2 5
- , a e a o i
u l r
, i a o
T 0 2 4 C t
s L o
T s g c 0 g L e
r t 1
[
n
i e d
l
r t
l l d s , e t
e a
e p n e r g t
n i
f e
m B e n e i
t
m , t o u
t p h e g
c i d
a e
g
s h i
t p
t
c l m g r
i e i
r
D d d i o e l a a e w
c d b
r i n 3 e w r s a i w g e
r
e b a 8 g , t c
i t
r s h k h
z
t t
i p , i h h s
n o s t
r e s c 0 t ]
r o 6
, s 3 r t
h c r o e v a a 2 n 7
o r 1
B f 5 1 3 2 a o c V 2 m e 5 o [
s e
a s r
l f a e e
e e o m w g
β e s
t g d
d h k i
a
r s n r
e c 0 b
t n d a n a 0 e
S i
l e e u 9
r r
e y m t e h
l , g t
e
t B a h a s i
d , t y c
s b r l c r
i
n n
e e n a a t e t o
w g a e , l
l c
i r p h
g d
h d p
s 9 b k
c e t
r e m
v a e e
i ]
7
c i
r 9
, 0 a e e t e t
t
i 1
S b 1 1 2 C d s V 4 d w f [
31
r d ,
t y e e r
e a d
r n e s e
d w i u
n r g t p a h
u o
s N x l m g w t
l a a i
e ,
n e
s o
b a c d n
i g n
t k o e
n l c t
m g a e i
t r s h
g 0 d
0 g ]
p e s i
o d
j i
r 8
, s e
t
r
e v a e 0 e t d
t i
r 7
M b 0 3 S V 1 m w i
6 f b 1
[
d s , s s
n e r g
n a %
o m s n e i
u t m l s
t h ; 5 a
p s
e i
g u
m l m g g ≈ r a
d r
t
a
e l a a i
e k f u
t
c m
i
r a b a c
i d w o l
n l c t 0 s u
b
t e a i r s h 0 o r a
o g p e
e
t
r e s c 0 i t d
s o ]
8
o r 3
, s 8 t e v a a 1 t f
a
F l 4 1 2 S V a o m
2 m e ≈ r 1
[
r d ;
e e r f
d
r n e o
i
g t p o l s
u s f y s
o l b e
d e x l m
i a a
e g o l a a t f r e a g
t d
i z a m
c d r o u l s r m
n i
l r
y r n b
l
a i
t
r s s t
a a u
t i t
p o
p b
t
H
4 a 4
p , e
c
i
t e s s % c
d e
r c p 7 ]
p 8 s 7 e r v a a 0 u u
c r m , 9
i u o
m , t e , r o a 2
o
S s f 1 1 4 S V 1 m m 1 t
s m n t
I s d 1 1
[
h
t
i
d e w
e v
i
, y t d r
e
m e 0 a c e
i g y
e i n
,
3 t l - t p
t
a i
h d
r a o 5
s
- a
c
i f m
a
h
c b m t , e
l i
t m , d ]
r t r 0 7 b r e D
s M 1
o o e
o G , 1 a e R 2
F f 1 1 C V 1 T M [
g
n o
i
e s
c i
r g p t
n a k m %
h 0
r P
a
, s t
s i
0
4
a 3
d ,
F e a w
d a 0
l -
e g s 2
d
d i l m r f n
a
e r 2
, %
s u 0
e r r a
c
i d p o
e
s p a t 3 ,
5
d b
t 5 e t e s s e
r c m -
a
t o 9 d r n m a r c u 3 ]
o
S f
,
1 -
r
i
g
e
V
u a e
T d m p n r
I t o ,
r 4
s f 8
[
,
e h l
t
d i a
r %
g h n 4 , t
o
d
i n t
i a w g u
t , )
r
b a w g s k c 0 l
a
p 0 u r %
t
o s D k D 0 r
t m e
t ,
5
l 0 M s o
o
f a
r
M
T 0 l 0
T g
0 k f a 0 %
r l
(
l 0 1 0 d 2 )
o t l 5 a e g m l
n n l l a 0
c s 6 s i d a
i
r e
c a a a
c i
r i c r 1 i
e s t r e
s 7 a n %
5
o
p , f n
r c
i
é s t t t e a t
l i
f 1 4 2 6 h e
t r r a a e
s s
v a d
r
c m 3 d r
o r
a 8
, 9
, 2
, 0 c
r a e e l n n a o n % e
v
]
8
S P 0 1 2 1 A L V V 1 m 2 m a I d t a 3 ( [
32
r -
a
r e % l %
n d
r h 5 a
b ; a i t
i c 3
l , i , %
m
i 4 7
1 p
s
g w 6
a 0
r r t 0
6
o 3 ,
, 3 l / e m e m -
s
d D t
s C
ê ,
2
; ; a
r h
c
D
M n a o v o %
l
0 7 t r a r d r
n e
I 8 n a T M
l g T d f n f 3
g , 0
, e l e g a g o
e d c l l a
r k s i
o
i
r 1 3
; ;
w
o
l a a
c e 0 a
c a n n t
i
r b 5 4 4 m l r i
t t 0 i e p % p
% ]
d t 8 0 a e
t r a 8 t
r r
c m 4 m 7
e o
o , 7
, 7
, 1 h a e l 4 e a
n d t a 2
o d , 1
P f 0 1 2 1 S L V 1 1 v I 2 [
33
from the dynamic equations of motion and are available in the literature [18],
[23].
3 μ
ξ opt = Eq. 6-3
8 (1 + μ )
3
34
g
5. Calculation of pendulum length L = , where g is the acceleration of
(2 π f )
d
2
where μ is the previously chosen TMD mass ratio and κ is the geometry coefficient,
defined by
B + 2 Hcos β
κ = Eq. 6-9
Leff
with
AH
Leff = 2
H + B Eq. 6-10
AB
35
δ opt
*
δ opt = Eq. 6-11
1 + μ * (1 − κ 2 )
⎨ Eq. 6-12
⎪ B + 2 Hcos( β ) AH
⎪H = − B
⎩⎪ 2 κ 2 AB
Note that since β = π / 2, B is obtained directly from the first equation. Also,
since Ah / Ab = 1 and cos( β ) = 0, H can be extracted from the second
equation.
4. Calculate the cross section areas Ah and Ab from the mass constraint:
( A B + A 2 H )γ = M μ
b h liquid
struct
*
Eq. 6-13
M struct μ *
Ah = Ab = Eq. 6-14
(B + 2 H )γ liquid
The optimum damping of the TLCD should be the same as the analogue TMD.
The TLCD has intrinsic damping due to fluid turbulence. In addition, by inserting
a control valve or an orifice plate in the horizontal tube, the TLCD damping can
be further enhanced. However, there is no specific literature with information
concerning the quantification of TLCD damping, so it must always be obtained
from tests on the TLCD prototypes.
36
1. Take the mean or frequent value of the amplitude of the deck displacement
response X s (estimated, after inclusion of TLD).
2. Calculate the non-dimensional excitation parameter Λ=X /L
s , where L is the
length of the tank in the direction of motion.
3. Calculate the damping coefficient ξ = 0,5 Λ 0 ,35
4. Calculate de frequency ratio χ between the non-linear and the linear TLD
frequency defined by Lamb’s formula:
χ = 1,038 Λ0,0034 for Λ < 0,03 (weak wave breaking)
χ = 1,59 Λ 0,125 for Λ > 0,03 (strong wave breaking)
5. Calculate the water depth, which includes the stiffness hardening parameter
χ , assuming that best tuning is accomplished by setting the TLD frequency
equal to the structure’s (f s):
L ⎛ 4 π Lf s2 ⎞
h0 = tanh −1 ⎜⎜ 2
⎟⎟ Eq. 6-16
π ⎝ g χ ⎠
g – acceleration of gravity (9,81 m/s2)
6. Choose tank width or number of tanks according to the necessary mass ratio
for structural damping. The water mass ratio should be chosen around the
same magnitude as in a TMD, i.e. from 0,01 to 0,05.
For numerical analysis, an equivalent TMD can be used. For very small deck
displacement amplitudes (below 1cm) the active mass, md , may be as low as
around 80% of the total liquid mass [28]. The stiffness k d is obtained from
k d =
χωd,lin )2md . The damping coefficient is ξ d , the same as the TLD.
(
In sum, TLD tuning may be accomplished by taking a mean or frequent value of
the amplitude of ground displacement expected for the structure when in use,
and the remaining parameters (tank length and/or water depth) can be derived
from there.
7 Worked Examples
7.1 Simply supported beam
The verification for reversible serviceability is shown for a pedestrian bridge
having a span length of 50 m.
The bridge has the following properties:
37
The owner demands that medium comfort is guaranteed for weak pedestrian
traffic (d = 0,2 P/m2) and that for very dense traffic (d = 1,0 P/m2), which is
expected for the inauguration of the bridge, minimal comfort in vertical direction
should be guaranteed and a pedestrian-bridge interaction with lateral vibration
should be avoided.
Loading scenario Required comfort
d = 0,2 P/m2 alimit,vert ≤ 1,0 m/s2
n = 50×3×0,2 = 30 alimit,hor ≤ 0,1 m/s2
38
39
40
p(t )
[N/mm²]
41
should be considered with load acting up and down according to the investigated
mode shape directions.
The different load directions are simulating a phase shift of 180° or π for the
pedestrians walking over the bridge. This can be interpreted as full
synchronization between every single pedestrian and the belly of the mode shape
(direction), which he is reaching or just walking over.
The design situation is defined by the combination of a traffic class and a comfort
class. Generally, different design situations should be considered although this
example is limited only to one design situation. As the footbridge connects the
Minden town centre with a recreation area in a park traffic class TC2, weak traffic
with 0,2 P/m² is chosen (according to section 0) in combination with the comfort
class CL1, maximum comfort, with less amplitudes than a = 0,5 m/s².
Table 7-3: Description of the design scenario
42
43
44
Considering the location of the footbridge close to a school, although not linking
very relevant areas in town, several scenarios should be investigated. In the
current example only two design situations are analysed, corresponding to: 1-
the inauguration of the footbridge, with a traffic class TC4 (d = 1,0 P/m2) and a
minimum comfort class (maximum vertical accelerations of 1-2,5 m/s2 and
lateral accelerations of 0,3-0,8 m/s2); 2- commuter traffic (TC2, d = 0,2 P/m2)
and medium comfort class (maximum vertical accelerations of 0,5-1 m/s2 and
lateral accelerations of 0,1-0,3 m/s2). Although the measured damping ratios
after construction of the footbridge (presented in Table 7.4) are higher, a value
of 0,6 % was considered at design stage.
The harmonic load models for pedestrian streams are then defined in accordance
with the guidelines and are systematised in Table 7-6 for the two design
situations. It should be noted that for the design situation 1 the added mass
associated with pedestrians represents 7,6 % of the total bridge mass, therefore
the footbridge natural frequencies should be re-calculated with the footbridge
loaded. That has not been done within the current example for simplification.
Table 7.6: Harmonic load models for pedestrian streams
45
8 References
[1] BS5400, Part 2, Appendix C, Vibration Serviceability Requirements for Foot
and Cycle Track Bridges. British Standards Institution, 1978
[2] DIN-Fachbericht 102, Betonbrücken. Deutsches Institut für Normung, 2003.
[3] ENV 1995-2, Eurocode 5 - Design of timber structures – bridges. European
Committee for Standardization, 1997.
[4] Guidelines for the design of footbridges. fib bulletin 32, November 2005.
[5] EN 1990, Eurocode 0 – Basis of structural design. European Committee for
Standardization, 2002.
[6] Charles, P.; Bui, V., Transversal dynamic actions of pedestrians &
Synchronisation. Proceedings of Footbridge 2005 – 2nd International
Conference, Venice 2005
[7] Schneider, M., Ein Beitrag zu fußgängerinduzierten Brückenschwingungen,
Dissertation. Technische Universität München, 1991
[8] Maia, N. et al., Theoretical and Experimental Modal Analysis. Research
Studies Press, UK, 1997.
46
47
[29] Yu, J.-K., T. Wakahara, and D. Reed, A non-linear numerical model of the
tuned liquid damper. Earthquake engineering and structural dynamics,
1999. 28: p. 671-686.
[30] Statistisches Bundesamt:
http://www.destatis.de/basis/d/gesu/gesutab8.php, Mikrozensus 2004
[31] Živanović, S. et al., Vibration serviceability of footbridges under human-
induced excitation: a literature review . Journal of Sound and Vibration 279
(2005), pp. 1-79
[32] SETRA/AFGC, Comportement Dynamique des Passerelles Piétonnes
(Dynamic behaviour of footbridges), Guide (Draft). December 2004.
[33] Peeters B., System Identification and Damage Detection in Civil
Engineering, Ph.D. Thesis. Katholieke Universiteit Leuven, 2000.
[34] Brincker R., Zhang L. and Andersen P., Modal identification from ambient
responses using frequency domain decomposition, Procceedings of IMAC-
XVIII, International Modal Analysis Conference, pp.625-630, San Antonio,
Texas, USA, 2000.
[35] Van Overschee P., De Moor B., Subspace Identification for Linear Systems:
Theory-Implementation-Applications. Kluwer Academic Publishers,
Dordrecht, The Netherlands, 1996.
[36] Fujino Y., Pacheco B., Nakamura S. and Warnitchai P., Synchronization of
Human Walking Observed during Lateral Vibration of a Congested
Pedestrian Bridge. Earthqauke Engineering and Structural Dynamics, Vol.22,
pp.741-758, 1993.
[37] http://www.bwk.kuleuven.ac.be/bwm/macec/index.html
[38] http://www.svibs.com/
48
⎡ n
⎤
F p,vert (t) = P ⎢1 +
⎣
∑ α
i =1
sin(2 π i f s t − φi )⎥
i,vert
⎦
Eq. 9-1
n
F p,lat (t) = P ∑ α
i =1
i,lat
sin(π i f s t − φi )
Eq. 9-2
F p,long (t) = P ∑ α
i =1
i,long sin(2 π i f s t − φi )
Eq. 9-3
49
Bachmann
α1 = 0,4 – 0,5; α2 = α3 = 0,1 for f p = 2,0 - 2,4 Hz Walking – vertical
& Ammann
α1 is
frequency
Kerr α1, α2 = 0,07; α3 = 0,2 Walking – vertical
dependant
50
Figure 9-2: Typical vertical force patterns for slow jogging and running [1]
The proposed load model is a single load P (t ,v ) which is moving across the bridge
with a certain velocity v of the joggers. That is the reason why this load model is
very difficult to apply with currently used commercial structural analysis
programs and may only be modelled by specialised software (e.g. ANSYS,
DYNACS).
51
P [N]
1250 — —
Vertical reduction coefficient ψ
52
53