Beruflich Dokumente
Kultur Dokumente
Abstract: In 1999 and 2000, the two vibration incidents at the Paris Passerelle Solferino Bridge and the London Millennium Bridge triggered
a major revision of existing knowledge concerning footbridge response to pedestrian-induced actions. In the last 15 years, an incredibly large
amount of research has emerged on the topic. Although researchers have provided many valuable scientific contributions regarding the under-
Downloaded from ascelibrary.org by University of Central Florida on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.
standing and modeling of pedestrian-induced vibrations of footbridges, there is still a need to determine what real improvements have been
achieved in design procedures. This article provides a critical overview of the methodologies proposed over the last four decades, as well as
their code implementations, and summarizes the actual advances available that consultants can use to their advantage in bridge design.
DOI: 10.1061/(ASCE)BE.1943-5592.0000825. © 2015 American Society of Civil Engineers.
Author keywords: Footbridges; Walking-induced vibrations; Crowd–structure interaction; Walking loading models; Response models.
Ground reaction forces (GRFs) are the forces exerted by a walker Fig. 2. GRFs exerted by a walker to a footbridge and discrete Fourier
on the walking surface. For example, the vertical GRF exerted by a series representation of the vertical component
walker on a footbridge is shown in Fig. 2. GRFs are characterized
by a rather large level of randomness, which is the result of intersub-
ject variability (i.e., the differences existing between different
walkers) and intrasubject variability (i.e., the nonperiodic behavior
of each walker) (Eriksson 1994; Ohlsson 1982). In addition, GRFs mean of 2 Hz and a standard deviation of approximately 0:20 Hz
are modified by the interaction among walkers and of the walkers (Matsumoto et al. 1972; Ricciardelli and Pansera 2010). One possi-
with the structure (as discussed later in the article). ble approach to the modeling of GRFs is to neglect intrasubject vari-
Although it is not fully periodic, walking is a rather repetitive ac- ability, and assume that the force is periodic in time and moving in
tivity, and so are the GRFs. Natural (or undisturbed) walking fre- space at a constant speed; in so doing, the vertical and lateral com-
quency is reasonably described by a Gaussian distribution with a ponents can be written as
" #
X1
L
FV ðx; tÞ ¼ W 1 þ DLFV; j ðsin ð j 2 p fw t t Þ f V; j Þ d ðx vðt t ÞÞ Hðt t Þ H t t
j¼1
v
" #
X1
L
FL ðx; tÞ ¼ W DLFL;j ðsin ðj p fw t t Þ f L; j Þ d ðx vðt t ÞÞ Hðt t Þ H t t
j¼1
v (1)
where W is the weight of the walker; DLFV;j and DLFL;j are the jth x is the position of the pedestrian on the bridge and v ¼ fw l is the pe-
dynamic load factors (DLFs); fw is the walking frequency; f j is the destrian speed, with l being the stride length; HðÞ is the Heaviside
phase lag of the jth harmonic; d ðÞ indicates the Dirac delta function; function; t is arrival time; and L is the span length. The equations in
the vertical GRF and by Pizzimenti and Ricciardelli (2005) and of j 1 , and for L in the range of 10–50 m. The model of Blanchard et
Ricciardelli and Pizzimenti (2007) for the lateral GRF. al. (1977) is a simple correction to a static load, and was easily
Brownjohn et al. (2004b) obtained the PSDFs of the first six ver- incorporated into codes of practice (discussed later in this article).
tical DLFs from the Fourier spectra of measured forces; converted From a mathematical point of view, the same approach can be
to GRF, this becomes applied, and indeed was applied, also to the lateral response; in that
h i2 case, however, the idea of the correction of a static deflection loses
j f =ð jfw Þ1jCV; j
AV; j þ BV; j exp DV; j its physical meaning. What the authors failed to recognize is that
SFV; j ð f Þ ¼ 32 W 2 (2) the maximum transient response does not depend on L and j 1 sepa-
fw
rately, but rather on their product (Ricciardelli and Briatico 2011).
Nevertheless, the graphs they produced are quite accurate and con-
where AV;j , BV;j , CV;j , and DV;j are curve-fitting parameters. For
each harmonic, f is in the range of ðj 0:25Þ fw –ðj þ 0:25Þ fw . sistent with the fact that different pairs of L and j j having the same
Živanovic et al. (2007) extended the results of Brownjohn et al. product must produce the same value of w .
(2004b) by considering also the DLFs of the first five subharmonics Generally speaking, the maximum acceleration in the ith mode
of the walking force. in resonant conditions (i.e. jfw ¼ fi ) can be written as
Pizzimenti and Ricciardelli (2005) and Ricciardelli and Pizzimenti DLFj W
(2007) obtained the PSDFs of the lateral GRFs around the first five har- €y max;i;j ¼ w ðɛi Þ ¼ €y stat;i;j w ðɛi Þ (6)
2 j i mi
monics of the walking frequency as
2 !2 3 where ystat;i;j ¼ ðDLFj WÞ=ð2 j i mi Þ is the stationary acceleration
F~ L; j 2AL;j
2
f =ð j fw Þ 1 5
SFL; j ð f Þ ¼ pffiffiffiffiffiffiffi exp 42 (3) amplitude in the ith mode due to the jth resonant harmonic, mi is the
f 2p BL;j BL; j ith modal mass, and w ðɛi Þ is a transient resonant response
Coefficient (TRRC) that accounts for the load motion. The TRRC is
where F~ L;j is the characteristic value of the area of the PSDF around
2 a stationarity index, and has been found to depend on the parameter
the jth harmonic, AL; j is a parameter used to normalize the PSDF, (Ricciardelli and Briatico 2011), as follows:
and BL;j is a bandwidth parameter. For each harmonic, f is in the 1 2L 1
range of 0:9 j fw –1:1 j fw . ɛi ¼ ji ¼ n ji (7)
i l i
Transient Response to a Single Crossing where n is the number of effective load cycles. In particular, large
values of ɛi are associated with long-span footbridges, a short stride
The simplest approach to the analysis of walking-induced foot- length, a low vibration mode, and high damping, in which case the
bridge vibrations is that of evaluating the transient response to the TRRC approaches 1. On the other hand, small values of ɛi are asso-
crossing of a walker in resonance with one of the bridge vibration ciated with short footbridges, a long stride length, a high vibration
frequencies. This condition is likely to occur for footbridges hav- mode, and low damping, in which case the TRRC is close to 0 and
ing vertical vibration frequencies lower than approximately 5 Hz the response highly nonstationary. Nonetheless, few researchers
or horizontal vibration frequencies lower than approximately recognized ɛi as the governing parameter for the TRRC; instead,
2:5 Hz, and may prove to be the worst load scenario. No feedback many used the variables n and j i separately when studying the
of the footbridge vibration to the walking parameters is consid- transient response. In the following discussion, the maximum
ered, as discussed later in the article. acceleration derived by different authors will be written in the
The first study that recognized the crossing of a resonant walker form of Eq. (6).
as a critical condition is that of Blanchard et al. (1977). The action Blanchard et al. (1977) did not mention, and therefore seem not
of the walker was modeled as a vertical sinusoidal force acting at to have been aware of, the work of Fryba (1973) published four
the footbridge fundamental frequency, moving at a speed corre- years earlier, in which a closed-form solution of w was given as a
sponding to a stride length l ¼ 1:8 m: function of j i and v; this solution proves rather inaccurate. In partic-
ular, it is based on the wrong assumption that the maximum acceler-
l ation occurs when the walker crosses midspan, but this is not the
FV ðtÞ ¼ DLFV;1 W sinð2p f1 tÞ f f1 t (4)
2 only source of inaccuracy. The solution of Fryba (1973) does not
acknowledge the fact that the TRRC depends on ɛi , as it expresses it
where W ¼ 700 N, f1 is the footbridge fundamental frequency, as function of L and j separately; in fact, it can be manipulated to
DLFV;1 ¼ 0:257, and f ðxÞ is the normalized fundamental mode take the form
occurs when the walker is at midspan was removed, and the exact
time when this occurs was evaluated numerically. Nevertheless, side for all values of ɛ1 . This model was adopted by some codes of
the solution of Rainer et al. (1988) also proves inaccurate, prob- practice (discussed later in this article).
For footbridges having more than one span, numerically
ably due to the inaccuracy of the response equation it is based on.
Based on the measurements carried out on two footbridges, obtained correction factors exist. In particular, Blanchard et al.
(1977) proposed a correction coefficient for up to three spans;
Allen and Murray (1993) suggested that the constant value w ¼ 0:7
Pimentel and Fernandes (2002) proposed a correction coefficient
can be used, regardless of the bridge characteristics.
for the case of two spans, to be used in the model of Rainer et al.
Grundmann et al. (1993) suggested the approximated value
(1988). The coefficients proposed are in agreement with each other
w ðn; j 1 Þ ¼ 0:6½1 exp ð2p n j 1 Þ (9) and are for reducing factors: For the case of two spans, they are in
the range of 0:7–0:96 for ratio between the lengths of the spans in
derived from interpolation of the solution of the modal equation of the range of 1–0:2.
motion (in square brackets), which was then reduced to 60%. This
model was the first to acknowledge the dependency of w on Response to Multiple Walkers
ɛ1 ¼ n j 1 , but the reason for the reduction was not explained.
For floors, Young (2001) proposed an equation for the TRRC An alternative situation is that of a footbridge crossed by more than
similar to that of Grundmann et al. (1993), as follows: one walker. In that case, the characteristics of the walkers have to
be described on a probabilistic basis, and the response can be eval-
w ðn; j 1 Þ ¼ 1 exp ð2p 0:275n j 1 Þ (10)
uated either in the frequency domain through random vibration
theory, or in the time domain through Monte Carlo simulations.
Finally, Ricciardelli and Briatico (2011) found an approximated The first attempt to evaluate the load exerted by a number N of
closed-form for the TRRC, as follows: walkers is that of Matsumoto et al. (1978), who noted that when the
qffiffiffiffiffiffiffiffiffiffiffiffiffi
footbridge is crossed by a stream of walkers having equal frequency
ɛi 2 þ exp ɛ p þ arctan 1
w ðɛi Þ ¼ 1 þ ɛ i and phases uniformly distributed between 0 and 2p (uncorrelated
1 þ ɛ2i i
2 ɛi walkers), the RMS response varies with the square root of N (or with
(11) the square root of the walker density d ¼ N=BL, B being the breadth
of the deck). An opposite situation is that of fully correlated walkers
which proves quite accurate, with an error with respect to the nu- (i.e., walkers stepping in phase), for which the RMS response varies
merical solution never exceeding 6 103 for ɛi in the broad range with N.
of 104 –10. From experiments on a 10-m footbridge, Eriksson and Ohlsson
In Fig. 3 the values of w ðɛ1 Þ proposed by different authors are (1988) found that when the walkers step in phase the total vertical
compared, indicating a clear disagreement among one another. The response is obtained multiplying the RMS response to a single
Fig. 3. Comparison of the TRRC from different authors; the light-gray area shows the typical range of ɛ1 for common footbridges
Fig. 4. Variation of the mean and standard deviation of the walking frequency with the walker density, d (data from Butz et al. 2007)
d r ðxs Þ ¼ 3 104 y2s 2:28 102 ys þ 0:56 (22) cient calibrated based on Millennium Bridge data. The model is ap-
plicable for vibration frequencies in the range 0:75–1:25 Hz, and it
Eq. (21) can be used to calculate the footbridge response is suggested that a value of DLFL1 ¼ 0:044 be used in combination
accounting for walker–structure interaction. with an average body mass of 70 kg and a standard deviation of the
walking frequency of 0:1 Hz.
Roberts (2005) solved the walker–footbridge coupled equations
Stability of motion, setting the displacement of the walker equal to that of the
footbridge. With uniform distribution of fw =f1 between 0:8 and 1:2,
When the footbridge vibrates laterally, walker–structure interaction the critical number of pedestrians turns out to be
can bring dynamic instability.
The full-scale measurements carried out by Arup in the after- c1 m1 j 1
Nc ¼ 4 (26)
math of the Millennium Bridge failure showed that the amplitude of mo
the in-quadrature component of the lateral force was proportional to
the bridge velocity (Dallard et al. 2001), thus having the characteris- where c1 is a coefficient taking into account the averaging of the fre-
tics of a viscous damping force. This led to the now well-known quency ratio; it depends on the footbridge damping, and for practi-
Arup formula, as follows: cal values of this in the range of 0:01–0:02, it varies between 7:27
and 9:06; other values are given in Roberts (2005) in the form of a
m1 f1 j 1 table.
Nc ¼ 8p (23)
k Ingólfsson et al. (2012), based on the criteria of the Service
d’Études Techniques des Routes et Autoroutes (SETRA 2006),
Eq. (23) allows calculation of the number Nc of pedestrians that, Danbon and Grillaud (2005), and Charles and Bui (2005), consid-
when uniformly distributed on a footbridge having a sinusoidal first ered as critical the condition that the acceleration amplitude reaches
mode shape, produces an in-quadrature component of the lateral a limit value a1 . Assuming that the walkers are uncorrelated, at reso-
force that balances the structural damping force, thus producing nance they obtained
instability. The coefficient k in Eq. (23) was found experimentally
to be equal to 300 Ns=m, in a range of vibration frequencies of p2 m21 j 1 a21
Nc ¼ (27)
0:5–1:0 Hz. 10:82 DLFL1
2 m2 g2
o
In the following discussion, other stability criteria that appeared
after that of Eq. (23) will be discussed, and for purpose of compari- It is suggested that the limit acceleration be set equal to
son, these have all been translated in terms of critical number of 0:10 m=s2 , and that for sparse and moderately dense crowds the
pedestrians and put in a form similar to that of Eq. (23) and with the value DLFL1 ¼ 0:05 be used.
same symbols. They all refer to vibrations occurring in the first Finally, Piccardo and Tubino (2008) investigated parametric res-
mode. onance as the cause of instability, and from solution of the Mathieu
Newland (2003) wrote the modal equilibrium equation of the equation they found
footbridge subjected to the inertia force produced by resonant
walkers, assuming that these are all in quadrature with the bridge m1 f12 j 1
Nc ¼ 32p 2 (28)
motion, and obtained an expression of the critical number of b a1 mo g
walkers, as follows:
in which a1 is the slope of the variation of the DLF with the ampli-
m1 j 1
Nc ¼ 4 (24) tude of the oscillation of the footbridge. Results of experiments pre-
ab mo sented in Dallard et al. (2001) suggest a value of a1 ¼ 2 m1 .
Eqs. (23)–(28) contain some similarities, and some differences,
where mo is the mass of each walker, a is the ratio of the amplitude that have been widely discussed in Ricciardelli (2014).
of oscillation of the walker’s center of mass to the amplitude of os- Eqs. (23)–(28) can all be written in the form
cillation of the footbridge, and b is the probability of a pedestrian
being synchronized with the footbridge motion. In a range of vibra- m1 j 1
Nc ¼ a (29)
tion frequencies of 0:75–0:95 Hz, a value of a ¼ 2=3 is suggested. mo
As to b , reference is made to the experiments carried out by Arup
(Dallard et al. 2001), which show a strong variability, between 0:28 with a depending on the criterion adopted; these are summarized in
and 0:87, when the amplitude of oscillation is in the range of Table 1, together with the values it takes when the basic parameters
1
32p 2
b a1 g
a
mo ¼ 70 kg.
b
f1 ¼ 0:5–1 Hz.
c
Phase lag ¼ p =2; a different phase lag increases this value.
d
fs ¼ 0:75–1:25 Hz.
e
This depends on the structure, as an example, for the Millennium Bridge
south span a ¼ 8:5, and for the Millennium Bridge center span a ¼ 11:3.
suggested by the authors are used, and with the minimum and maxi-
mum values coming from the broader combination of the values of
the basic parameters available in the literature.
First, it is observed that there is no consistency in the depend-
ency of a on the oscillation frequency. In Eqs. (24), (26), and (27), a
is independent of f1 ; in Eq. (23), it is proportional to f1 ; and in Eqs. Fig. 5. Variation of the motion-induced added mass, j p , and damping,
(25) and (28) it is proportional to the square of f1 . Then, it is noticed r p : (a) with the range of variation of the phase lag of the motion-
that there is a quite large scatter in the values of a. Some differences induced force; (b) as a function of the ratio of the oscillation frequency
can be explained based on the hypotheses used to derive the stability to the undisturbed walking frequency over a fixed floor (data from
criterion. As an example, the larger values of a in Eq. (26) with Ricciardelli et al. 2014)
respect to those in Eqs. (24) and (25) can be ascribed to the fact that
Eq. (26) was derived under the assumption of uniformly distributed
phases of the walkers, whereas Eqs. (24) and (25) were derived Assuming that unstable vibrations start from rest, the values of
under the assumption that the walkers are all in quadrature with the r p ðfc Þ and j p ðfc Þ appearing in Eqs. (30) and (32) can be obtained
footbridge motion. Some other differences are more difficult to from extrapolation of Eq. (21) to a zero amplitude of vibration, as
explain. follows:
An alternative criterion was presented in Ricciardelli (2014). This 3 2 3
leads to the finding that under the same hypotheses made earlier of si- f f f
j p ðf Þ ¼ 1:77 þ 7:77 9:55 þ 3:19
nusoidal vibration mode and uniformly distributed walkers, the fw fw fw
vibration frequency of the combined walker–footbridge system is
1:11 f
r p ðf Þ ¼ 9:17 sin exp 2:89 þ 0:56 (33)
fs f =fw fw
fc ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (30)
1 þ m r p ðfc Þ
which are plotted in Fig. 5(a).
The results coming from Eq. (32) can then be checked with those
with m ¼ ðd BLmo Þ=ð2m1 Þ being the ratio of the total walker mass of other stability criteria given in Table 1. When the minimum value
to the footbridge mass. Eq. (30) must be solved iteratively, as the of the pedestrian-induced damping j p ¼ 0:4 in Fig. 5(a) is used,
added mass coefficient at the right-hand side of the equation needs then a ¼ 5. This result compares very well with Eq. (23), which
to be evaluated at the vibration frequency. A similar equation had confirms the quite good agreement of the results coming from the
already been used in Bocian et al. (2012). full-scale measurements on the Millennium Bridge (Dallard et al.
Then, the stability condition is 2001) and those coming from laboratory tests using a treadmill
(Ricciardelli et al. 2014), also in consideration of the totally differ-
j s þ m j p ðfc Þ > 0 (31) ent experimental approach used for the two studies. If the average
mass of 74:4 kg and the average walking frequency of 0:857 Hz of
where j p is the damping coefficient associated with the walker the walkers participating in the tests is used, then the maximum neg-
action, referred to the walker mass (Ricciardelli et al. 2014). The ative j p of 0:40 is translated into cp ¼ 281 Ns=m, whereas the
critical number of walkers can be calculated as value of j p of 0:36 occurring for fs =fw ¼ 1 is translated into
cp ¼ 288 Ns=m. These differ only very little from the value of
m1 j 1 300 Ns=m derived from full-scale tests. Comparison can also be
Nc ¼ 2 (32)
j p ðfc Þ mo made with the value cp ¼ 140 Ns=m found numerically in
Macdonald (2008), from a model excluding the possibility of syn-
which can be put in the form of Eq. (29), with a ¼ 2= j p ðfc Þ. chronization. The ratio of the cp value found in Macdonald (2008)
Mj
EN EC5 Vert 0:23 0:6 L B 200
Mj
EN EC5 Lat 0:18 0:6 L B 50
Mj
FIB Lat/Vert 3 Rainer et al. (1988) or Grundmann et al. (1993)
FIB Vert 0:225 N Rainer et al. (1988) or Grundmann et al. (1993)
pffiffiffiffiffiffiffi
SETRA and HIVOSS Vert jffiN
10:8pffiffiffi d < 1 w=m2 2 280 1
1:85 N d 1 w=m2 p m1 2 j
pffiffiffiffiffiffiffi
SETRA and HIVOSS Lat jffiN
10:8pffiffiffi d < 1 w=m2 2 35 1
1:85 N d 1 w=m2 p m1 2 j
pffiffiffiffi c rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
HIVOSS Lat/Vert N k ðf Þ
kp mC2 k1 ðfj Þ j j 2 j kf
j
Note: Ne ¼ 1 refers to the single walker, Ne ¼ 3 refers to the group of walkers, Ne 6¼ 1; 3 refers to a stream of walkers, Dir = the direction of vibration,
Vert = vertical, Lat = lateral).
a
v ¼ 3 m=s.
b
M is the total mass of the footbridge; n ¼ L=0:9 m.
c
In the HIVOSS model, the parameters contained in €y max ð1Þ are function of d , that is, N.
Fig. 6. Comparison of Wðf1 Þ for the vertical component; the gray and light-gray areas show the typical range of frequency for the first and second har-
monics for a typical walker
mode, and kf represents the variance of the loading of a single pe- In Figs. 6 and 7, the values of Wðf1 Þ for the vertical and lateral
destrian; these values are given in table form for the vertical and lat- components proposed by the different codes and guidelines are
eral components for crowd densities equal to: d < 0:5 walkers=m2 , compared. The gray and light-gray areas show the typical range of
d ¼ 1 walkers=m2 and d ¼ 1:5 walkers=m2 . In this model it is not frequency for the first and second harmonics for a typical walker.
possible to isolate the function Wðf1 Þ, because it is incorporated into The frequencies at which Wðf1 Þ take large values qualify the ranges
k1 ðfj Þ and k2 ðfj Þ. where resonance may occur, and Wðf1 Þ ¼ 1 means perfect resonant
Fig. 7. Comparison of Wðf1 Þ for the lateral component; the gray and light-gray areas show the typical range of frequency for the first and second har-
monics for a typical walker
pffiffiffiffiffiffiffiffiffiffiffi
response. ENV EC5 (CEN 1995), EN EC5 (CEN 2005), and FIB Ne ¼ 10:8 j N ; the second appliesptoffiffiffiffilarge densities and is inde-
(2005) give only one plot for Wðf1 Þ, which applies to the first and pendent of damping: Ne ðNÞ ¼ 1:85 N . The two equations pro-
second harmonics; therefore, the lower values appearing in Figs. 6 vide the same results for j ¼ 0:03. The scatter among the results pro-
and 7 in the second harmonic range are the effect of the ratio vided by the different equations is quite large; for example, for
between the DLFs, that is, DLF2 =DLF1 , being the resonant acceler- N ¼ 20, Ne is in the range of 2–8, and for N ¼ 100, Ne is in the range
ation, and €y max ð1Þ, computed using DLF1 in both cases. As a matter of 10–19.
of fact, the differences between the different curves in Figs. 6 and 7
are minor.
Fig. 8 compares the values of Ne ðNÞ proposed by different guide-
Lateral Stability
lines and codes of practice, indicating a clear disagreement among
the various proposals. The light-gray area shows the range of Ne The problem of lateral stability is also considered by some codes
ffiffiffiffi of full correlation (Ne ¼ N) and of full uncorrela-
between thepcases and guidelines. In particular, FIB (2005), SETRA (2006), HIVOSS
tion (Ne ¼ N ). FIB (2005) suggests to use Ne ¼ 3 for N 10, (Heinemeyer et al. 2009), and AASHTO (1997, 2009) give a value
Ne ¼ 0:225 N for a low-density stream, and Ne ¼ 0:27 0:46 N to for the critical number of walkers as defined in Dallard et al. (2001)
account for synchronization in higher-density streams; in the latter
pffiffiffiffi (described earlier in this article).
two cases the values of Ne lie below the lower limit given by N , up The U.K. National Annex to Eurocode 1 (BSI 2008) pre-
to N ¼ 70 and N ¼ 20, respectively. SETRA (2006) and HIVOSS scribes that all the modes with lateral natural frequencies below
(Heinemeyer et al. 2009) give two equations, depending on d (see 1:5 Hz have to compare their pedestrian mass damping parameter,
Table 2); the first applies to low densities and depends on damping: as follows:
Conclusions
References
This article provided a critical review of the available methods for
assessing footbridge serviceability against walking action. In partic- Allen, D. E., and Murray, T. M. (1993). “Design criterion for vibrations due
ular, single-walker models, multiple-walker models, interaction to walking.” AISC Eng. J., 30(4), 117–129.
models (inter-walker and walker-structure), and instability models AASHTO. (1997). Guide specifications for design of pedestrian bridges,
were addressed, together with current design procedures incorpo- Washington, DC.
rated into codes of practice and guidelines. Models originally pre- AASHTO. (2009). Guide specifications for design of pedestrian bridges,
sented in different forms were made homogeneous for purpose of Washington, DC.
comparison. The investigation was limited to the models allowing Blanchard, J., Davies, B. L., and Smith, J. W. (1977). “Design criteria and
practical use, although they are also of great interest, the more analysis for dynamic loading of footbridges.” Proc., Symposium on
Dynamic Behaviour of Bridges, Transport and Road Research
advanced approaches that are not yet ready for implementation into
Laboratory, Berkshire, England, 90–106.
design procedures were not dealt with. Some inconsistencies and
Bocian, M., Macdonald, J. H. G., and Burn, J. F. (2012). “Biomechanically
inaccuracies were pointed out. inspired modelling of pedestrian-induced forces on laterally oscillating
Two general conclusions can be drawn, the first being that in structures.” J. Sound Vib., 331(16), 3914–3929.
spite of the impressive amount of research developed in the last 15 Brownjohn, J. M. W., Fok, P., Roche, M., and Omenzetter, P. (2004a).
years, very few changes have appeared in design procedures. The “Long span steel pedestrian bridge at Singapore Changi Airport—Part
second is that when similar procedures are compared with one 2: Crowd loading tests and vibration mitigation measures.” Struct. Eng.,
another, they often show a quite large and unjustified scatter in the 82(16), 28–34.
results. Brownjohn, J. M. W., Pavic, A., and Omenzetter, P. (2004b). “A spectral
Single-walker models were compared in terms of the TRRC, density approach for modelling continuous vertical forces on pedestrian
and it was shown that the best-fitting numerical solutions is that of structures due to walking.” Can. J. Civ. Eng., 31(1), 65–77.
BSI (British Standards Institution). (1978). “5400—Part 2: Steel, concrete
Ricciardelli and Briatico (2011); this, however, is applicable only to
and composite bridges. Specification for loads.” BS5400, London.
the case of a sinusoidal mode shape. Other approaches seem to be BSI (British Standards Institution). (2008). U.K. national annex to
rather inaccurate. Eurocode 1, London.
For multiple walkers, stochastic frequency domain response Butz, C. (2006). “Beitrag zur Berechnung fußgänger induzierte Brückensch-
models exist for both vertical and lateral vibrations. Although pro- wingungen.” Ph.D. thesis, Shaker Verlag Aachen, Aachen, Germany (in
posed by different authors, these are in formal agreement with one German).
another. Time-domain models also exist, based on the results of nu- Butz, C., et al. (2007). “Advanced load models for synchronous pedestrian
merical simulations. excitation and optimised design guidelines for steel foot bridges
As to the possibility of accounting for interaction, it appears that (SYNPEX).” Project RFS-CR-03019, Final Rep., Office for Official
inter-walker interaction models seem to not yet be ready to describe Publications of the European Communities, Luxembourg.
the variation of GRFs and to model synchronization; the only infor- CEN (European Committee for Standardization). (1995). Eurocode 5,
Brussels, Belgium.
mation available for application seems to be the relationship
CEN (European Committee for Standardization). (2005). Eurocode 5,
between the walker density and the mean and RMS walking fre- Brussels, Belgium.
quency (Fig. 4) proposed by Butz et al. (2007).Walker–structure Charles, P., and Bui, V. (2005). “Transversal dynamic actions of pedes-
interaction is accounted for through the variation of the DLFs and trians. Synchronization.” Proc., Footbridge 2005, 2nd Int. Conf.,
the in-phase and in-quadrature components (with displacement) of Venice, Italy.
the GRFs with the floor acceleration and velocity; however, it Dallard, P., et al. (2001). “The London Millennium Footbridge.” Struct.
appears that only the latter lend themselves to stability analyses. Eng., 79(22), 17–21.
Stability criteria available in the literature were manipulated so Danbon, F., and Grillaud, G. (2005). “Dynamic behaviour of a steel foot-
as to express the critical number of walkers though one common bridge. Characterization and modelling of the dynamic loading induced
equation, and the values of the coefficient a appearing in such equa- by a moving crowd on the Solferino Footbridge in Paris.” Proc.,
Footbridge 2005, 2nd Int. Conf., Venice, Italy.
tions were compared, using the calibration parameters recom-
Eriksson, P. E. (1994). “Vibration of low-frequency floors: Dynamic forces
mended by the authors, as well as a wider range of these; a quite and response prediction.” Ph.D. thesis, Dept. of Structural Engineering,
large scatter of the results was found. Moreover, the data available Chalmers Univ. of Technology, Göteborg, Sweden.
for the amplitude and phase of the motion-induced force caused by Eriksson, P. E., and Ohlsson, S. V. (1988). “Dynamic footfall loading
a walker on a laterally vibrating floor allowed calibration on an orig- from groups of walking people.” Proc., Symposium Workshop on
inal criterion, in which stability is checked by looking at the sign of Serviceability of Buildings, Vol. 1, National Research Council,
the total damping of the combined footbridge–walker system; the Ottawa, Canada, 497.
tation, harmonization and further development of the Eurocodes.” JRC- response of footbridges.” Can. J. Civ. Eng., 51(1), 66–71.
ECCS Technical Rep., Office for Official Publications of the European Ricciardelli, F. (2005). “Lateral loading of footbridges by walkers.” Proc.,
Communities, Luxembourg. Footbridge 2005, 2nd Int. Conf., Venice, Italy.
Highways Dept. (2002). Structures design manual for highway and rail- Ricciardelli, F. (2014). “Lateral stability of footbridges subjected to crowd
ways, Highways Dept., Government of the Hong Kong Special loading.” Proc., 9th Int. Conf. on Structural Dynamics—EURODYN
Administrative Region, Kowloon Bay, Hong Kong. 2014, A. Cunha, E. Caetano, P. Ribeiro, G. Müller, eds., Porto, Portugal.
Ingólfsson, E. T., Georgakis, C. T., and Svendsen, M. N. (2008). “Vertical Ricciardelli, F., and Briatico, C. (2011). “Transient response of supported
footbridge vibrations: Details regarding and experimental validation of beams to moving forces with sinusoidal time variation.” J. Eng. Mech.,
the response spectrum methodology.” Proc., Footbridge 2008, 3rd Int. 10.1061/(ASCE)EM.1943-7889.0000241, 422–430.
Conf., Porto, Portugal. Ricciardelli, F., Mafrici, M., and Ingólfsson, E. T. (2014). “Lateral pedes-
Ingólfsson, E. T., Georgakis, C. T., Ricciardelli, F., and Jönsson, J. (2011). trian-induced vibrations of footbridges: Characteristics of walking
“Experimental identification of pedestrian-induced lateral forces on forces.” J. Bridge Eng., 10.1061/(ASCE)BE.1943-5592.0000597,
footbridges.” J. Sound Vib., 330(6), 1265–1284. 04014035.
Ingólfsson, E. T., Georgakis, C. T., and Jönsson, J. (2012). “Pedestrian- Ricciardelli, F., and Pansera, A. (2010). “An experimental investigation into
induced lateral vibrations of footbridges: A literature review.” Eng.
the interaction among walkers in groups and crowd.” Proc., 10th Int.
Struct., 45, 21–52.
Conf., Recent Advances in Structural Dynamics 2010, M. J. Brennan, I.
ISO. (2007). “Bases for design of structures—Serviceability of buildings
Kovacic, V. Lopes, K. Murphy, B. Petersson, S. Rizzi, and T. Yang, eds.,
and walkways against vibration.” ISO 10137, Geneva.
Southampton, U.K.
Macdonald, J. H. G. (2008). “Lateral excitation of bridges by balancing
Ricciardelli, F., and Pizzimenti, A. D. (2007). “Lateral walking-induced
pedestrians.” Proc. R. Soc. A, 465(2104), 1055–1073.
forces on footbridges.” J. Bridge Eng., 10.1061/(ASCE)1084
Matsumoto, Y., Nishioka, T., Shiojiri, H., and Matsuzaki, K. (1978).
-0702(2007)12:6(677), 677–688.
“Dynamic design of footbridges.” IABSE Proc., International
Roberts, T. M. (2005). “Lateral pedestrian excitation of footbridges.” J.
Association for Bridge and Structural Engineering, Zurich, Switzerland,
Bridge Eng., 10.1061/(ASCE)1084-0702(2005)10:1(107), 107–112.
1–15.
Rönnquist, A. (2005). “Pedestrian induced lateral vibrations of slender foot-
Matsumoto, Y., Sato, S., and Shiojiri, H. (1972). “Study on dynamic design
of pedestrian over-bridges in consideration of characteristics of pedes- bridges.” Ph.D. thesis, Norwegian Univ. of Science and Technology,
trians.” Proc. Jpn. Soc. Civ. Eng., 1972(205), 63–70. Trondheim, Norway.
Murray, T. M., Allen, D. E., and Ungar, E. E. (1997). Floor vibrations due SETRA (Service d’Études Techniques des Routes et Autoroutes). (2006).
to human activity: Steel design guide 11, AISC, Chicago. Footbridges: Assessment of vibrational behaviour of footbridges under
Nakamura, S. (2004). “Model for lateral excitation of footbridges by syn- pedestrian loading, Paris.
chronous walking.” J. Struct. Eng., 10.1061/(ASCE)0733 SIA (Societe suisse des Ingenieurs et des Architectes). (1989). SIA160
-9445(2004)130:1(32), 32–37. (Swiss Standard SIA160: Actions on structures), Zurich, Switzerland.
Nakamura, S., Kawasaki, T., Katsuura, H., and Yokoyama, K. (2008). Strogatz, S. H., Abrams, D. M., McRobie, A., Eckhardt, B., and Ott, E.
“Experimental studies on lateral forces induced by pedestrians.” J. (2005). “Theoretical mechanics: Crowd synchrony on the Millennium
Constr. Steel Res., 64(2), 247–252. Bridge.” Nature, 438(7064), 43–44.
Newland, D. E. (2003). “Pedestrian excitation of bridges—Recent Sun, L., and Yuan, X. (2008). “Study on pedestrian-induced vibration of foot-
results.” Proc., 10th Int. Congress on Sound and Vibration, bridge.” Proc., Footbridge 2008, 3rd Int. Conf., Porto, Portugal, 2–4.
International Institute of Acoustics and Vibration, Stockholm, Tilden, C. J. (1913). “Kinetic effects of crowds.” Proc., American Society of
Sweden, 533–547. Civil Engineers, Vol. 76, ASCE, New York, 2107–2126.
Oeding, D. (1963). “Verkehrsbelastung und Dimensionierung von Tredgold, T. (1825). A practical treatise on railroads and carriages,
Gehwegen und anderen Anlagen des Fußgängerverkehrs.” Ph.D. Taylor, London.
thesis, Bundesminister für Verkehr, Abt. Strassenbau, Bonn-Bad Venuti, F., and Bruno, L. (2009). “Crowd-structure interaction in lively
Godesberg, Germany (in German). footbridges under synchronous lateral excitation: A literature review.”
OHBDC (Ontario Highway Bridge Design Code). (1983). Highway Phys. Life Rev., 6(3), 176–206.
Engineering Division, Ministry of Transportation and Communication, Wheeler, J. E. (1982). “Prediction and control of pedestrian-induced vibra-
Ontario, Canada. tion in footbridges.” J. Struct. Div., 108(9), 2045–2065.
Ohlsson, S. V. (1982). Floor vibrations and human discomfort, Chalmers Young, P. (2001). “Improved floor vibration prediction methodologies.”
Univ. Technology, Division of Steel and Timber Structures, Göteborg, Proc., Arup Vibration Seminar, Vol. 4, Institution of Mechanical
Sweden. Engineers, London.
Okamoto, S., Abe, M., Fujino, Y., and Nakano, Y. (1992). “Characteristics Živanovic, S., Pavic, A., and Reynolds, P. (2005). “Vibration serviceability
of human walking on a laterally shaking floor.” Proc., Jpn. Soc. Civ. of footbridges under human-induced excitation: A literature review.” J.
Eng., 441(18), 177–184 (in Japanese). Sound Vib., 279(1–2), 1–74.
Piccardo, G., and Tubino, F. (2008). “Parametric resonance of flexible Živanovic, S., Pavic, A., and Reynolds, P. (2007). “Probability-based pre-
footbridges under crowd-induced lateral excitation.” J. Sound Vib., diction of multi-mode vibration response to walking excitation.” Eng.
311(1–2), 353–371. Struct., 29(6), 942–954.