Pedestrian Bridges Combining Stainless Steel and Glass-Fiber Reinforced Polymers as

Structural Members

Daia ZWICKY

Juan A. SOBRINO ALMUNIA

Prof., Dr., dipl. Bauing. ETH

University of Applies Sciences
Fribourg, Switzerland

Dr. Ingeniero
PEDELTA s.l.
Barcelona, Spain

Summary
The paper reports on the design and specific details of two similar footbridges in Zumaia (Bask country, Spain) and
Medelln (Colombia). These two bridges are probably some of if not the firsts to combine the new construction materials
stainless steel and fiber-reinforced composites, both as structural members, in a medium span pedestrian structure.
The bridges consist each of two main longitudinal single-span girders, creating open sections over 28.0 m and 25.2 m,
and decks with clear widths of 5.0 m and 3.0 m, respectively. The longitudinal girders are made of Vierendeel trusses in
stainless steel and with variable heights of approx. 1.1 m to 1.4 m, combined with glass-fiber reinforced polymers
(GFRP) as structural webs of the trusses. The Vierendeel trusses have stiff bottom chords with hollow sections, stiff
posts with Pi-sections with slightly variable spacing for the larger bridge and very slender top chords with a stiffened
stainless steel plate. The bridge deck of the larger bridge is made of transverse GFRP I-girders of 300 mm height, with a
spacing of 1.0 m, and longitudinal standard light deck planks of 40 mm height. The transverse girders of the smaller
bridge are welded from stainless steel plates to I-sections of 165 mm height with a spacing of 1.5 m, while the same
deck planks are provided.
The paper comments on the design peculiarities of the very light structures, such as the modeling of structural details for
static and dynamic investigations, buckling analysis of the top chords, ultimate limit and serviceability limit state
considerations and design concepts.
Keywords:

1.

new materials, stainless steel, GFRP, design procedure, dynamics, structural modeling, buckling analysis,
blind rivets.

Introduction

Both new footbridges designed by Pedelta intend to provide new inputs to urban development, although in very
different surroundings. The new Zumaia footbridge spans over a side-channel of the river Narrondo in the city centre of
Zumaia in the province of Gipuzkoa (Baskque country, Spain). The bridge provides an additional pedestrian crossing of
the channel in resident areas between two existing road bridges, which are located approx. 750 m from each other.
The new Juan Bobo footbridge in Medelln (Colombia), on the other hand, lies in the heart of the comunas (ghettos) and
connects highly populated areas. This new footbridge shall provide a sign that the social situation is changing in the
area. Since the local government can not provide the funding for the bridge, the enterprises Outokumpu (Finland) [1]
and Fiberline (Denmark) [2] could be convinced to sponsor the structural materials for the construction as well as
specialized work such as welding of stainless steel. The rest of the works shall be executed by indigenous companies in
order to locally anchor the works.
To choose stainless steel as the main structural material has quite different motivation: for the Zumaia footbridge the
vicinity to the sea is the main reason, as already applied for a road bridge in Spain [3]; for the Juan Bobo footbridge in
Medelln, however, it is the excellent durability intending a practically maintenance-free service life, also see [3].

2.

Description of the footbridges

2.1

Zumaia footbridge, Spain

The new bridge has a span of 28.0 m (Fig. 1) and an open section. The section consists of two simply supported
Vierendeel trusses of variable height of 1.071.37 m and a transverse deck (Fig. 2). The truss is made of stainless
duplex steel of class 1.4462; the openings of the Vierendeel trusses are filled with structural sheets of glass fiber
reinforced polymer (GFRP) as structural webs.

Fig. 1 Longitudinal section of Zumaia footbridge.

The box cross-section of the bottom chord has a variable height of 300 mm to 400 mm, and constant width and
thicknesses of 360 mm and 12 mm, respectively. The box is shaped skew to the inward of the bridge. The posts have a
Pi-section, which is open to the deck, and have a slightly variable spacing of 2.20 m to 2.50 m (narrower towards bridge
end). Between the webs of the posts, steel plates 15 mm x 200 mm, the lighting for the bridge deck will be installed. The
flange of the posts is a steel plate 15 mm x 300 mm. The top chord is a steel plate 28 mm x 300 mm with a web 15 mm x
80 mm. The elliptically shaped handrail is made of durable wood (missanda / ordeal tree) and is 0.951.05 m above the
deck. The handrails rest on twin posts of stainless duplex steel, which are welded to the top chord of the truss, but are
not considered as structural members.
The GFRP sheets in the openings of the Vierendeel-truss consist of two plates with 8 mm thickness each. They are
connected to the posts and the top and bottom chords by stainless steel bolts or structural blind rivets, e.g. POP rivets.
The total width of the bridge is approx. 5.60 m, and the deck provides a walkway width of 5.0 m. The deck is built from
GFRP transverse girders, type I 300x150x15 with 1.0 m spacing, and standard GFRP walkway planks (type Fiberline
HD40 or similar). The deck has a total height of approx. 340 mm and a transverse slope of 1% and a longitudinal slope
of 0.75%; the whole structure is counter-sagged by 300 mm. The transverse girders at the ends of the bridge and two

Fig. 2 Cross-section of Zumaia footbridge.

intermediate girders are made of stainless duplex steel; for the end transverse beams, box sections are chosen while the
intermediate stainless steel transverse girders have I-sections. The stainless steel transverse girders are welded to the
bottom chords, in order to provide the stability of the top chords of the open section by frame action. The GFRP
transverse girders are connected to the bottom chords with three M12 bolts, steel grade A4, structurally considered as
pinned connections. The GFRP walkway planks are treated with a slip-resistant surface (sand-epoxy coating) and
mounted onto the transverse girders with standard connectors.
The abutments of the bridge are made of a simple reinforced concrete beam each. On the left riverbank, the new bridge
can be supported by the new sidewall of the channel. On the right riverbank, the foundation beam will be supported by
two groups of three repeatedly injected micro-piles 200 mm (nominal pile diameter), reinforced with a steel tube
139 mm x 12 mm. The micro-piles are drilled through the existing sidewall and decoupled from it by a PE tube
200 mm. The piles are inclined 1:5 away from each other. The piles closer to the river are loaded in compression, and
therefore, embedded in the rock below the riverbed with a length of 1.2 m (6) and have a total length of approx.
16.6 m. The piles away from the riverbed are loaded in tension. This force is anchored on a length of 5.0 m below the
base of the retaining wall, leading to a total length of the piles of approx. 11.0 m.
2.2

Juan Bobo footbridge, Colombia

The structure of the Juan Bobo footbridge is very similar to the Zumaia footbridge; Fig. 3 shows a visualization of the
newly developed bridge type combining stainless steel and GFRP sheets. The main difference is the reduced span of
25.2 m and the reduced deck width of approx. 3.60 m; the deck provides a walkway width of 3.0 m. The Vierendeel
trusses have the same height, since it is mainly determined by the vertical position of the handrail. The stainless duplex
steel provided for the structural steel parts is also chosen differently as class 1.4162 (LDX 2101).
Further changes to the structure are that all transverse girders are in stainless steel, and that the posts have a constant
spacing of 2.29 m. Due to the smaller span, some of the structural steel plates could also be reduced in thickness: the
plates of the bottom chords are 10 mm thick, and there is only one GFRP plate of 8 mm thickness in the openings of the
Vierendeel trusses. They are connected to the posts and the top and bottom chords by structural blind rivets.
The transverse beams are welded stainless steel girders with I-section (flanges 5 mm x 80 mm, web 4 mm x 155 mm) at
a spacing of 1.5 m. The whole structure is counter-sagged by 250 mm. The transverse girders at the ends of the bridge
and two intermediate girders are made of stainless steel box sections 300 mm x 300 mm x 10 mm. These box girders
are welded to the bottom chords, in order to provide the necessary frame action. The transverse I-girders are connected
to the bottom chords with two M12 bolts, steel grade A4, structurally considered as pinned connections. The abutments
of the bridge are made of simple reinforced concrete footings supported on drilled concrete piles.
Although the structure may look rather simple, the geometry of the structural steel parts is rather complex since all
members are inclined and have variable height; this leads as an example to differences in the horizontal positions of
the posts on the bottom chord if the top chord shall be straight and the openings shall be plane for the GFRP panels.
The top and bottom flange of the bottom chords have to be cut slightly curved, if the webs shall be plane.

Fig. 3 Visualization of bridge type combining stainless steel frames and GFRP sheets as structural webs.

3.

Design procedure and structural modeling

3.1

Procedure for ULS and SLS design

The main problem in the proof of structural safety of this new combination of structural materials is the lack of ductile
behavior of the GFRP sheets in the ultimate limit state in comparison to the highly ductile stainless duplex steel. The
literature only provides design formulas for local shear buckling of GFRP sheets in the elastic state, e.g. [4]. However, it
is believed that the sheets could provide a post-critical resistance analogously to the webs of thin-plated steel girders,
provided that sufficient stiffening elements are available but no scientific information is accessible on this matter today.
This may be mainly due to the fact that exhaustive research on FRP structural materials is primarily conducted in the
aircraft industry and therefore usually subjected to confidentiality. The right to use to such information, however, might
provide considerable input to the development of new types of FRP structures in bridge and structural engineering.
For the ULS design of the structure, it is thus assumed that only the Vierendeel trusses in stainless steel are effective for
the longitudinal girders. The GFRP sheets are considered on the safe side to be totally ineffective at ULS. The GFRP
structural members of the deck deck planks and transverse girders though are considered to be effective in the ULS
for the load transfer from the deck to the longitudinal main girders.
For SLS checks, the GFRP sheets as structural webs of the Vierendeel trusses are considered to be fully effective since
experience shows that deflections or other serviceability limit states such as vibrational behavior usually governs the
design of footbridges and GFRP structures.
These reflections on ULS and SLS design lead to the following design procedure:
ULS and SLS design of deck structure deck planks and transverse girders independently from main girders
Determination of minimum steel sections for main girders from ULS proof with 2D frame model
SLS analysis for frequent load cases for the minimum thicknesses of GFRP sheets with 2D frame model
Dynamic analysis on global structural model, including deck structure, according to state-of-the-art literature
Verification of ULS and SLS of final structure
Design of connections of GFRP sheets to stainless steel frame for rare load combinations
ULS and SLS loads are adapted from the Spanish code [5] and Eurocode 1 [6]; specific dynamic load cases are
considered according to [7].
3.2

Structural modeling

The main girders are modeled as plane Vierendeel frames; to account for the effective height of the panels, the steel
members are introduced with offsets. However, no eccentricities are introduced for the connections of the panels to the
posts. This modeling is regarded to be on the safe side since the length of the panels is over-estimated, hence too soft.
The deck is modeled with rigid connections for the four stainless steel transverse girders, pinned connections for the
GFRP transverse girders and representation of the planks by a beam grid. The effective layout of the deck planks
regarding the plank joints due to limited transportation length of the planks is considered by introducing bending
hinges in the beam grid. This way, the deck tends to be modeled too soft for its horizontal and torsion stiffness, which is
on the safe side regarding the SLS and the dynamic analysis.
The supports are set up at the edges of the bottom chords of the longitudinal girders and first posts such, that no
restraints are introduced on the structure (single span beams longitudinally and transversally). The global model of the
Juan Bobo footbridge is shown in Fig. 4, also illustrating the hinges introduced to consider the joints of the deck planks.
All calculations are performed with the FE-program ROBOT Millennium, version 19.0.7 [8]. This program allows the
application and combination of relevant features for the projects presented here, such as: bar members with tapered
cross-sections; user-defined (i.e. thin-walled) cross-sections; offsets (eccentric joints) of bar members; orthotropic FEpanels with user-defined stiffness matrix of variable shape; and most importantly for the dynamic behavior verification
modal and time-history analysis of 3D structures. The program also offers common applications of calculation of stress
and deformations, sectional forces etc.

Fig. 4 Representation of global 3D-frame model of Juan Bobo footbridge.

3.3

Analysis and assessment of dynamic behavior

The dynamic behavior of the footbridges is analyzed according to the procedure suggested in [7], considered state-ofthe-art for dynamic assessment of pedestrian-induced vibratory behavior of footbridges. The publication was initiated
after the experiences of the Millennium Bridge in London and the Solferino Bridge in Paris. It considers the main
parameters for the dynamic behavior of structures mass, stiffness and damping as well as the common assessment
criterion for pedestrian bridges, i.e. risk of resonance and accelerations. The main features of this procedure are:
Assessment of the risk of resonance for vertical and horizontal vibrations (longitudinal and transversal) by classes
of natural frequency ranges, for two load cases: bridge with dead loads only or bridge deck additionally loaded
with 1 pedestrian/m2 (additional mass 70 kg/m2), corresponding to a very dense (walking) crowd.
Definition of comfort classes in terms of accelerations and their effect on pedestrians as assessment criterion
for the dynamic behavior of a footbridge.
Definition of dynamic load cases for different bridge classes, considering pedestrian traffic volume (crowd
densities) by a number of equivalent pedestrians and allowing for reduction factors outside the range of maximum
risk of natural frequencies between 1.7 Hz and 2.1 Hz, as well as the structural damping ratio .
Comfort of pedestrians on lively structures is usually judged by accelerations or displacement velocities, although being
a highly subjective issue depending on the pedestrians age and fitness, his position (stationary or moving), period of
exposure to vibrations and the level of threshold considered (perception, disturbance, hazardous), e.g. [9]. The peak
values of acceleration a, displacement velocity v and displacement amplitude are linked to each other for low damping
by the corresponding natural frequency f by
v=

a
2 f

and

(2 f )

v
2 f

(1)

A vertical displacement velocity v = 24 mm/s is said to be perceptible by pedestrians; adjustment of the pedestrians gait,
however, already occurs for vertical displacements = 1020 mm [10]. This corresponds to at least a = 1.6 m/s2 or v =
125 mm at f = 2 Hz. It should be noted that the fulfillment of rather restrictive displacement criteria, e.g. L/700, does not
necessarily provide a sufficient dynamic behavior of the structure since dynamic displacements may lead very soon to
adjustments of pedestrian gait and lock-in effects; pedestrians are also much more sensitive to lateral displacements
than vertical displacements.
The procedure applied here [7] considers the following comfort levels for vertical accelerations:
a < 0.5 m/s2: practically imperceptible
0.5 m/s2 < a < 1.0 m/s2: merely perceptible
1.0 m/s2 < a < 2.5 m/s2: felt but not unbearable
a > 2.5 m/s2: not acceptable

In order to avoid forced synchronization in lateral horizontal vibrations (lock-in effects), the lateral accelerations should
be limited to a < 0.1 m/s2 [7]. The accelerations depend inversely on the assumed damping ratio , normally a very small
value below 5%; the damping should thus be chosen conservatively and its influence be judged by sensitivity analysis.
Generally, accelerations only have to be assessed if there is a risk of resonance; for vertical vibrations and horizontal
vibrations, this risk is negligible if f > 5 Hz or f > 2.5 Hz, respectively.
4.

Main results from dynamic and static analysis

4.1

Results for Zumaia footbridge

4.1.1 Dynamic analysis

The structural layout is altered in several steps in order to provide sufficient dynamic behavior; the main changes
concern the increase of the overall static height of the main girders, an increase in plate thickness of the bottom chord
and un-staggered thicknesses of the GFRP panels.
Another point of concern regarding structural modeling is the influence connections of the GFRP transverse girders on
the dynamic behavior; as possible models pinned joints, eccentrically pinned joints, elastically clamped and totally rigid
joints are studied. The results from modal analysis show that the type of joints has a minor influence on the natural
frequencies in the critical range below 5 Hz. The further investigations are thus performed with clamped joints for the
transverse girders (regarding their strong axis).
Fig. 5 shows the first three modal shapes, and Table 1 summarizes the results of the dynamic analysis; the scatter in the
accelerations in Table 1 is due to the assumed range of the structural damping ratio , in terms of a sensitivity analysis.

Fig. 5 Modal shapes of Zumaia footbridge: vertical bending, mainly torsion and lateral bending with torsion (left to right).

For a crowd density of 1.0 pedestrians/m2 being very dense, according to [11] the accelerations of the first bending
mode can be judged as felt but not unbearable, Table 1 and section 3.3; for a crowd density of 0.8 pedestrians/m2 still
being a dense crowd, according to [11] the accelerations will be practically imperceptible. The significant decrease in
accelerations for the latter case is mainly due to the reduced risk of resonance for natural frequencies above 2.1 Hz and
the corresponding reduction factors, section 3.3 and [7]. The corresponding displacement velocities amount to v =
4974 mm/s and v = 2521 mm/s, respectively, Eq. (1).
The second and third mode have probably significantly higher natural frequencies, since the deck planks are modeled in
a simplified way as a beam grillage, section 3.2; since the stiffened GFRP planks are usually also coupled in the
transverse direction, they will provide a significantly increased lateral bending and torsion stiffness. If these modes are
considered as lateral vibration modes, the risk of resonance could even be judged negligible.
Apparently, the accelerations strongly depend on the assumed damping ratio ; this ratio shall therefore also be
determined in a dynamic load test.
Table 1

Results of dynamic analysis of Zumaia footbridge, for structural damping ratio of = 23%.

Natural frequency [Hz]

1.0 ped./m2
0.8 ped./m2
2.51
2.56
3.11
3.15
3.57
3.63

Modal shape

Risk of resonance

Vertical bending
Torsion with lateral bending
Lateral bending with torsion

Medium
Low
Negligible

amax [m/s2]
1.0 ped./m2
0.8 ped./m2
1.160.77
0.410.33
1.020.68
0.870.71
---

4.1.2 SLS design

Deflections

The overall serviceability behavior regarding deformations is expected to be excellent. The GFRP panels are notably
effective for dead loads and should therefore be mounted before placing the bridge in its final position. The maximum
deflection for frequent load combinations amounts to approx. L/1000. This figure is higher than the recommended value
of [12] of L/1200 with no need to perform a dynamic analysis; it can thus be concluded that deflection limitations have to
be very restrictive if considered as the only criterion for sufficient dynamic behavior. The GFRP deck plank profiles are
also chosen heavier than from regular deflection limits, since users reported earlier that the deck planks might feel soft.

Buckling of GFRP sheets

The GFRP sheets are designed such that there is no buckling for rare load cases. The critical horizontal normal buckling
stress of the two GFRP sheets t = 8 mm per main girder amounts to cr,web = 36.5 MPa in total, while the average critical
buckling shear stress can be calculated as cr,web = 9.9 MPa (depending strongly on the effective panel height) in total.
The governing normal and shear stresses do not occur in the same panels, see Fig. 6; assuming coinciding action
nevertheless, a safety reserve of already 25% can be calculated if compared to a combined verification criterion
analogously to steel structures [12] regarding buckling safety in the SLS.

Fig. 6 Stresses in GFRP panels for rare load combinations horizontal normal stresses (left) and shear stresses (right).

Distributed vertical stresses on the webs occur due to the bending curvature, which may lead to buckling of the GFRP
webs. This vertical compression stresses are over-compensated by the curvature of the top flanges, Fig. 1. In any case,
the resulting vertical compression stresses are very low (0.06 MPa) and clearly below the minimum critical vertical
normal stress of 2.8 MPa, according to [4]. Analogous conclusions can be drawn for local vertical normal stresses in the
GFRP sheets due to local bending of the adjacent steel members. The critical normal stresses are at least twice as high
as the maximum local vertical stress for rare load combinations.
Approximately 90% of the total shear force is carried by the bottom chord of the Vierendeel frames and not as would be
expected by the GFRP webs. However, it seems to be a reasonable assumption for preliminary designs to assign all
the shear forces to the GFRP panels and choose the sheet thickness accordingly in order to allow the verification of the
stability of the sheets. Evidently, the final stresses in the panels depend on the stiffness proportions of the system and
have to be checked for the specific structure.

Connections of the GFRP panels to the steel structure

The governing shear forces for the final design of the connections of the GFRP panels to the steel frames are
determined from the differences in normal forces in the adjacent chords. The connections are designed with a reserve of
10% for rare load combinations to prevent local failures and assuming equal participation of all rivets along the panel
edges. The construction detail is such that two cuts per rivet are statically effective. The design leads to different spacing
of the blind rivets 4.8 in stainless steel, with a minimum distance of 30 mm in the outermost panel and maximum of
100 mm at mid-span, chosen from engineering judgment. It is proposed to vary the spacing of the rivets from panel to
panel in order to visualize the flow of forces in the panels and the main girders, respectively.
Further checks concern the verification of local plastification of steel members in the SLS; according to [12], a reserve to
the characteristic yield strength fyk of 10% should be respected for rare load combinations, while this reserve shall be
25% for frequent load combinations. The live loads are clearly dominant since qk / gk = 4.0 / 1.45 = 2.77, allowing to
conclude that the frequent load combination is not governing since 1 = 0.50 < 0.75 / 0.90 = 0.833. The reserve
calculated for the rare load combination amounts to at least 20%.

4.1.3 ULS design

GFRP panels and their connections

The sheets are assumed to be not effective in the ULS, section 3.1. Since the panels are neglected, there is no need to
design the connections for the ULS. However, the calculations show that apart from the first one or two panels from the
bridge end, the panels may be even effective for the ULS, not yet considering a probable post-critical resistance as for
steel webs, section 3.1. The weakest links are then the connections of the panels to the steel structure; this is
appreciated since the failure of the rivets will be governing (not the failure of the GFRP panel itself), which is regarded to
lead to a less pronounced brittle failure. This might not be the case for connections with more compact stainless steel
bolts, probably exhibiting a more brittle failure. This is the main reason that it is suggested to apply stainless steel blind
rivets instead of bolts for the connections of the sheets to the steel frame.

Steel members

The design of the structure for acceptable dynamic behavior is governing for the dimensions of the steel members. The
minimum sections of the main girders were designed in detail in an earlier design phase, section 3.1. The initially chosen
dimensions of the steel members show a minimum reserve of 15% in the ULS; with the structural changes due the
modifications for dynamic behavior, lower sectional forces result in general.
Vertical buckling analysis of the top chords leads to the application of a web 15 mm x 80 mm to the flange 28 mm x
300 mm, section 2.1. Horizontal buckling of the top chord of the open bridge section is checked considering the frame
action of the open section, including the torsion stiffness of the bottom chord (Fig. 7 left); the analysis is performed with
Engessers analogy of an elastically bedded beam (Fig. 7 right), with the spring constants determined as shown in Fig. 7
(left) by the Cross method, leading to a buckling length of approx. 3.2 m or 1.4x post spacing s, Fig. 1.
.023
.138
.023

1.0
.141
.141
.141

1.0
1.0

.023

.115

N<0

.138
s

.721

Fig. 7 Frame action in open bridge section considering torsion stiffness of bottom chord (left), and Essengers analogy (right).

Deck structure

The dimensions of the deck planks are governed by SLS considerations, section 4.1.2. For the ULS, however, the
accidental presence of a vehicle on the bridge deck is considered, even though not prescribed by the ownership. It is
decided to determine the allowable mass of a vehicle of compact dimensions from the provided resistance of the deck
planks; the geometry of the load model is chosen analogously to the service vehicle model of [6].
The bending capacity of the deck planks is governing for this accidental design situation; the shear capacity of the deck
planks is clearly higher. The vehicle may have a maximum design wheel load of Qd 14.5 kN on a footprint of 0.2 m x
0.2 m. This corresponds to a vehicle mass of at least 4.4 t, if it is assumed that 2/3 of the total mass acts on one axle. If
the vehicle is more balanced, the bearable mass can be increased up to 5.9 t. These vehicle mass is judged
reasonable for the accidental crossing of a larger vehicle, e.g. a small truck or a heavy service vehicle of compact
dimensions. However, the crossing of a vehicle shall be prevented by the means of at least two posts at every end of the
bridge, limiting the local access width to approx. 1.6 m.
4.2

Results for Juan Bobo footbridge

The results presented here focus on the dynamic analysis; the rest of ULS and SLS verifications is similar to the Zumaia
footbridge, section 4.1, and will not give any additional information. Table 2 summarizes the results of the dynamic
analysis of the Juan Bobo footbridge, and Fig. 8 shows the modal shapes. Note that the scatter in accelerations in Table
2 results from a wider range of assumed structural damping ratios .

Table 2

Results of dynamic analysis of Juan Bobo footbridge, for structural damping ratio of = 0.43%.

Natural frequency [Hz]

1.0 ped./m2
0.8 ped./m2
2.44
2.47
3.22
3.28
4.69
4.72

Modal shape

Risk of resonance

amax[m/s2]

vmax [mm/s]

Lateral bending
Vertical bending
Torsion

Low
Low
Low

0.020.008
1.590.61
0.320.12

1.40.5
77.529.4
10.74.0

Fig. 8 Modal shapes of Juan Bobo footbridge: lateral horizontal bending, vertical bending and torsion (left to right).

The modal shape with the lowest natural frequency is a lateral bending mode, as can be expected from the reduced
deck width. Despite the low risk of resonance and the soft modeling of the deck as beam grillage, it is decided to
investigate the dynamic behavior for the 2nd and 3rd modal shape in more detail by the means of a time-history analysis.
The first modal shape is not more deeply analyzed since the resulting accelerations are considerably below the
acceleration limit for forces synchronization, section 3.3. Fig. 9 shows results for the 2nd natural mode.

Fig. 9 Time-history analysis for 2nd mode: acceleration in [mm/s2] (top) and displacement velocity in [mm/s] (bottom) at mid-span
for = 2%.

The results from time-history analysis amount to only 63% of the accelerations determined from simple hand calculations
according to [7]. This is mainly attributed to the fact that manual calculations do not consider that stiffness and masses
are concentrated in the main girders and not equally distributed over the whole deck width. The natural frequency of the
deck alone is estimated to fdeck 16.1 Hz determined from f = /(2L)2(EI/m) for a single span beam , i.e. the deck
structure is not excited at its resonance frequency, leading to a considerably lower dynamic amplification since.

5.

Conclusions

The two footbridges presented here combine several new features: stainless steel and glass-fiber reinforced polymers
(GFRP) as structural materials as well as blind rivets for the connections of the GFRP panels to the steel structure. The
panels are applied as structural webs of stainless steel Vierendeel trusses.
As with many pedestrian structures, the design of the footbridges is dominated by the dynamic behavior i.e. sufficient
stiffness to limit accelerations being assessed by state-of-the-art literature and refined time-history analysis. This type
of analysis is clearly recommended since it implicitly considers local distributions of stiffness and dynamic mass; for the
presented footbridge the maximum accelerations could be reduced by almost 40%, compared to a simplified analysis.
The static analysis shows that the GFRP panels are less effective in the SLS than expected; unfortunately, no scientific
information is available on a post-critical resistance of GFRP panels in the ULS, analogous to webs of stiffened thinplated steel girders even if such a structural resistance can be easily imagined. However, the panels provide a shear
stiffness increase with negligible weight increase and, above all, high structural damping. This way, GFRP panels can
provide a significant enhancement to SLS behavior of footbridges, although they are not considered for the ULS
verifications. The contribution of the panels to the improvement of the dynamic behavior is due to the good damping
properties of the material itself as well as the high number of connections with stainless steel blind rivets at narrow
spacing instead of the usual connections with stainless steel bolts at larger spacing.
The horizontal buckling resistance of top chords of open bridge sections can be improved by considering not only the
frame action of posts and transverse girders but also the contribution of the torsion stiffness of the bottom chords.
6.

Acknowledgements

The sponsoring of the structural materials for the Juan Bobo footbridge in Medelln (Colombia) by Outokumpu (Finland)
and Fiberline (Denmark) is highly appreciated. The main author would like to sincerely thank Juan Sobrino for the
wonderful working experience he had during his stay at Pedelta s.l., Barcelona, in the period of September 2006 to
March 2007, and that allowed him to work abroad on very interesting and challenging projects in bridge engineering.
7.

References

[1]

http://www.outokumpu.com (visited on Jan. 4, 2008).

[2]

http://www.fiberline.com (visited on Jan. 4, 2008).

[3]

SOBRINO, J. A., Stainless Steel Road Bridge in Menorca, Spain, Structural Engineering International, Vol. 16,
Nr. 2, 2006 , pp. 96-100.

[4]

BANK, L.C., Composites for Construction Structural Design with FRP Materials, John Wiley & Sons Inc.,
Hobokan NJ, 2006, 551 pp.

[5]

Ministerio de Fomento, Instruccin relativa a las acciones a considerar en el proyecto de puentes de carretera
(Instructions on the actions to be considered in the design of road bridges), 1998.

[6]

Comit Europen pour la normalisation (CEN). Eurocode 1 actions on structures, 2003.

[7]

Association Franaise de Gnie Civil (AFGC), Assessment of Pedestrian-Induced Vibratory Behaviour of

Footbridges / Evaluation du Comportement Vibratoire des Passerelles Pitonnes sous l'Action des Pitons
(bilingual), 2006.

[8]

http://www.robobat.com (visited on Jan. 4, 2008).

[9]

BACHMANN H. et al., Vibration Problems in Structures Practical Guidelines, 2nd edition, Birkhuser, Basel /
Berlin / Boston, 1997.

[10]

STRASKY, J., Stress Ribbon and Cable-Supported Pedestrian Bridges, Thomas Telford, London, 2005, 232 pp.

[11]

International Federation for Structural Concrete / Fdration Internationale du Bton (fib), Guidelines for the
Design of Footbridges, Bulletin Nr. 32, 2005.

[12]

Ministerio de Fomento, Recomendaciones para el proyecto de puentes metlicos para carreteras

(Recommendations on the design of steel road bridges), 1996.