Footbridge over the Bow River in Banff

Gerald Epp Gerald Epp, born 1956, received his

President, StructureCraft Bachelor’s and Master’s degree in Structural
Engineering from the Univ. of British
Partner, Fast + Epp Columbia. He co-founded Fast + Epp in
Vancouver, BC, Canada 1989, and established sister design-build
[email protected] firm StructureCraft Builders in 1998.
Gerald sits on the committee for the CSA
086 Timber Design Code
.

Adam Gerber Adam Gerber, born 1988 is a recent

Project Engineer graduate from the Univ. of British Columbia
department of Civil Engineering and
StructureCraft Builders currently working towards his M.Eng. He
Delta, BC, Canada has worked with StructureCraft Builders for
[email protected] 8 years, starting as a carpenter before
pursuing a career in Structural Engineering.

Lucas Epp Lucas Epp, born 1985, received his civil

Project Engineer engineering degree from the Univ. of British
Columbia. He worked for Fast + Epp and
StructureCraft Builders StructureCraft Builders prior to joining a
Delta, BC, Canada prominent engineering consultancy in
[email protected] Auckland.

Summary
Set over the Bow River in the mountain setting of Banff is a slender 113m long timber bridge
both serving pedestrian traffic and providing a sensitive sanitary crossing replacement. Its
minimal form was designed to blend into the natural beauty of Canada’s first national park.
Completed in June 2013, the bridge was designed and constructed by StructureCraft Builders in
collaboration with Fast + Epp Structural Engineers for the Town of Banff, under a design-build
contract. Key to success of the project was careful attention to detailing, mitigation of dynamic
behavior, and a high degree of pre-fabrication.

Figure 1 - Bridge superstructure after erection

Keywords: timber footbridge, long span, dynamics, tuned mass damper, design-build, Banff
1. Introduction
Given the highly visible and historically significant location, the client was very careful about
the form the footbridge would take, the views it would create both from its deck and from
surrounding vistas (of which it would form a part), the materials it would be made of, and the
connection it would make with local trails. Driving the agenda, however, was a more pressing
concern that existing sanitary pipes installed below the river at this location some fifty years
earlier could fail, spilling raw sewage into the pristine Bow River. So the bridge also needed to
carry new pipes, creating an important dual use for the new crossing. The bridge is designed to
carry medium weight emergency vehicles, addressing another need in the community, as there is
only one other crossing to the populated south side of the river. Being situated in a national park,
environmental constraints were particularly stringent for both the final structure and the
construction process.
In late 2011, the Town created a design-build competition attracting a number of respondents;
choosing a winner who addressed the constraints, with a cost which fit within the budget. The
final design features an 80m clear span, which for a timber bridge is one of the longest of its
kind. This with a relatively slender 4m width create the primary design challenge for the bridge:
its dynamic behaviour. Not uncommon with long span footbridges, the primary vertical and
lateral frequencies fall directly in the range of susceptibility to pedestrian excitation. To
sufficiently damp the response, several damping systems were investigated, with the final system
comprising two custom-designed tuned mass dampers suspended visually with cables beneath
the bridge. Monitoring of the structure in-place has allowed tuning of the masses to the measured
natural frequencies of the bridge.
2. Design
The design-build delivery method enabled both the designers and the Town to pursue a variety of
concepts based on a simple set of guidelines:
1. Design within a 135m crossing length, with maximum 5% slope for accessibility.
2. Carry two sanitary and one water line, critical for the Town of Banff.
3. Create a minimal, unimposing design.
4. Minimize work within the river channel.
5. Accommodation of medium sized emergency vehicles.
The structural system is simple: Propped by drilled piers located (to minimize disturbance) just
outside the normal river channel, two 40m tapered haunch glulam girders cantilever from either
side to support a central 34m suspended span. Tension rods tie the cantilevers down into concrete
abutments at either end of the bridge. The bridge cross section comprises twinned sets of stepped
glulam girders trussed with diaphragm steel webbing. These are designed to follow the flow of
forces, and range in depth from 2.6m at the piers to 0.9m at the suspended span. The 4m wide
path is made of solid timber panels which span between the tops of the girders, and are
removable to provide access to the service pipes hidden below.
Figure 2 - Site plan

Figure 3 - Bridge elevation

The concrete abutments at either end of the crossing build up the

pathway to the bridge level, but also house the machine room for
the mechanical systems required to run the lift station,
eliminating the need for any additional above-grade structures.
Thanks to careful positioning of the bridge, existing trails on
both sides of the river were maintained without any substantial
grade or alignment changes.
Piling in a national park and close to a township was a delicate
process, requiring careful selection of foundations. Noise and
soil constraints ruled out driven piles, and environmental
restrictions within the river basin required the installation of a
containment system around the entire piling area. The two sets
of piles in the river take large compression loads, as the pivot of
the cantilever, while the two front piles at each abutment hold
down the cantilever’s back-span. Each of the river piles has a
steel pipe anchored to augured concrete piles and fastened to the
precast concrete pile caps above.
The glue-laminated beams contribute to the bridge’s character
and natural feel, however wood beams of this scale are not
without their challenges. At up to 2.6m deep, 43m long and
11.5T per piece, simply finding manufacturers capable of
producing beams this big was a challenge. The girder supplier
needed to perform a full survey of their own facility to ensure
Figure 4 - Section at north pier
 that once loaded onto a tractor and steerable rear axle
rig, the beams with their stepped profile maintained
adequate clearance leaving their yard.
Durability was a topic of significant interest to the
client. The Town wanted to ensure that the bridge
continued to look and perform well for the extent of its
design life. A tremendous amount of energy was spent
detailing for durability. The glulam pairs were
separated by laminated veneer lumber (LVL) blocking
and capped with a heavy gauge galvanized steel
Figure 5 - Pile installation through the ice flashing. The spacing allows full ventilation between
members and the flashing creates a drip edge shedding
water 100mm from the face of the glulams. All steel
components are hot dip galvanized (HDG) or stainless
steel, and rubber spacers or grommets separate the two
where they interface. The glulams were coated with a
high quality system which behaves like a membrane,
allowing vapour to pass through while preventing liquid
water from penetrating the wood. A darker tint was
selected to provide improved UV protection and all three
coats were applied in shop prior to shipping. The town
has committed to maintenance of the coatings with the
goal of increased longevity and performance of the
structure.
The steel diaphragm trussing serves a dual purpose, both
connecting the pairs of glulam beams and creating
support for the three service pipes sitting just underneath
Figure 6 – Concealment of service pipes
the bridge deck. Rollers are attached to the steel webs,
allowing the pipes to be easily deployed across the
bridge, and accommodating thermal movement of the
pipes during the year.
The central drop span sits on neoprene bearing pads on
notches in the receiving ends of the cantilevered Glulam
beams. This detail is achieved by using long screws to
reinforce notch. Diaphragm chord forces are transferred
through the connection into the beams via steel drag
straps along the tops of the beams, concealed beneath the
deck panels and flashing. This forms an elegant,
concealed connection which was easily assembled on site
and left plenty of tolerance during erection.
Figure 7 - Drop-in span detail
3. Structural Analysis
The structural analysis and design of the bridge and associated abutments was in some ways
relatively simple – as a determinate system, the basic forces in the cantilevers and drop span due
to gravity loading can be calculated quickly by hand. To capture the more complex behaviour
due to stepping of glulam, a finite element model was built incorporating the beams and pile caps
as shell elements, and the trussing and piles as beam elements.
The long span and slender section of the bridge make it susceptible to both vertical and lateral
excitation from human traffic on the bridge. The use of timber, which is less stiff than steel, also
contributed to creating a more flexible structure. Early discussions with the client highlighted
the fact that this was likely to be a lively bridge, and mitigation measures were proposed.
The 3D finite element model was also used to investigate the response of the bridge to dynamic
excitation. Preliminary modal analyses showed that the fundamental frequencies of the bridge
both vertically and laterally fell within the critical range for human excitation. The primary
modes of concern and their natural frequencies were as follows:
Table 1 - Critical modes and frequencies

Mode Description Predicted Measured Mechanism

Frequency (Hz) Frequency (Hz)
First Lateral 0.8 0.8 Walking
Second Vertical 1.9 1.5 Walking
Second Torsional 3.1 3.3 Jogging

It seemed evident that some form of damping system would be advantageous for user comfort.
Detailed dynamic loading models for each of the potential excitation scenarios were imposed on
the finite element model, and analysis of the time history acceleration response provided the first
estimates of performance under pedestrian excitation and the level of damping required.
Several damping systems were investigated, including
viscous dampers fastened to the bridge diaphragm, and
combined viscous and tuned mass dampers suspended
below the bridge deck. The final system adopted
comprised two asymmetric custom-designed tuned
mass dampers suspended on cables at quarter points
beneath the central span of the bridge. The dampers
were created using a cradle which had the capacity to
hold different numbers of tuning mass plates. This
cradle was suspended from the bridge above with two
sets of cables, of particular lengths and stiffnesses.
Figure 8 - Tuning of mass dampers before erection
4. Fabrication
A parametric 3D solids model of the entire bridge was created early on in the design process,
allowing rapid and detailed investigation of a multitude of design decisions, providing visual
feedback to both
designer and client. The
model grew through the
design process to include
every component of the
bridge – from the
surrounding topography
to the giant glulam
beams to the minimal
guardrails. The
Figure 9- 3D parametric model parametric nature of the
model allowed it to respond seamlessly to changes in global form and reflect them down to the
smallest level of detail.
The model was critical to many parts of the design process, from early visualization and
rendering, to cut-and-fill calculations on the site, trucking viability, and fine tuning the aesthetics
of the tuned mass damper beneath the bridge. Most importantly the model was used to create
detailed fabrication drawings for each piece of the bridge. The link between model and drawings
ensured drawings remained accurate even as the design evolved.
As the Glulam beams are curved, the parametric model contained both ‘states’ of the beams –
pre-bent (straight, prior to cutting) and post-bending (arched and cut to length), allowing
accurate dimensional checks to be performed at each of the manufacturing stages and an accurate
appraisal of erection weights.
Comprehensive part numbering and component scheduling within the model provided accurate
material take-offs and ensured all pieces were assembled correctly on the site.
Fabrication of all timber elements occurred in the shop. The direct proximity of the fabrication
shop and the design team allowed full-scale ‘rapid prototyping’ of many of the key components
of the bridge. Handrails, stanchions, and deck panels all went through several mock-ups and
design iterations before settling on a final design optimised for both form and aesthetic.
Throughout the design and fabrication process, the delivery model led to a natural desire to
create efficiency: similar to the manufacturing industry, conceptualisation, design, modelling,
testing, fabrication, and erection takes place in-house, allowing each step of the process to be
influenced by the other steps.

Figure 10 - Glulam delivery at fabrication shop

5. Erection
A tight, remote site, huge structural elements, and harsh winter weather all combined to make the
bridge erection a challenge. This, coupled with a desire to complete the lifts before spring thaw
and a firm arrival date for a large mobile crane put a huge emphasis on ease and accuracy of
assembly in the field.
The main structural elements of the bridge were too large to be transported to the site in modules
ready for installation; and fitting up the pieces over the river with a smaller crane would have
presented significant environmental and safety challenges. In order to accurately assemble and
erect the bridge, the individual elements were prefabricated in the shop and shipped to site as a
kit of parts. All cutting, drilling, sanding and finishing was performed indoors under controlled
conditions so that members were protected from the elements both in transit and on site. Jigs
were built to ensure accurate assembly of the main bridge components in the field. The stepped
glulam girders were paired up and joined by long fully threaded screws through pre-drilled holes.
Once the beams were levelled and squared to each other, the diaphragm webbing was connected
with friction grip bolts to the receiving plates which were fastened to the innermost glulam face.
During the final preparations for the big lift,
temporary walking planks and safety lines were
installed to provide immediate access to the bridge
sections after installation. Additionally, several
survey targets were installed to the girders, both at
the abutment end and at the tips which would
receive the drop-in span. This allowed real-time
verification and placement of the bridge sections and
fine tuning of the hold-down rods at the abutment
ends to compensate for dead load deflection of the
Figure 11 - Erection of south girders girders and the drop-in span.
In all, the entire bridge superstructure was erected
in 3 lifts over 2 days with an additional day spent
relocating the crane to the North bank of the river.
The tapered girder assemblies weighed in at over
50T while the drop-in span, complete with tuned
mass dampers, maxed out the 500T crane at just over
20T and a radius of 45m. The bridge sections rested
within 12mm of their target coordinates and the
drop-in span “fit like a glove,” according to the
erector.

Figure 12 - Erection of drop-in span

6. Dynamic Testing
The desire for field testing of the bridge arose early in the design
phase and gained traction as the importance of its dynamic behaviour
came into focus. The testing was primarily required to provide
experimental validation of the structure’s natural frequencies; but a
number of additional tests were performed which provided
estimations of modal damping as well as measurements of pedestrian
induced accelerations for comparison with user comfort criteria.
This information allowed for precision tuning of the dampers and
quantification of their effect on user comfort.
A variety of tests are available to determine dynamic properties of Figure 13 - Dropping over tie-down
rods at abutment
structures, however, the Ambient Vibration Test stood out as the
most appropriate and convenient method. Due to the current availability of high precision
sensing equipment and sophisticated data analysis software packages, experimental modal
estimation, which once required extensive equipment, custom software, and precise knowledge
of the input forces and frequencies, can now be performed with a few accelerometers and a
laptop. The test setup involves positioning of accelerometers at locations of maximum estimated
modal displacement which are set to record simultaneous data sets of the accelerations measured
under ambient vibration of the structure.
The testing of the Banff Bridge utilized a commercially available software package for the
determination of the natural frequencies and associated mode shapes. The software applies local
degrees of freedom to the global geometry where accelerometers are to be placed. Higher
accelerations resulting from resonance at the structure’s natural frequencies appear as peaks on a
frequency vs acceleration plot, which made the modal frequencies easily identifiable. Finally, by
selecting the peaks on the graph, the software animates the dominant motion of the structure as
measured by the accelerometers. In this way we were able to visually confirm which mode
shapes we were measuring and compare with the analytical model.

Figure 14 - 2nd torsional mode (from AVT) Figure 15 - 2nd torsional mode (analytical model)

Once the true natural frequencies were known, it was time to

tune the dampers. Instead of simply tuning to the measured
frequencies, baseline testing was performed to estimate the
modal damping of the structure before and after tuning.
Resonance tests, which seek to excite only specific modes,
were performed by jumping up and down at the exact natural
frequency of the mode being excited. With the assistance of a
metronome, it was relatively easy to activate the modes of
interest by jumping at a location of high modal displacement.
Once resonance was achieved, the excitation was stopped and
measurements of the free decay response were taken. Figure 16 - Frequency domain decomposition graph
 Estimation of modal damping was determined by fitting of an exponential curve to the free decay
response according to equation (1).

 + 1 =   (1)

Acceleration Time History Exponential Damping Curve

1.1 0.05
1.08 0.045
1.06 0.04

Acceleration (g)
0.035
Acceleration (g)

1.04
1.02 0.03
1 0.025
0.98 0.02 y = 0.0332e-0.416x
0.96 0.015
0.94 0.01
0.92 0.005
0.9 0
0 5 10 15 0 1 2 3 4 5 6
Time (s) Time (s)

Figure 17 - Excitation and free decay of 2nd torsional mode Figure 18 - Fitting of exponential curve to free decay response

Since the two tuned mass dampers were designed to mitigate motions excited by the 2nd vertical
and 2nd torsional modes of vibration, these were of particular interest. The 2nd vertical mode was
predicted to be at 1.9 Hz and the damper’s vertical stiffness & mass were proportioned to have
the same natural frequency with room to add or remove mass for tuning. The measured
frequency of the 2nd vertical mode was significantly lower than anticipated at 1.5 Hz; meaning
the structure was softer than predicted, likely due to uncertainty in modelling the boundary
conditions. Frequency follows equation (2) and being unable to alter the stiffness provided by the

 = 2 (2)

cables suspending the carriage, the mass needed to be increased by over 600 kg (or 60% of its
predicted mass) for which we had made sufficient allowance in the damper carriage. By
contrast, the 2nd Torsional mode was predicted to be 3.1 Hz and was measured at 3.3 Hz. This
required an increase of only 50 kg to 375kg total. Once the masses of the dampers had been
calibrated to the bridge’s natural frequencies, an additional set of resonance tests was performed
for both the 2nd vertical and 2nd torsional modes. The accelerations and structural damping were
compared with the baseline tests performed before tuning.
The tuned mass dampers improve the dynamic performance by preventing the accelerations from
reaching unacceptable levels by participating in and out of phase at the natural frequencies; thus
preventing a resonant response. The results of the before and after tuning resonance tests
showed that the damping ratio increased slightly but remained at a relatively low 1.5-2%
depending on the mode under consideration. Vertical accelerations were reduced by over 50% to
less than 1 m/s2 during excitation of the 2nd vertical mode.
 In order to calibrate the model to match the field tested frequencies, very small adjustments to
suit the boundary conditions between the drop span and haunched girders as well as the bearing
surface between the girders and the precast pier cap brought the analytical frequencies precisely
in line with measured values. This sensitivity to actual site conditions is a major reason why
there is a continued need for field testing of vibration controlled structures.

7. Conclusion
Footbridges are an excellent opportunity to exploit the use of timber, especially in beautiful
settings. However great care in design and execution must be taken to account for its unique
properties, and to ensure the structure will remain an enhancement to the setting which all can
enjoy for many years to come. Proper credit goes to all involved, including a sympathetic client
with great aspirations!

Figure 19 - Finished bridge south bank Figure 20 - Finished bridge south elevation

Figure 21 - Opening day crowds Figure 22 - Illuminated walkway