th

4 INTERNATIONAL CONFERENCE

DESIGN AND CONSTRUCTION OF THE NORTH BANK BRIDGE

David GREENWOLD Simon FRYER Karl HAGLUND
Principal Engineer Associate Director Project Manager, New Charles
Ammann & Whitney Infrastructure Bridge River Basin
Boston, MA, USA Engineering Massachusetts Department of
dgreenwold@ammann-whitney.com Buro Happold Ltd Conservation and Recreation
London UK Boston, MA, USA
Simon.Fryer@BuroHappold.com Karl.Haglund@state.ma.us

Summary
The North Bank Bridge is a 213 m long bridge being built as part of a new riverfront park in Boston and Cambridge,
Massachusetts, USA. As part of a large scale restoration of the Charles River Basin, the project will transform a
60,000 m2 industrial site into inviting and vibrant parkland, while facilitating the flow of non-motorized traffic through this
heavily travelled area. Ammann & Whitney designed a sinusoidal bridge for the highly constrained site which was
conceptualized in cooperation with Buro Happold in a highly community oriented process. The tubular steel truss bridge
features fibre reinforced polymer decking. Analysis of the bridge included a thorough investigation of lateral vibrations
and global buckling. The project was initiated as a remediation measure for the Central Artery / Tunnel Project and was
ultimately constructed with American Recovery and Reinvestment Act funds. Construction is distinguished by the
collegial atmosphere and the Contractors efficient method of producing the irregular truss geometry.

Keywords: footbridge; tubular; truss; lateral vibrations; global buckling; fibre reinforced polymer (FRP)

1. Background

1.1 Site History

The Alster-damm in Hamburg made that city one of the most beautiful in the world, according to a Boston resident who
in 1859 published an etching of the Alster River in a memorial addressed to the state legislature. He was convinced that
Hamburg was an ideal model for Bostons future. Instead of reeking, dangerously polluted mudflats exposed at every low
tide, a dam near the mouth of the Charles River would create a great water park in the heart of the city. Fifty years later
an earthen dam was completed near the mouth of Boston Harbor, and landscaped esplanades were laid out on both
sides of the river. The Charles, once a foul threat to public health, became the centrepiece of Americas first regional
park system. With the river as foreground, Harvard University, the Massachusetts Institute of Technology, and Boston
University erected new campuses along the esplanades between 1913 and 1939.
At the mouth of the Charles, however, the esplanades were cut off from the harbour by a half dozen railroad bridges and
trestles that crossed the river to connect Greater Boston with northern New England. And when the Central Artery, a six-
lane elevated highway, was built through the centre of the city in the 1950s, it further isolated the Esplanade from the
harbour and divided and isolated neighbourhoods throughout the city.
Less than twenty years later, engineers began discussing the demolition of the highway. Long expanses of brick and
granite warehouses had been transformed into residences and new residential towers followed. The city had begun
linking sections of the harbours edge into parkland, and the regional parks agency drew up plans to connect the Charles
with the harbour for the first time. If the highway were replaced by a tunnel, according to these visionaries, the city would
be made whole once more.
This vision became reality with the execution of Bostons Central Artery / Tunnel Project (CA/T). The elevated highway
through the city was relocated to a tunnel on the same alignment, and transitioned to gorgeous new bridges over the
Charles River Basin. Federal regulations required mitigation for the impact of this work on existing and planned parks.
Further, the new bridges required the use of land owned by the regional parks agency. To settle these state and federal
requirements, the state highway department agreed to a program of pedestrian and bicycle paths and the construction of
800,000 m2 of new public spaces, including the construction of the North Bank Bridge.
 th
4 INTERNATIONAL CONFERENCE

1.2 Project History

The bridge design began with the derivation of multiple concepts and engagement with a very active local community.
Once a concept was chosen, the final design phase began. The unorthodox nature of the design required a thorough
analysis with several layers of review, including an independent review by Buro Happold. Responsibility for the project
was unavoidably transferred between Authorities, creating managerial challenges. Project partners faced significant
hurdles in transforming the industrial site into parkland related to existing infrastructure and contaminated soils.
In 2007, the Authority solicited bids for the park project and the low bid was approximately 20% higher than estimated.
Asserting that the bids exceeded the funds on hand, the Authority determined not to proceed with the work. When the
American Recovery and Reinvestment Act (ARRA) became law in 2009, high priority was given to projects that were
shovel-ready, with completed bid documents ready for advertisement. The North Bank Bridge was chosen to receive
funding. This time, the low bid was below the estimate and construction began in 2010.

2. Design
2.1 Conceptual Design

The first phase of design was carried out in

collaboration with Buro Happold and Julian
Hakes. Through a series of site visits and
design charettes, several concepts were
developed and presented to the community.
A number of options were studied in an attempt
to address the multitude of geometric
constraints. The constraints, taken from west to
east, are: clearance above an amphibious
vehicle (Boston Duck Tours) launch ramp,
clearance above railroad tracks bisecting the
park, clearance between a historic building
(Tower A) and a parallel highway ramp (Ramp
Figure 1 Concept sketch of bridge and site, from southwest looking northeast CT), clearance below the Leverett Circle
Connector Bridge, clearance above the Millers
River, and clearance below the Zakim Bridge. Taken together, these and other physical features represented a significant
challenge in arranging the structure.
The initial options included a rainbow arch and asymmetric solid industrial design but the preferred solution was
dubbed the sinusoidal bridge because of its snaking, undulating form in plan and elevation (see Figure 1). The bridge is
supported by a truss structure that is positioned alternately below and above the deck level according to the site
constraints.
At the west and east, approach embankments are provided to a height of approximately 3 m at either end of the
alignment. The structure commences with a minimal depth as the trusses sweep below the deck in the approach spans.
They then rise above the deck over the railroad so that the necessary 5.6 m clearance is achieved while minimising the
overall length of the structure. Through the transition, the trusses fold in close to the walkway, accommodating the
narrow gap between Tower A and Ramp CT. The alignment was also set to allow for the re-use of existing foundations,
minimising cost, risk, and disruption.
The selection of steel pipes for the truss members was made as these sections can be readily bent to form complex
curves in plan and elevation without a concern that imperfections will be noticeable. Above the railroad, a high sided
protective mesh screen is required and fit neatly within the elevated trusses.
The design is satisfying for an engineer for a number of reasons. Firstly, it is a simple form that is born out of the site
constraints. The bridge is a direct response to the project brief and would not work on another site. Secondly, the
bridges beauty comes from the expression of the structure alone and not because of any add-ons or cladding. Thirdly,
the collaborative team work that occurred in evolving the design concept was an enjoyable process but one which also
enabled the relevant stakeholders to give input at an appropriate stage to produce a memorable structure.
 th
4 INTERNATIONAL CONFERENCE

2.2 Final Design

2.2.1 General
The final design was founded on a holistic, iterative approach encompassing overall geometry; foundation and
substructure design; and analysis of global buckling, lateral vibrations, and individual members and joints; all while
prioritizing preservation of the approved concept, the user experience, maintainability, and constructability.
The tubular steel truss is made up of deck chords and sweeping outer chords connected by posts (dubbed vertical
members) spaced at 3.2 m on centre, and intermittently reinforced with diagonals. At each vertical, there is a floorbeam
that supports the fibre reinforced polymer (FRP) deck. The deck acts as a composite member with the truss. The shear
action of the deck is augmented with steel lateral braces as required. Throughout the length of the bridge, the sweeping
outer chords of the truss remain outboard of the deck chords so that the angle of the verticals to the walkway varies as
the outer chord follows its sinusoidal path.
The 7 span continuous structure has a total centreline length between abutment bearings of 212.73 m (see Figure 3).
The span arrangement is symmetrical about the main span with span lengths from west to east of 25.40 m, 25.40 m,
31.75 m, 47.63 m (railroad span), 31.75 m, 25.40 m, and 25.40 m.
The horizontal alignment of the bridge centreline consists of a simple reverse curve, with the point of reverse curvature
located at the middle of the bridge. The first curve has a radius of 304.8 m and the second has a radius of 396.2 m.
The profile of the bridge deck is symmetrical about the apex at the midpoint of the bridge. The profile consists of
constant approach grades of 4.9% (in accordance with the Americans with Disabilities Act), connected by a 41.3 m long
crest vertical curve (see Figure 4). The approach ramps leading up to the bridge spans at each end of the bridge carry
the pathway from grade up to the bridge at approximately a 5% slope.
The truss chords are 324 mm diameter pipes with a thickness of 25.4 mm. The verticals are 324 mm diameter pipes with
a thickness of between 9.5 mm and 12.7 mm. The diagonals are 168 mm diameter pipes with a thickness of 14.3 mm.
The floorbeams are 254 mm wide x 152 mm with a thickness of 12.7 mm. The lateral braces are 127 mm x 127 mm with
a thickness of 12.7 mm. The FRP decking is approximately 127 mm thick and is composed of top and bottom skins
joined by closely spaced webs, oriented longitudinally.
The original concept called for Vierendeel trusses which would require rigid joints. The guideline for rigid joints is to use a
ratio of 1:1 for chord:vertical diameter, and a ratio of 2:1 for chord:vertical thickness. Though the Vierendeel concept was
ultimately found to not be feasible, the use of the above ratios was continued to the greatest extent possible after adding
diagonals to the trusses to capitalize on the resulting joint strength.

Figure 2 Rendering of bridge and site, from southwest looking northeast

 th
4 INTERNATIONAL CONFERENCE

Figure 3 Plan of bridge and site

Figure 4 Elevation of bridge and site

2.2.2 Setting the geometry

In determining the geometry, there were
several major goals: follow the conceptual
design, avoid site constraints, re-use existing
foundations, and meet the accessibility
requirements of the Americans with Disabilities
Act. Further, the geometry is a major
determinant of the structures behaviour,
requiring an iterative approach.
To address the above a controllable,
repeatable approach was adopted in which the
geometry was determined parametrically. The
centreline was set as two circular arcs in plan,
set in relation to the existing foundations and
major site features and constraints.
Accessibility requirements largely determined
Figure 5 Typical section of bridge the vertical geometry of the centreline. The
deck chords were set to simply follow the
centreline. The outer chord geometry was then determined using an interesting algorithm.
First, a series of sinusoidal arcs were set in elevation relative to centreline. Next, a set of lines joined by an arc were
defined in cross section (see Figure 5). Given a longitudinal position along the centreline, the vertical location of the
outer chord was determined from the sinusoidal arcs. Given this vertical location, the horizontal location in section was
 th
4 INTERNATIONAL CONFERENCE

then determined from the set of lines joined by an arc. These sections were then radially applied to the three dimensional
centreline to get a set of three dimensional working points. All of this work was performed with a simple spreadsheet.
The data was used to automatically draw a solid model of the bridge. This model was then inserted into a model of the
site which included laser survey at the point of minimum clearance at the control tower. With this composite model, the
aesthetics and clearances were examined. The data was also used to rapidly generate the analysis model. Any issues
with the geometry found in the site model or the analysis model were easily addressed by adjusting the parameters of
the algorithm in the spreadsheet and performing another iteration. The finalized working points were presented as a
table in the Contract Drawings.
2.2.3 Global analysis model
The bridge was modelled with finite elements
using Lusas software. Foundations,
substructure, and steelwork were modelled as
beam elements and the deck was modelled
with shell elements (see Figure 6). The deck
acts compositely with the truss, and the model
includes the effects of staged erection of the
deck. The model was used to determine
member forces and to examine the structures
lateral vibration characteristics and stability.
Z Further details on the model can be found in
Y X the relevant sections below.

Figure 6 Lusas model of bridge

2.2.4 Foundations
The site has relatively poor soil conditions which are unsuitable for shallow foundations. Numerous subsurface
obstructions are known to exist, and vibrations due to pile driving were a concern at Tower A. Also, the new structure
incorporates existing drilled shafts that were constructed during the CA/T project for temporary ramps. It is generally
undesirable to mix multiple foundation types for a single structure because of the significant differences in lateral
stiffness. Drilled shafts were used extensively on the CA/T, so there was knowledge and equipment available in the area.
For these reasons, drilled shafts were chosen for the new foundations.
Analysis of the shafts used methods specified for use on the CA/T. Non-linear soil / shaft interaction was examined in a
stand-alone program. The program was first used to develop equivalent free standing cantilever lengths of shafts. These
lengths were used in the global analysis model. Forces thus found with the global analysis model at the top of the shafts
were then applied to the stand-alone models. Design forces were extracted and standard moment / axial force
interaction methods were used for design.
Leading up to the ends of the bridges, pathways are on embankments which vary in height from 0 m to approximately
3 m. Construction of these embankments was anticipated to result in consolidation of the underlying organic deposits
and subsequent large settlements. The soil was therefore pre-consolidated to avoid supporting the embankments on
deep foundations or using a lightweight fill. Unlike on the bridge itself, any future settlement on the approaches can be
addressed rather simply.

2.2.5 Deck
The bridge incorporates an FRP deck for reasons discussed in Section 2.2.6 Vibrations. It is typical to attach FRP panels
to steelwork with a relatively lightweight clip detail. This approach results in non-composite behaviour between the
decking and the steelwork. On the North Bank Bridge, the FRP deck is made composite with the truss with a high
strength bolted detail in which the lower skin of the deck is bolted to steel angles which are welded to the floorbeams.
The bolts are tensioned after the bridge is erected and the deck has been surveyed and shimmed into proper position.
The deck is therefore not composite for dead load. The analysis model reflects this staging.
As mentioned above, the deck was modelled with shell elements. The out-of-plane bending and shear behaviours are
 th
4 INTERNATIONAL CONFERENCE

not important to the global behaviour of the bridge, so the shells were made to represent axial, in-plane shear, and in-
plane bending behaviours only. The deck elements use orthotropic properties to reflect the actual material properties in
each direction. An equivalent thickness was calculated for each of the three desired behaviours. The calculated
thicknesses were very close to one another and the lowest was used. A comparison of hand calculations and results
from stand-alone FEM models of the deck verified this approach.

2.2.6 Vibrations
Conscientious of the ongoing interest within the engineering community in pedestrian induced vibrations, this issue was
discussed at length with the Authority at the earliest stages of design. After a thorough discussion, we were directed to
design for a minimum frequency of 3 Hz for the first vertical mode per the American Association of State Highway and
Transportation Officials (AASHTO) Guide Specifications for Design of Pedestrian Bridges. Since the frequency of the
horizontal cyclical load due to walking is half that of the vertical, we were directed to design for a minimum frequency of
1.5 Hz for the first horizontal mode (see Figure 7).
Meeting these requirements required several iterations of truss geometry, truss member sizes, location of truss
diagonals, bearing layout, and substructure and foundation design. It was also found beneficial to substitute an FRP
deck for the previously assumed concrete deck. Due to FRPs high stiffness to weight ratio, this had a dramatic effect on
the vibration behaviour of the bridge.
At the end of this process, the bridge met the above criteria, but it was deemed prudent to perform a more thorough
analysis. As is now widely known, on a bridge with a low lateral frequency and the correct distribution of pedestrian
lateral loading, a resonant response can occur. If a small group within a large crowd of pedestrians happens to be
walking in phase, this group can set up a resonant response, which will encourage others to synchronize with them.
Eventually, this feedback will produce motions that can become disturbingly large.
Armed with this knowledge, we applied a horizontal load due to walking as a distributed cyclic load at a range of
frequencies and found a maximum response at 1.6 Hz. The load applied was roughly equivalent to a dense crowd
walking with 100% synchronization. Scaling the response to 20% was taken to represent a random synchronization of
20% of the dense crowd. This scaled reaction was very small. We could find no published guidelines, but we judged that
at these very small levels of movement no feedback would occur. In searching the literature, we did find serviceability
limits for horizontal vibration in ISO 2631 (1980). We found that in order to produce an unacceptable structural response
per these guidelines, a synchronization rate of 75% of the dense crowd would have to occur.

Z X

Figure 7 Plan showing first lateral mode of vibration, f = 1.6 Hz

2.2.7 Buckling
At several locations along the bridge (most notably at the main span), the outer
chord is in compression and bracing against buckling is provided only by the truss
verticals. The degree of lateral restraint provided by the verticals is dependent on
their cross sectional properties, length, and distance from a support. Since these
factors vary along the bridge, the bracing they provide to the compression chord
also varies. The compression chords themselves follow a non-planar path. The
buckling behaviour of the bridge is therefore of concern and is non-trivial to
determine. A full nonlinear buckling analysis was therefore carried out.
The nonlinear analysis used a distributed load approximating the self weight of the
bridge (approximately equivalent to live load) applied in a variety of patterns. The
Figure 8 Detail of buckled outer chords load was incrementally increased. At each iteration, the deformed geometry for
at main span the previous iteration was used as the basis of the stiffness matrix. For the
purposes of this analysis, materials were assumed to be elastic. The deformations
 th
4 INTERNATIONAL CONFERENCE

at each iteration were examined and the structure was found to behave linearly up to approximately fifteen times the
starting load (see Figure 8). This demonstrated that the behaviour of the structure is linear well beyond the material
limits. Therefore, a linear analysis was used for the general analysis and stresses were limited to the elastic range.
Similar analyses were carried out on segments of the bridge to verify constructability.

2.2.8 Fatigue
Welded tubular joints typically have
pronounced local stress concentrations,
Z commonly referred to as hot spot stresses. The
Y
range of stress at the hot spot must be kept
X
within a given allowable range. For most
locations, a stress concentration factor was
found in the literature. At floorbeam to deck
chord connections, no stress concentration
factors could be found so a finite element
model was used (see Figure 9). The
rectangular floorbeam and circular deck chord
were modelled with shell elements. Member
Figure 9 Shell element model of floorbeam to chord joint loads taken from the global model were applied
to find the hot spot stress ranges directly.

2.2.9 Low maintenance

The steelwork is composed of closed sections and the joints are completely sealed. This limits interior corrosion by
preventing the cycling of air and intrusion of water. To minimise maintenance of the exterior protection system, the
steelwork is metalized, rather than painted. Further low maintenance details include the FRP deck, stainless steel safety
infill, and LED lighting. Dissimilar materials were scrupulously isolated to protect against galvanic corrosion.

2.2.10 Pedestrian experience

The pedestrian experience was treated with great care. All butt welds are ground smooth. The multi-piece curb was
carefully detailed to allow access for the various trades during construction and a pleasing appearance to pedestrians in
service. Ammann & Whitneys railing concept is bridge specific. To minimise visual clutter, the posts are layed out with
the truss bays, leaning away from the user at the ultimate slope of truss. The horizontal infill of tensioned wires
maximizes transparency to the user, but creates a ladder effect which can be dangerous for children. A return at the top
of the railing post simultaneously eliminates this ladder effect and presents the handrail back to the user. Lighting is
provided by an LED light strip integrated into the handrail. Required safety infill is provided in each truss bay at the
railroad span. A tensioned stainless steel mesh infill is used, eliminating additional cross frames required by rigid infill.

Figure 10 Rendering of bridge and site from bridge users perspective

 th
4 INTERNATIONAL CONFERENCE

3. Construction
The specific capabilities of the Contractor suggested a series of small, but important modifications to the plans. The
Contractor, Consultant, and Authority worked closely to resolve each issue with either no reduction in quality or an
improvement. This spirit of cooperation carried over to more mundane aspects of construction, such as dealing with
unforeseen buried conditions. The Authority deserves a large share of the credit for this success for fostering an
environment of open communication.
The Contractor proposed a simple and effective method for fabricating the complex geometry of the truss chords. First, a
non-uniform rational basis (NURB) spline was defined using the working points provided on the drawings. This spline
was used as the reference geometry. The spline was then broken down into a series of lengths which matched the
length of pipe which could be easily procured. Each segment was then approximated by a planar element, consisting of
discrete bends and tangents. Each planar element is connected to its neighbour with the ends timed together such that
the overall reference geometry is approximated with these planar elements to within 2 mm. A three dimensional solid
model of the proposed finished geometry to be produced by this method was provided to the Authority by the Contractor,
examined by the Consultant for aesthetic integrity, and approved.
For ease of handling the pieces, the trusses are fabricated on their sides. The coordinates supplied in the contract
documents are converted into relative shop coordinates. A target and total station are used to precisely locate pipe
supports. Each chord segment is set in a minimum of three supports. Bending takes place on site so that any segment
found to not fit properly into the surveyed supports can be easily returned to the bending process. Completed truss
halves (north and south) are temporarily connected into sections and surveyed. Adjacent bridge sections are then pre-
assembled prior to shipping.
The sections of the bridge are shipped over the road with a field joint at the middle of each floorbeam. The bridge was
designed such that the full cross section could fit through all marine obstructions, but an optional floorbeam splice was
provided to expand the pool of bidders. All field splices are detailed with a retractable backing ring.

4. Acknowledgements
This project could not have been completed without the hard work of the people at the Central Artery / Tunnel Authority,
MassDOT, and the Department of Conservation and Recreation; nor without the support of Governor Deval Patrick and
the MassDOT Board; nor without the dedication of the community, including the Citizens Advisory Committee for the
New Charles River Basin, the Charles River Watershed Association, MassBike, the Charles River Conservancy,
WalkBoston, and the Conservation Law Foundation; nor without Ammann & Whitneys design partners: CRJA / Carol R.
Johnson Associates, Stantec, GPI / Greenman-Pedersen Incorporated, and Buro Happold.

5. Conclusion
The Central Artery / Tunnel Project removed a cleft in the city and allowed it to become whole again. As one of Central
Arterys mitigation commitments, the North Bank Bridge removes a similar cleft between the riverfront parks and the
harbour. It is our sincere hope that the bridge is found to be useful and enjoyable to the public.

Figure 11 Rendering of bridge and site