Construction stresses

in the world’s first precast

concrete network arch bridge

Hossein Yousefpour, Todd A. Helwig, and Oguzhan Bayrak

A
rches provide a structural system that can efficient-
ly support large loads while lending themselves to
excellent aesthetics. Historically, arches have been
widely used in bridge systems; however, in modern ap-
plications, they are usually reserved for signature bridges,
where aesthetics play an important role in the design. Be-
cause arches primarily resist loads through compression,
concrete is an ideal structural material for their application.

While construction techniques and analytical capabili-

ties have systematically improved, relatively few concrete
arch bridges were built in the past 50 years,1 mostly due to
their high construction costs. Concrete arches are usually
built using timber or steel falsework or cantilever methods,
which are time and labor intensive.2 As a result, structural
engineers have been especially interested in improving
■  The West Seventh Street Bridge in Fort Worth, Tex., built construction techniques for this efficient, aesthetic struc-
in 2013 and comprising 12 prestressed, precast concrete tural form.
network arches, was instrumented with vibrating-wire gauges
embedded in the arches prior to concrete placement. An innovative solution for constructing concrete arches
was used for a signature bridge on West Seventh Street
■  The instrumentation provided data on the stresses induced in in Fort Worth, Tex. The bridge consists of twelve 280-ton
the arches during posttensioning, handling, and transport as (2500 kN) concrete network arches, which were cast on
well as deck construction to ensure no cracking. their sides, rotated into a vertical position, transported, and
installed. To withstand the substantial stresses induced by
■  The measurements also provided a means for evaluating the rotation and transportation, the arches were prestressed in
accuracy of stress calculations that were made during design. both the tie and the rib.

30 S ept e mb e r – Oc t o b e r 2 0 1 5 | PCI Journal

Table 1. Examples of recently constructed steel network arches
arch bridge in the world.17 Despite Tveit’s early sugges-
Arch span Year
Bridge Country tions to build network arches using high-strength concrete
length, ft completed
ribs,3 the authors have not been able to find records of any
Bugrinsky Bridge Russia 1247 2014 previously built concrete network arches. Therefore, it is
likely that the West Seventh Street Bridge is also the first
Troja Bridge Czech Republic 657 2014 concrete network arch bridge in the world. Prefabrication
Lake Champlain Bridge United States 480 2011 of the concrete arches in this project significantly re-
duced on-site construction time and allowed better quality
The Brandangersundet control. However, the designers predicted that some of
Norway 722 2010
Bridge the most critical times in the life of the arches happened
Florabrücke Germany 435 2010 during construction. As a result, accurately estimating the
stresses during construction was critical. While sophisticat-
Blennerhassett Island ed finite element models were used to predict the stresses
United States 878 2008
Bridge in the structure, the possibility of damaging the arches
Palma del Río Bridge Spain 427 2008 during construction remained a concern. Therefore, a field
monitoring study was initiated to evaluate the perfor-
The Providence River mance of the arches during construction. As a part of field
United States 400 2007
Bridge monitoring, the arches were instrumented and data were
Note: 1 ft = 0.305 m.
collected and interpreted to ensure the safety of the arches
and verify the design assumptions.18

A network arch is a tied arch bridge with inclined hangers, This paper presents the findings from the instrumenta-
in which each hanger crosses at least two other hangers tion of the West Seventh Street Bridge with a focus on
in the plane of the arch. These densely arranged hangers short-term stresses during handling. A brief overview
provide a nearly continuous shear transfer between the rib of the innovative design of the bridge is presented. The
and the tie and therefore greatly reduce the bending mo- instrumentation, monitoring, and interpretation of data are
ments and deflections in the arch elements, which results in then described. Finally, comparisons are made between the
significant material savings.3 Network arches were first in- stresses measured in the structure and those predicted in
troduced by Per Tveit in the 1950s as lightweight steel arch design calculations.
bridges with prestressed concrete decks. The world’s first
network arch bridges, including the Steinkjer Bridge and The West Seventh Street Bridge
the Bolstadstraumen Bridge in Norway and the Fehmarn
Sound Bridge in Germany, were constructed in the early The new West Seventh Street Bridge was designed to
1960s. They had span lengths of 262 ft (79.9 m), 275 ft replace a century-old bridge that connected downtown Fort
(83.8 m), and 814 ft (248 m), respectively. In the following Worth to the cultural district. The bridge spans four lanes
years, many other steel network arches were built around of traffic, the Clear Fork of the Trinity River, and a number
the world (Table 1).3–15 of recreational trails.

Steel network arches are usually constructed using prefabri- The aesthetics of the replacement bridge were of impor-
cated segments. Due to their light weight, it is also com- tance to city officials. The majority of new bridges in
mon to assemble the whole steel skeleton with parts of the Texas are precast concrete girder bridges, which could
deck at a temporary construction site and transport the span also have provided an economical solution for this project.
into position. A typical construction sequence begins with However, the new West Seventh Street Bridge was expect-
assembling the tie element and the deck on temporary shor- ed to be a signature bridge and a pleasant gateway to five
ing. In the next step, another temporary structure is con- internationally renowned museums in the cultural district.17
structed on the deck to support the arch rib segments. Once On the other hand, due to high traffic demands, the new
all rib segments are erected and properly connected, the bridge needed to be constructed as quickly as possible.
hangers are installed and stressed as needed. To achieve the
desired forces in the hangers, a detailed stressing sequence Site conditions allowed engineers to use six uniform spans
is developed based on finite element simulations.16 The se- of 163.5 ft (49.83 m). Therefore, they conceived an inno-
quence might require multiple rounds of adjustment. How- vative solution comprising 12 identical precast, prestressed
ever, due to high static indeterminacy in these structures, concrete network arches. While the decision to use arches
the resulting stresses are highly dependent on the modeling was highly influenced by aesthetic considerations, the
assumptions, construction imperfections, and temperature. identical design of all arches and the possibility of precast-
ing resulted in a significant reduction in construction costs
The West Seventh Street Bridge, which was completed in and the time of street closure and made precast concrete
2013, is believed to be the first precast concrete network arches feasible. Figure 1 shows the layout of a typical span

PCI Journal | S e p t e m b e r – O c t o b e r 2015 31

 Road surface
Arch

Sidewalk
Floor beams
(17 per span) Precast concrete panels

Details of a typical span (Courtesy of Joel Blok)

6 ft
R = 1.1

Figure 1. The new West Seventh Street Bridge. Note: R = radius of curvature. 1 in. = 25.4 mm; 1 ft = 0.305 m.

and the detailed geometry of the precast concrete arches. The precast concrete arches were fabricated in a casting
The bridge carries four lanes of traffic and two sidewalks, yard less than 1 mi (1.6 km) from the bridge location. The
which are located outside the arches that support the spans. steps for constructing the arches of the West Seventh Street
Each concrete arch includes 52 hangers, which are located Bridge were as follows:
in two parallel planes, spaced 2 ft (0.6 m) apart (Fig. 1).
The 26 hangers in each plane are parallel to each other, all 1. The arches were cast on their sides. Table 2 gives the
with an angle of 35 degrees from the vertical. However, properties of the concrete used in the arches.
the hangers in two planes are inclined in opposite direc-
tions, resulting in a mesh that is typical of a network arch. 2. To prevent cracking during rotation, a first stage of
The deck is constructed using precast concrete panels with posttensioning was conducted on the arches: two ten-
a cast-in-place concrete topping slab and is supported by dons in the rib were stressed to 208 ksi (1430 MPa),
17 prestressed concrete floor beams, which are suspended and four tendons in the tie were stressed to 104 ksi
from the arches using posttensioned bars (Fig. 1). The floor (717 MPa). Each tendon consisted of nineteen 0.62 in.
beams are pretensioned elements with a nominal depth of (16 mm) strands. Figure 2 shows the posttensioning
5 ft 6 in. (1.7 m) at midspan, which is tapered to a depth of tendon layout.
3 ft (0.9 m) at the arches and further to 1 ft 9 in. (0.53 m) at
the ends. These floor beams are 1 ft 4 in. (0.40 m) wide and 3. The hanger elements were installed. Each hanger was
are spaced at 9 ft 71⁄2 in. (2.93 m). passed through a hanger tube in the tie and threaded

32 S ept e mb e r – Oc t o b e r 2 0 1 5 | PCI Journal

 Table 2. Properties of the concrete used in the arches
were supported by a gantry system through a series
Parameter Quantity
of equalizer beams and lifting ropes above the rib
Required 56-day compressive strength, psi 8000 and below the tie. All lifting points were first raised
equally. Once clear of the formwork, only the back
Target 28-day compressive strength for mixture design, psi 7900
lifting points (at the rib) were raised, allowing the
Typical slump, in. 9 arch to pivot to its final vertical orientation (Fig. 4).
After completing the 90-degree rotation, the arch was
Cementitious-material content, lb/ft3 25.9 moved laterally and positioned on temporary supports,
Fly ash replacement ratio (Class F), % 25 each at a distance of 7 ft (2.1 m) from the end of the
arch.
Water–cementitious material ratio 0.36

Aggregate/cement ratio 4.4 5. Because the small gap between the arches in their final
position did not allow any posttensioning, all stressing
Typical air content, % 1.4 needed to be completed in the precasting yard. There-
Note: 1 in. = 25.4 mm; 1 ft = 0.305 m; 1 lb = 4.448 N; 1 psi = fore, a second stage of posttensioning was conducted
6.895 kPa. in which the tie tendons were stressed to 208 ksi
(1430 MPa) and the rib tendons were detensioned to
104 ksi (717 MPa).
into a clevis at the rib. A nut was put at the other end
of the hanger and was hand tightened. 6. To prestress the hangers, an upward jacking opera-
tion was conducted (Fig. 4). Hydraulic rams were
4. The arches were rotated into a vertical position us- positioned under the tie at the locations of future floor
ing a lifting assembly in which six lifting frames beams and were simultaneously activated to push the
engaged the sides and bottom surface of the rib and tie up. When the rams were active, the sag was re-
the tie. These lifting frames were symmetric about moved from the hangers and the nuts were retightened
the midspan of the arches (Fig. 3). The lifting frames (Fig. 4). The rams were then deactivated. As a result,

81.58 ft Section
4.50 ft
18.00 ft 32.02 ft
Rib tendons

0.88 ft
(nineteen 0.62 in.
R = 80.00 ft

diameter strands)
R = 79.94 ft
R = 165.06 ft
R = 79.94 ft

R=
2.00 ft

R= R=
Section
4.50 ft

Tie
R = 79.94 ft

R = 79.94 ft
0.24 ft

0.71 ft

bottom tendons
0.42 ft

(nineteen 0.62 in.

diameter strands)

6.63 ft 11.13 ft 20.51 ft 9.63 ft 14.44 ft 4.81 ft

4.81 ft 4.81 ft 4.81 ft
R = 100.00 ft

R = 79.94 ft

R = 79.94 ft

R = 79.94 ft

R=
1.58 ft

Section
4.50 ft
0.93 ft

Tie
R = 79.94 ft

top tendons
R=

(nineteen 0.62 in.

diameter strands)

27.99 ft 6.98 ft 9.62 ft 14.44 ft

3.30 ft 4.81 ft 4.81 ft 4.81 ft 4.81 ft

Figure 2. The prestressing layout in the arches. Note: R = radius of curvature. 1 in. = 25.4 mm; 1 ft = 0.305 m.

PCI Journal | S e p t e m b e r – O c t o b e r 2015 33

 Lifting frame 3

Lifting frame 2

Lifting frame 1

13.00 ft 25.00 ft 29.33 ft 14.25 ft

Figure 3. Locations of the lifting frames used for arch rotation. Note: 1 ft = 0.305 m.

the self-weight of the tie induced a prestress in the For design calculations, engineers used an analysis model
hangers. To stiffen the rib near the knuckle and prevent of the bridge in which the rib and tie elements were mod-
excessive tensile stresses during the upward jacking eled using three-dimensional (3-D) thick beam elements.
operation, prestressed concrete strongbacks (Fig. 4) Thick shell elements were used for modeling the knuckle
were clamped to the rib before upward jacking. These region of the arches. The models also included the effects of
strongbacks remained attached to the arches until all staged construction and time-dependent behavior of concrete
floor beams were installed in each span. according to CEB-FIP 1990.19 The hangers were modeled us-
ing 3-D bars. Because the hangers were not expected to resist
7. Once all arches and new piers were constructed, the compressive forces, analysis results were unrealistic when
arches were moved from the precasting yard to the compression was predicted in the hangers. In such cases, the
new piers. Two self-propelled modular transporters hanger configuration was revised to make sure all hangers
carried each arch from the precasting yard to its final remained in tension, and the analysis was repeated.17
location (Fig. 4), where the arch was lifted by cranes
and installed on bearings. Using an initial eigenvalue buckling analysis of the un-
cracked completed structure in the bridge analysis soft-
8. When all arches were transported to their final loca- ware, the design team predicted the load factor for the low-
tions and properly braced, the street was closed and est buckling mode to be 13.3 for AASHTO LRFD Bridge
the old bridge was demolished. The floor beams Design Speciafications20 service I load combination with
were then installed, and the construction of the deck six lanes of traffic and wind load. The associated buckling
for the new bridge began immediately to minimize mode was the out-of-plane deformation of the rib, with a
traffic interruption. maximum deformation at the crown. Further studies using
nonlinear buckling analyses also showed that with a load
This accelerated construction procedure served to limit the factor of 2 for AASHTO LRFD specifications strength
street closure to 120 days. However, because the arches ex- III load combination, which is related to transverse wind.
perienced several posttensioning and handling operations, The maximum out-of-plane displacement of the rib at
the design team needed to make careful decisions to ensure the crown was limited to 2 in. (50 mm), and the load-
satisfactory performance of the arches during construction displacement relationship was nearly linear.17 As a result,
and in service. the uncracked arches were found to be sufficiently stable
against buckling. However, concerns were raised regarding
To facilitate rotation and transportation, the design team the possibility of cracking during construction.
tried to minimize the weight of the arches by making the
rib and the tie elements as slender as possible. Moreover, Cracking is a serviceability concern in most reinforced
to lower the center of gravity of the arches, a relatively and prestressed concrete structures. However, potential
large span-to-rise ratio of 7.6 was used, resulting in cracking of the arches during the construction of the
a height of only 23.5 ft (7.16 m) at the crown of the bridge could also significantly reduce the stiffness of the
arches. Due to the low rise of the arches and also eco- rib elements and result in potential out-of-plane instabil-
nomic and aesthetic considerations, the design included ity in the finished bridge, especially because no top lateral
no cross bracing for the rib, but the lateral stability of bracing was used. Therefore, several measures were taken
the bridge is provided by the frame action that is created during design to make sure the arches did not experience
by moment connections between the two arches and the excessive tensile stresses so that they would behave like
floor beams. uncracked elements.

34 S ept e mb e r – Oc t o b e r 2 0 1 5 | PCI Journal

 Arch rotation Arrangement of hydraulic rams and strongbacks
during upward jacking

Retightening
the hanger nuts
when the rams
were activated

Arch transportation

Figure 4. Construction operations on the precast concrete arches.

The rib is inherently a compressive element. However, the self-weight of the arches generated a small portion
the arches were cast horizontally and their ribs were not of the service load tension in the tie, and as a result, the
subjected to compression until the arches were rotated tie element was subjected to a significant compressive
into the vertical orientation. Therefore, excessive tensile force during the second stage of posttensioning. To avoid
stresses in the rib were likely during rotation. The design potential instability in the tie, the design included a series
team chose to minimize the risk of cracking in the rib by of small curves in the duct paths so that the tendons would
stressing the two rib tendons to 208 ksi (1430 MPa) prior be in contact with the wall of the ducts after a small lateral
to rotation. After rotation, the self-weight of the arches displacement.17 As a result, the second-order displacement
provided some compression in the rib, and the initial pre- of the tie was minimized.
stressing was no longer necessary. However, the stresses
due to the self-weight were not large enough to ensure that While significant efforts were made by the design team
the ribs remained fully in compression during the remain- to minimize the risk of cracking and prevent instability in
ing construction operations, especially upward jacking. As the arches, uncertainties existed about the modelling as-
a result, the designers detensioned the tendons to 104 ksi sumptions and the resulting stress calculations, especially
(717 MPa) after rotation to ensure compression in the arch because the structure was the first of its kind. On the other
rib while also avoiding unnecessary prestress. The in-plane hand, the work plan for the arch-handling operations was
stability of the tie element during construction was another primarily developed by the contractor and was out of the
important concern. The tie element transfers the horizontal direct control of the design team. As a result, ensuring that
thrust between the supports through tension, and a rela- the arches were not experiencing excessive tensile stresses
tively high prestress is needed to prevent cracking in the was possible through the instrumentation program high-
tie when the bridge is subjected to service loads. However, lighted herein.

PCI Journal | S e p t e m b e r – O c t o b e r 2015 35

 ib) me
2
ck le (r g fra
Knu Li f t i n
(rib) e 3
n g fram
Lifti (rib)

le
uck spa
n
Kn ie) i d
(t M ib)
(r
e 2
g fram
in
Lift (tie)
e 3
A vibrating-wire gauge tied to the arch reinforcement
g fram
in
Lift (tie) n
spa
Mid ie)
(t
Arrangement of vibrating-wire gauges in different sections
of the first two arches

Instrumented in arches 1 and 2 only

Instrumented in arches 1 to 4 only
Instrumented in all arches

Instrumented sections in different arches

Figure 5. Arch instrumentation details.

Instrumentation Figure 5 also shows the location of the individual VWGs

embedded in the first two arches. The instrumented sec-
The West Seventh Street Bridge was instrumented tions were selected in coordination with the design team
using 224 vibrating-wire gauges (VWGs). A VWG and included the midspan sections, sections located over
includes a strain transducer and a thermistor, which the lifting frames during rotation, and sections in the
allow for both strain and temperature measurements knuckle region. The distribution of the VWGs within each
with resolutions of 1 με and 0.5°C (0.9°F), respec- of these sections was selected based on the expected strain
tively. The VWGs were installed immediately prior to profile in the cross section. According to St. Venant’s prin-
the assembly of the outside forms for each arch. Each ciple, disturbed regions with nonlinear strain distribution
VWG was attached to a no. 3 (10M) steel reinforcing are assumed to exist within one member depth from the
bar, which was tied to the transverse reinforcement location of any discontinuity in load or geometry. In other
of the arch. Figure 5 shows a VWG installed in the parts of the structure, which are assumed to be nondis-
reinforcing cage of one of the arches prior to con- turbed, plane sections remain plane and the distribution of
crete placement. strains is linear. In nondisturbed regions, linear interpola-

36 S ept e mb e r – Oc t o b e r 2 0 1 5 | PCI Journal

tion or extrapolation can be used to calculate the strains sure the modulus of elasticity Ec and compressive strength
and stresses at any point in the cross section. However, f c' of the concrete used in the arches. To confirm satisfac-
in disturbed regions, the interpolation or extrapolation tory placement of concrete in the arches, the contractor
of strains is invalid due to the nonlinearity of the strain constructed a mock-up segment before casting the first
distribution.22 The nondisturbed regions of the structure, arch. The concrete from the mock-up was used to prepare
which included all instrumented sections except those in forty-eight 4 × 8 in. (100 × 200 mm) concrete cylinders.
the knuckle region, were instrumented using three or four All 48 concrete cylinders were tested for modulus of elas-
VWGs (Fig. 5). As a result, the stresses at every corner ticity according to ASTM C469.23 Because the specimens
of these sections could be found using the plane section were loaded only up to 40% of their compressive strength
assumption. However, for the knuckle region, the strain in the modulus test, they were assumed to remain in the
profile was expected to be highly nonlinear and the VWGs elastic range. Therefore, 40 of these cylinders were also
were expected to represent the local strains. Therefore, in tested for compressive strength according to ASTM C39.24
these sections, only two VWGs were installed in the antici-
pated locations of maximum stresses. The results of these tests were used to develop Eq. (1),
which is a mixture-specific equation with a format
The first two arches were instrumented at every section frequently used to correlate Ec and f c' for high-strength
where the designers predicted high stresses so that the safe- concrete.25
ty of the arches could be ensured during construction. The
number of instrumented sections was gradually reduced for
subsequent arches. Figure 5 shows the instrumented sec-
=Ec 39 f c' + 1350 (1)
tions in different arches of the bridge.

To improve the flexibility of monitoring for the highly where

mobile arches, a wireless data-acquisition network was
used. The wires from the embedded VWGs were con- Ec = modulus of elasticity of concrete, ksi
nected to a data-collection box, which received the data
and then sent them to the data-acquisition system through f c' = compressive strength of concrete, psi
wireless communication. The wireless connectivity not
only eliminated the lengthy wires on the construction To consider the effects of different curing temperatures in
site but also reduced the number of channels needed on the structure compared with test cylinders, the in-situ Ec
the data loggers. The data-acquisition system was also values were estimated based on the compressive strengths
connected to a cellular modem, which enabled remote obtained from the maturity method26 and Eq. (1). The
monitoring of the structure. authors used the results of a maturity study by the con-
tractor, which correlated the compressive strength with
The arches were monitored during construction until the the maturity of the concrete used in the arches. Using the
bridge was opened to traffic. Depending on the speed of temperature measurements from VWGs, three average
construction activities, different scan rates were used for temperatures were calculated at any point in time: the
monitoring the VWGs. The maximum possible scan rate average rib temperature, the average tie temperature, and
for a single VWG with the available interface analyz- the average knuckle temperature. Using these temperatures
ers was once every 2 seconds; however, the sensors were and the results of the maturity study by the contractor, the
scanned sequentially, so the scan rate was reduced. To compressive strength of the concrete was independently
achieve a suitable scan rate, more data loggers were added calculated for the rib, the tie, and the knuckle. However,
to the network when several arches were under construc- because the first construction operation was conducted
tion simultaneously. The resulting configuration allowed when the arches were six days old, the difference between
the researchers to scan the gauges every 150 seconds, the strengths estimated for the rib, the tie, and the knuckle
which made it possible to detect changes during rapid was found to be insignificant during the first stage of post-
construction operations, such as posttensioning. Scanning tensioning and later construction operations. Therefore,
continued hourly when no construction activity was in the average of these three compressive strength values was
progress to capture the effects of temperature fluctuations used to develop a continuous correlation equation for f c'
and time-dependent effects on the structural behavior. versus time (Eq. [2]). In developing this mixture-specific
equation, a format was implemented that is similar to what
Estimating the mechanical was used in a model by Gardner and Lockman, known as
properties of concrete GL2000.27

Realistic values of modulus of elasticity and compressive  t 0.75 

strength of concrete were essential in processing the data. f c' ( t ) = 7500  0.75 
(2)
Therefore, a material test program was conducted to mea-  1.75 + 0.8t 

PCI Journal | S e p t e m b e r – O c t o b e r 2015 37

 where This method neglects the stress changes that occur when
no construction activity is in progress. For example, the ef-
t = concrete age after casting, days fects of prestress losses remain undetected by this method.
To include the effects of long-term changes on stresses in
This equation was combined with Eq. (1) to estimate the the arches, the authors used a more sophisticated data-
in-situ Ec values at each age. More details related to the processing method that incorporates the time-dependent
procedure for maturity calculations and the development of deformations of concrete in stress calculations. However,
Eq. (1) and (2) can be found elsewhere.18 the effects mentioned will be covered in a later publication
and are expected to have a minimal effect on the short-
Processing of data term stress changes that are the focus of the present paper.

A considerable processing effort was needed to interpret In order to eliminate the effects of thermal changes on
the data obtained from the instrumentation. The raw data calculated stresses, the structure was always compared
included strains in the VWGs and temperatures at the loca- between data points with equal temperatures. During step-
tions of these sensors, which were used to calculate the wise construction operations, such as posttensioning, the
stresses at the corners of the instrumented cross sections in stress changes due to stressing of individual tendons could
the structure. be calculated because thermal changes during stressing of
each tendon were negligible. However, for slower opera-
The first step in stress calculations was to calculate tions, such as arch rotation or transportation, stress chang-
the strains at the corners of the instrumented cross es were obtained by comparing the readings taken during
sections. For sections in nondisturbed regions of the the nights before and after the specific event. Temperature
arches, plane sections were assumed to remain plane, effects were minimized by identifying times before and
and strains at any point in the cross section could be after the event with equal temperatures. Finding two data
calculated using analytic geometry. However, in the points that included exactly the same temperatures at the
disturbed regions of the arches, such as the knuckle locations of all VWGs was not practical. However, if the
region, the distribution of strains was highly nonlinear following criteria were satisfied, thermal effects were
and therefore the strains could be calculated only at the assumed negligible between the two records: both points
locations of the VWGs. were recorded overnight (after sunset and before sunrise)
so that the effects of sunlight on nonuniform heating of
The measured strains included several time-dependent the arches were minimized; the average temperature of the
and environmental effects, including temperature changes, arch, found from averaging all temperature measurements
creep, and shrinkage. These components needed to be ex- from the VWG thermistors, was not more than 1°F (0.6°C)
cluded from the strain history so that stress-related strains different between the two records; and the temperatures
could be multiplied by the modulus of elasticity of the from none of the VWG thermistors differed more than 5°F
concrete to calculate the stresses. (2.8°C) between the two records.

To minimize the effects of creep and shrinkage in the Results and discussion
calculated stresses, the strain changes due to each construc-
tion stage were calculated separately and multiplied by Posttensioning response
the corresponding modulus of elasticity at the time of that
construction operation. The total stresses were estimated Figure 6 shows the stress changes in a typical arch dur-
by adding the stress increments due to each construction ing the first stage of posttensioning, and Fig. 7 shows the
operation, as expressed in Eq. (3). stress changes during the second stage of posttensioning
on the tie, which was conducted before detensioning the
n rib tendons. The instrumentation was capable of detecting
σ ( t ) = Σ Eci ∆ε i (3) the stressing of individual tendons inside the rib and the
i =1
tie. As a result, the recorded data provided the opportunity
to evaluate the response of the structure and to check the
where analytical models of the bridge with respect to each tendon
separately. Moreover, the interaction of rib and tie elements
Eci = modulus of elasticity of concrete at the time of the ith could be evaluated by comparing the stress changes in the
construction operation rib while the tie was being posttensioned and vice versa.

∆εi = strain change in concrete due to the ith construction Figures 6 and 7 show that the responses of the arch rib and
operation the tie were relatively independent of each other during
posttensioning. For each of the tendons, posttensioning was
σ = stress in concrete applied in increments of 52 ksi (360 MPa). During the first

38 S ept e mb e r – Oc t o b e r 2 0 1 5 | PCI Journal

 Stresses at midspan, rib, top Stress at midspan, rib, bottom

Stress at midspan, tie, top Stress at midspan, tie, bottom

Figure 6. Measured stresses in arch 2 during the first stage of posttensioning. Note: R = prestressing the rib; T = prestressing the tie. 1 ksi = 6.895 MPa.

stage of posttensioning, each increment in the rib tendons their axial stiffness, axial deformations happen in each of
increased the stresses at the edges of the rib at midspan by these elements without inducing noticeable bending stress-
more than 0.10 ksi (0.7 MPa) (Fig. 6 top left, R intervals), es in the other. Although the stiff knuckle region provides
but such an increment in the rib caused minimal stress some local restraint for the deformations of the rib and
changes at midspan in the tie (Fig. 6 bottom left, R inter- the tie, the effects of the slight interaction between these
vals). The maximum stress change anywhere in the tie when elements are mostly limited to the knuckle region and do
a prestress increment was applied to the rib was smaller than not materially affect the overall response of the rib or the
0.02 ksi (0.14 MPa). During the second stage of posttension- tie. The authors also observed this behavior through finite
ing on the tie, stress changes show a similar trend, and negli- element simulations of the arches, which are not within the
gible stress changes could be observed in the rib (Fig. 7). scope of this paper.

The small interaction between the rib and the tie during Another important observation was that despite the curved
posttensioning is in contrast with the response of the arches shape of the arch rib, posttensioning did not induce signifi-
to future external loads. For vertically oriented arches, cant bending in this element. This behavior is attributed to
external dead or live loads mobilize arch action, which the circular profile of the arch and careful selection of the
induces compression in the rib and tension in the tie to tendon paths and anchorage orientation by the designers.
provide static equilibrium with vertical reaction forces at The response of the arch rib to posttensioning is similar
the supports. Posttensioning induces compressive stresses to a compressive ring, which is in compression due to the
in the rib as well. However, the geometric design of the effect of pressure applied along its radius (Fig. 8). This
arches and tendon profiles were carefully chosen so that characteristic of the arch design was important in reducing
the anchors for the rib tendons had a relatively small ec- the long-term bending deformations in the arch rib under
centricity from the centroid of the rib. As a result, static sustained posttensioning forces.
equilibrium between internal stresses in the rib and the
rib anchorage forces does not require large shear or axial Rotation response
forces in the tie element or significant bending moments
in either the rib or the tie. On the other hand, because the Figure 9 shows the three stages of rotation. The first stage,
bending stiffness of the tie and the rib is much smaller than vertical lifting, occurred when all lifting points were raised

PCI Journal | S e p t e m b e r – O c t o b e r 2015 39

 Stresses at midspan, rib, top Stress at midspan, rib, bottom

Stress at midspan, tie, top Stress at midspan, tie, bottom

Figure 7. Stresses in arch 2 during the second stage of posttensioning on the tie. Note: T = prestressing the tie. 1 ksi = 6.895 MPa.

equally at the beginning of rotation. The second stage, C and G, which were located at the bottom of the rib and
supported rotation, represents the change of the arch from the tie before rotation. However, Fig. 10 shows that during
a horizontal position to a vertical position while the arch vertical lifting of arch 2, corner G experienced an increase
was supported by the lifting frames. The third stage, arch in compressive stresses, and the decrease in compressive
setting, represents the release of the arch from the lifting stresses at corner C is also short lived. In other words,
assembly and its installation on temporary supports. vertical lifting results in an increase in total internal com-
pressive force at the midspan, both in the rib and in the tie.
Figure 10 shows typical stress changes at the midspan of This observation might be due to the interaction between
arch 2 during the rotation operation. During vertical lifting, the steel formwork and the arch. During vertical lifting, the
the response of the rib and the tie was governed by bending arch is released from the bottom formwork, and therefore
between the lifting frames. However, the magnitude of the friction between the arch and the formwork is elimi-
stress changes during vertical lifting was generally small. nated. As a result, local redistribution of the prestressing
Corners A and E (Fig. 10), which were located at the top of force could occur in the arches.
the rib and the tie before rotation, experienced an increase
in compressive stresses during vertical lifting. Similarly, The stress changes during supported rotation were gradual
a decrease in compressive stresses is expected for corners and relatively small. Although the arch was loaded by

Radial pressure from the tendon

Tendon Tendon
anchorage force anchorage force

Figure 8. Forces acting on the arch rib due to posttensioning.

40 S ept e mb e r – Oc t o b e r 2 0 1 5 | PCI Journal

 Vertical lifting Supported rotation Setting on supports

Figure 9. The three stages of the rotation operation.

its self-weight during this stage, it was supported by six stresses during supported rotation was more critical for
lifting frames, which were distributed along the length of ensuring the safety of the arches.
the arch, and therefore, the arch action was not mobilized
during supported rotation. Most of the observed changes To calculate the stresses during rotation, the construc-
in this stage were caused by the biaxial bending of rib and tion engineering team developed models of the arches in
tie elements as they behaved similarly to continuous beams which the rib and the tie were modeled using tapered beam
during supported rotation. elements while shell elements were used to model the
knuckle region. To obtain more realistic results, the models
The stress changes during setting of the arch on temporary also included all lifting frames, wire ropes, and equalizer
supports were quick and relatively large. During this stage, beams that were used for lifting and rotating the arches. To
the arch action was fully mobilized. Therefore, this stage consider the dynamic effects, an amplification factor of 1.5
was associated with a significant increase in compressive was also applied to all dead loads.
stresses in the rib and a significant decrease in those in the
tie. Although these stress changes were relatively large, the Figure 11 shows maximum and minimum stresses ob-
arches were designed to withstand much larger demands served in the arches during supported rotation. None of
due to bridge deck and live loads in the completed struc- the arches experienced tension under this operation, and
ture. As a result, the arches easily tolerated these stresses therefore the arches remained crack-free during rotation.
in the vertical orientation. Consequently, monitoring the The measured stresses were in reasonable agreement with

Figure 10. Stress changes at midspan in the rib and the tie during rotation. Note: In each graph, the × shows the corner for which the stress is plotted. 1 ksi = 6.895
MPa.

PCI Journal | S e p t e m b e r – O c t o b e r 2015 41

 Figure 11. Maximum and minimum corner stresses during the supported rotation of the arches. Note: The black lines show the range of observed stresses in differ-
ent arches. 1 ksi = 6.895 MPa.

the calculations by the construction engineering team, and than design predictions but below 50% of the compressive
where a difference exists between measured and predicted strength of concrete.
stresses, the predicted response generally overestimated the
risk of cracking. The observed variability was not surprising. Although the
upward jacking operation was conducted consistently, each
The rotation operation was conducted slowly, and VWG arch had a different set of unknown forces in the hand-
strain records and site observations gave no indication of tightened hangers before the operation. Moreover, because
considerable dynamic effects during rotation. There is a the arches are highly indeterminate, different locked-in
possibility that the 50% increase in the self-weight of the forces were expected in the hangers prior to upward
arch in the calculations has contributed to obtaining more jacking, depending on temperature and time-dependent
realistic stresses. However, additional detailed analysis is effects. As a result, upward jacking, which was designed to
needed to draw conclusions that could potentially be used change the forces in the hangers from an unknown initial
for other projects. state to a known final state, is expected to have variable
effects on different arches. However, upward jacking tends
Upward jacking to be more consistent than the sequential stressing of the
hangers, which is the method commonly used for steel
The upward jacking operation was the most critical stage network arches. In sequential stressing, the same factors of
for the knuckle region of the arches. The design team had variability will be present, in addition to more uncertainty
predicted that when the hydraulic rams were activated dur- regarding modeling assumptions and potential temperature
ing upward jacking, tension would have been induced at changes that might happen during the time-consuming
the top of the rib in the knuckle region if the arches had not stressing of the hangers.
been temporarily strengthened. Therefore, monitoring the
stresses during upward jacking was of critical importance Stresses at the end
in ensuring their safety against cracking. of each construction operation

Figure 12 shows measured stresses in the arches when Figure 13 shows the stresses at the end of the main con-
the hydraulic rams were activated, but the hanger nuts struction operations on the arches in the precasting yard.
were not retightened. The stresses in the knuckle region Design calculations were generally successful in predicting
were in agreement with the design predictions. However, the stresses in the structure during precasting. While vari-
as illustrated by the black lines, the variability of stresses ability was observed in the concrete stresses between dif-
among different arches was relatively large. While the ferent arches, design calculations could identify the major
average measured stress at the top of the rib in the knuckle parts of stress changes in the structure and provide safety
was compressive, some arches experienced tension in for the arches during handling. The largest differences
this region. These tensile stresses were well below the between measured and predicted stresses were observed
modulus of rupture of the concrete and therefore did not at the bottom of the tie, where design predictions under-
pose any risk of cracking to the arches. The stresses at the estimated the compressive stresses. However, because the
midspan of the arches in the tie were also highly variable stresses were below the linearity limit of the concrete, such
(Fig. 12). The stresses at the bottom of the tie were greater a difference should not be a source of concern.

42 S ept e mb e r – Oc t o b e r 2 0 1 5 | PCI Journal

Figure 12. Maximum arch stresses during upward jacking operations, before deactivating the hydraulic rams. Note: The black lines show the range of observed
stresses in different arches. 1 ksi = 6.895 MPa.

The total stresses in the finished bridge are highly influ- • The short-term construction stresses in the iden-
enced by the posttensioning and dead-load stresses. After tical arches could be highly variable from one
the bridge was opened to traffic, a static live-load test was arch to another, particularly after the arches were
conducted on one of the most heavily instrumented spans rotated into the vertical orientation. Uncertainties
of the bridge using four sand trucks, each weighing ap- due to the presence of hand-tightened hangers and
proximately 50 kip (220 kN). The largest live-load stress different temperature conditions among the arches
measured in the instrumented sections of the arches was may have contributed to such variability. In this
approximately 0.20 ksi (1.4 MPa),18 which is relatively project, the observed variability did not endanger
small compared with the construction stresses in Fig. 13. the safety of the arches. However, reliable stress
The details of the live-load test on the bridge will be dis- predictions for network arches must consider
cussed in a later paper. uncertainties due to unknown hanger conditions,
regardless of the sophistication of the model. A
Conclusion successful handling design would best be obtained
by assuming multiple conditions of hanger forces
This paper evaluated the short-term stresses during con- and making sure that the structure will not undergo
struction of the first precast concrete network arch bridge excessive stresses due to an unforeseen stiffness
in the world. The West Seventh Street Bridge was moni- distribution.
tored using embedded vibrating-wire gauges throughout
construction to develop an improved understanding of the • Before finalizing the design of other structures that
structure’s behavior and to ensure the safety of the arches. might be sensitive to cracking similar to these arches,
The primary conclusions of this paper are as follows: a material study is highly recommended. The modulus
of elasticity, creep, and shrinkage of the concrete will
• Design calculations were generally successful in cap- affect the stress. Therefore, these parameters must be
turing the essence of the structure’s response during realistically estimated before design. Although such a
posttensioning and handling. As a result, the arches study is often impractical in initial design calculations,
were successfully constructed without experiencing it is possible to analyze the model with the updated
tensile stresses that could pose a risk of cracking or parameters once the final mixture is determined to
compressive stresses larger than 50% of the compres- verify the design.
sive strength of the concrete. The observed differences
between average measured stresses and predicted To the authors’ knowledge, the study presented herein is
stresses could be in part due to assumptions made re- the first on the construction responses of a concrete tied
garding the modulus of elasticity of concrete in design arch bridge of any type. The data obtained in this study
calculations and the inherent complexities in convert- are a useful validation tool for future modeling of concrete
ing the strain measurements to stresses. arches.

PCI Journal | S e p t e m b e r – O c t o b e r 2015 43

 Figure 13. Stresses at the end of each construction operation. Note: The black lines show the range of observed stresses in different arches. 1 ksi = 6.895 MPa.

44 S ept e mb e r – Oc t o b e r 2 0 1 5 | PCI Journal

Acknowledgments in March 2014 after Lectures in 50+ Countries.” Ac-
cessed January 13, 2015. http://home.uia.no
The authors gratefully acknowledge the Texas Department /pert/index.php/The_Network_Arch.
of Transportation (TxDOT) for providing financial sup-
port for this study. Dean Van Landuyt and Courtney Holle, 9. Institute Stroyproekt JSC. 2014. “Ob River Cross-
the TxDOT engineers who designed the arches, provided ing—Bugrinsky Bridge in Novosibirsk.” Accessed
a great deal of guidance. In addition, there were numer- January 13, 2015. http://www.stpr.ru/eng
ous engineers and representatives from TxDOT and the /projects/1045/.
contractor, Sundt Construction, who provided significant
assistance throughout the investigation. Many thanks are 10. Janata, V., D. Gregor, L. Šašek, P. Nehasil, and T.
also due to Jose Gallardo, Ali Morovat, David Garber, Kos- Wangler. 2012. “New Troja Bridge in Prague–Struc-
tas Belivanis, Vasilis Samaras, and Hemal Patel, current tural Solution of Steel Parts.” Procedia Engineering
and former graduate students at the University of Texas at 40: 159–164.
Austin, for assisting with the instrumentation. The findings,
opinions, and recommendations presented in this article are 11. T. Zoli. 2012. “A Bridge by the People, for the Peo-
those of the authors and do not necessarily reflect the views ple.” Civil Engineering: The Magazine of the Ameri-
of TxDOT. can Society of Civil Engineers 82 (6): 48–57.

References 12. Larssen, R. M., and S. E. Jakobsen. 2011. “Taller,

Longer, Lighter: Meeting Growing Demand with
1. Salonga, J. 2010. “Innovative Systems for Arch Bridg- Limited Resources.” In Proceedings of 35th Annual
es Using Ultra High-Performance Fibre-Reinforced Symposium of IABSE/52nd Annual Symposium of
Concrete.” PhD thesis, University of Toronto, Toronto, IASS/6th International Conference on Space Struc-
ON, Canada. tures. London, September 2011. Switzerland: IABSE/
IASS
2. Mondorf, P. 2006. “Arch Bridge Construction.” In
Concrete Bridges. 600–638. New York, NY: Taylor 13. Gauthier, P., and L. Krontal. 2010. “Erfahrungen mit
and Francis. Netzwerkbogenbrücken im Eisenbahnbrückenbau”
[Experience with Network Arch Bridges in Railway
3. Tveit, P. 1987. “Considerations for Design of Network Bridge Construction]. Stahlbau 79 (3): 199–208.
Arches.” ASCE Journal of Structural Engineering 113
(10): 2189–2207. 14. Wollmann, G., and T. Zoli. 2008. “Bridge across Ohio
River and Blennerhassett Island.” Structural Engineer-
4. Tveit, P. 2014. “Systematic Thesis on Network ing International 18 (1): 28–30.
Arches.” Accessed January 13, 2015. http://home.uia
.no/pert/index.php/Systematic_Thesis. 15. Mato, F. M., M. O. Cornejo, and J. N. Sánchez. 2011.
“Design and Construction of Composite Tubular
5. Tveit, P. 2007. “An Introduction to the Optimal Net- Arches with Network Suspension System: Recent
work Arch.” Structural Engineering International 17 Undertakings and Trends.” Journal of Civil Engineer-
(2): 184–187. ing and Architecture 5 (3): 191–214.

6. Tveit, P. 2010. “Optimal Network Arches for Road 16. Smit, T. 2013. “Design and Construction of a Railway
and Rail Bridges.” In Proceedings of the 6th Interna- Arch Bridge with a Network Hanger Arrangement.”
tional Conference on Arch Bridges, College of Civil Master’s thesis, Delft University of Technology, Delft,
Engineering, Fuzhou University, Fuzhou, China. http:// Netherlands.
www.arch-bridges.com/paper2010/pdf
/35-Optimal%20network%20arches%20for%20Road 17. Van Landuyt, D., C. Holle, and J. Aparicio. “Precast
%20and%20rail%20bridges.pdf Network Arch Bridge.” In 2013 PCI Convention and
National Bridge Conference. Chicago, IL: PCI. CD-
7. Graβe, W., S. Teich, P. Tveit, and S. Wendelin. 2004. ROM.
“Network Arches for Road Bridges.” In Proceedings
of the 4th International Conference on Arch Bridges, 18. Yousefpour, H., T. Helwig, and O. Bayrak. 2014.
Barcelona, Spain. http://caminstech.upc.edu/sites “Structural Monitoring of the World’s First Precast
/default/files/Network%20Arches%20for%20Road Network Arch Bridge during Construction.” Center for
%20Bridges_0.pdf Transportation Research technical report 5-5253-03-1.
Austin, TX: University of Texas at Austin.
8. Tveit, P. 2014. “The Network Arch: Bits of Manuscript

PCI Journal | S e p t e m b e r – O c t o b e r 2015 45

 19. CEB (Comité Euro-International du Béton). 1993. 10. Farmington Hills, MI: ACI.
CEB-FIP Model Code 1990. Lausanne, Switzerland:
Thomas Telford. 26. ASTM International. 2011. Standard Practice for
Estimating Concrete Strength by the Maturity Method.
20. AASHTO (American Association of State Highway ASTM C1074. West Conshohocken, PA: ASTM
and Transportation Officials). 2012. AASHTO LRFD International.
Bridge Design Speciafications. 6th ed. Washington,
DC: AASHTO. 27. Gardner, N. J., and M. Lockman. 2001. “Design Pro-
visions for Drying Shrinkage and Creep of Normal-
21. Blok, J. 2012. “Stress Monitoring and Sweep Control Strength Concrete.” ACI Materials Journal 98 (2):
Studies for Innovative Prestressed Precast Arches.” 159–167.
Master’s thesis, The University of Texas at Austin,
Austin, TX.

22. Birrcher, D., R. Tuchscherer, M. Huizinga, O. Bayrak, Notation

S. Wood, and J. Jirsa. 2008. ‘Strength and Service-
ability Design of Reinforced Concrete Deep Beams.” Ec = modulus of elasticity of concrete
Center for Transportation Research technical report
0-5253-1. Austin, TX: University of Texas at Austin. Eci = modulus of elasticity of concrete at the time of the
ith construction operation
23. ASTM International. 2010. Standard Test Method for
Static Modulus of Elasticity and Poisson’s Ratio of f c' = compressive strength of concrete
Concrete in Compression. ASTM C469. West Con-
shohocken, PA: ASTM International. R = radius of curvature

24. ASTM International. 2012. Standard Test Method for t = concrete age after casting
Compressive Strength of Cylindrical Concrete Speci-
mens. ASTM C39. West Conshohocken, PA: ASTM ∆εi = strain change in concrete due to the ith construction
International. operation

25. ACI (American Concrete Institute) Committee 363. σ = stress in concrete

2010. Report on High-Strength Concrete. ACI 363R-

46 S ept e mb e r – Oc t o b e r 2 0 1 5 | PCI Journal

About the authors in Fort Worth, Tex. This innovative bridge, com-
pleted in 2013 as a replacement for a century-old
Hossein Yousefpour is a PhD bridge, consists of 12 prestressed, precast concrete
candidate in structural engineering network arches. The bridge was instrumented with
at the University of Texas at 224 vibrating-wire gauges that were embedded in
Austin. the arches prior to concrete placement. The gauges
were monitored during posttensioning, handling, and
transport as well as deck construction. The instru-
mentation provided data on the stresses induced in
Todd A. Helwig, PhD, PE, is an the arches, which were used to ensure a safe environ-
associate professor in the Depart- ment throughout construction. The measurements
ment of Civil, Architectural, and also provided a means for evaluating the accuracy
Environmental Engineering at the of stress calculations that were made during design.
University of Texas at Austin. He The results obtained from this study provide a unique
holds the J. Neils Thompson insight into the behavior of concrete arches built by
Centennial Teaching Fellowship an accelerated construction method.
in Civil Engineering.
Keywords
Oguzhan Bayrak, PhD, PE, is a
professor in the Department of Accelerated bridge construction, arch, bridge, monitor-
Civil, Architectural, and Environ- ing, network arch, post tensioning.
mental Engineering at the
University of Texas at Austin. He Review policy
holds the Charles Elmer Rowe
Fellowship in the Cockrell School This paper was reviewed in accordance with the Precast/
of Engineering and serves as the director of the Phil Prestressed Concrete Institute’s peer-review process.
M. Ferguson Structural Engineering Laboratory.
Reader comments
Abstract
Please address reader comments to journal@pci.org or
This paper reports some of the major findings from Precast/Prestressed Concrete Institute, c/o PCI Journal,
the instrumentation of the West Seventh Street Bridge 200 W. Adams St., Suite 2100, Chicago, IL 60606. J

PCI Journal | S e p t e m b e r – O c t o b e r 2015 47