Innovative Infrastructures - Toward Human Urbanism 1

EVALUATION OF CRACKS OF HATHAWAY PRECAST CONCRETE

SEGMENTAL BOX GIRDER BRIDGE
Dongzhou HUANG, Ph. D., P.E. Bo HU, Ph. D., P.E.
Atkins North America MMM Group Limited
Tampa, Florida, USA Calgary, Alberta, Canada
dongzhou.huang@atkinsglobal.com HuB@mmm.ca

Summary
The Hathaway Bridge, an eastbound and westbound bridge pair, is located in Panama City, Florida, USA. Each of
the pair consists of a single box precast concrete segmental girder with a deck width of 24,4 m (80 ft). The lengths
of the eastbound and westbound bridges are 1031,43 m (3384 ft) and 1162,93 m (3815 ft) respectively. To date,
these are the largest single-cell precast segments without internal struts fabricated in the USA. Unfortunately,
during construction many web cracks developed in the external anchorage areas. The purpose of this investigation
is to identify the main causes of the cracks and to provide some design recommendations for large precast concrete
segmental box girders. First, a brief description of the bridge is given. Then, analytical models for the bridge, post-
tension forces, and construction forces are presented. The analytical results show that the external longitudinal post-
tension forces can cause a significant high tensile stress in the inside face of the web around the anchorages. The
information presented in this paper can assist bridge engineers in the design of concrete segmental box girder
bridges.

Keywords: concrete segmental bridge; box girder; cracking analysis; finite element method; bridge model;
prestressing force model; anchorage zone design.

1. Introduction

Due to initial cost and life cycle cost advantages, speed of construction, appealing aesthetics, and adaptability to
curved roadway alignment, segmental concrete bridge construction has become a preferred bridge alternative for
major transportation projects worldwide. The Hathaway Bridge, completed 2004, is a pair of concrete segmental
single box girder bridges each consisting of 544 variable-depth segments weighing from 140 to 182 metric tons with
deck widths of 24,4 m (80 ft). These are the largest single-cell precast segments without internal struts fabricated in
the USA to date. During construction, many web cracks developed in the external anchorage areas.

Concrete cracking has always been an intriguing design and research topic of bridge engineering: Marshall and
Mattock (1962) performed experimental research on cracks in precast prestressed concrete I-girders. Huang and
Shahawy (2005) investigated the end zone cracks in precast prestressed U-beams. Huang et al (2011) investigated
the horizontal cracks in spliced I-girder bridges. However, little has been written on the cracking analysis of
segmental box girder bridges. The purposes of this investigation are to identify the main causes of cracking in
precast concrete segmental box girders and to propose a practical design method for limiting such cracking. First, a
brief description of the bridge is given. Then, analytical models for the bridge, post-tension forces, and construction
forces are presented. Finally, the analytical results and some recommendations are given.

2. Description of bridge

The Hathaway Bridge (Fig. 1) in Panama City, Florida, USA, consists of an eastbound and a westbound pair with 12
and 14 continuous spans respectively having a maximum span length of 100,58m (330 ft). The cumulative span
lengths of the eastbound and westbound bridges are 1031,43 m (3384 ft) and 1162,93 m (3815 ft) individually. The
elevation view of the Eastbound Bridge is shown in Fig. 2. Each of the eastbound and westbound bridges consists
of a single box precast concrete segmental girder with a total deck width of 24,4 m (80 ft), accommodating 4 travel
2 18TH CONGRESS OF IABSE, SEOUL, 2012

lanes of 14,63 m, 2 shoulders of 6,10 m and one sidewalk of 2,44 m. The girder depth varies from 5,49 m (18 ft) at
interior pier to 3,05 m (10 ft) at mid-span. Web thickness is 406,4 mm, top flange thickness is 228,6 mm, and the
bottom slab thickness varies from 533,4 mm at piers to 228,6 mm at mid-span. The design concrete strength is
41,37 Mpa (6.0 ksi). The principal dimensions can be found in Fig.2, Fig. 3, and Table 1.

Fig.1. Hathaway Bridge

Fig. 2. Elevation of Eastbound Bridge

(a) (b)

(b)
Fig. 3. Typical Section and Elevation, (a) Typical Section, (b) Section A-A, (c) Typical Cantilever Section

The bridge was built by cantilever construction method. The cantilever tendons layout for the typical span of 100,
58 (330 ft) long is shown in Fig. 4. There are 28 tendons in top flange which are symmetrically located about the
girder section centerline and designated as A to P respectively. Each of the tendons A through F and H have 24-
15,2 mm (0.6 in.) diameter strands stressed to 5417,7 kN (1218 kips). Each of the remaining tendons has 24 -15,2
mm diameter strands stressed to 4817,2kN (1083 kips). All of the tendons are anchored close to the web and 0,343
m from the top of the box segment. In Fig.4, a typical segment 8 in Span 9 is designated by four letters or numbers
which represent the Eastbound, span number, up or down referenced to the pier, and the segment number,
respectively.
 Innovative Infrastructures - Toward Human Urbanism 3

Table 1. Variable Dimensions of Box Section (mm)

Dim. Node
0 1 2 3 4 5 6 7 8 9 10 11 12
A 5486 5055 4669 4324 4023 3762 3543 3364 3226 3127 3067 3048 3048
B 533 486 443 403 367 332 302 275 249 229 229 229 229
C 4224 4440 4632 4805 4956 5086 5196 5285 5355 5404 5434 5442 5442
D 2296 2488 2659 2811 2945 3079 3151 3227 3285 3323 3353 3362 3362

(a)

(b)
Fig. 4. Layout of Top Tendons, (a) Elevation, (b) Tendon Layout

The layout of the bottom slab tendons is illustrated in Fig. 5. There are 16 tendons which are symmetrically located
about the centerline of the box girder and designated as A to H respectively. Each of the tendons A to F was
stressed to 5213,1 kN (1172 kips) and each of the Tendons G and H was stressed to 4585,9 kN (1031 kips). The
anchorage of the tendons is located 0,64 m from the end of the segment (match face) and 0,49 m from the top of the
bottom slab with a inclined angle of 16,50. The anchorage details of the segment 8 are shown in Fig. 6 and the
anchorage details for the other segments are similar.

(a)

(b)

Fig. 5 Layout of Bottom Tendons, (a) Elevation, (b) Tendon Layout

(a) (b)
Fig. 6. Bottom Tendon Anchorage, (a) End View, (b) Elevation
4 18TH CONGRESS OF IABSE, SEOUL, 2012

To apply a minimum compression to the match-cast joints during the “open time” of the epoxy, a total
of eight 28,56 mm (1 1/8”) diameter erection post-tensioning bars were used for each of the segments.
Each of the bars was stressed to 660,5 kN (135 kips). There are two bars in the cantilever slab, 2 bars
at each of the top corners of the box girder and 1 bar at each of the bottom corners. The locations of
the erection post-tensioning bars and the web reinforcement can be found in Fig. 7.

Fig. 7. Web Reinforcement and Location of Temporary Erection Bars.

There are 544 variable-depth segments, ranging from 140 to 182 metric tons which were barged to the site and
erected in balanced cantilever by an overhead launching truss with two winches for a faster and more balanced
erection. A critical construction loading case for the 100,6 m (330 ft) span erection are shown in Fig.8.

Fig. 8. Critical Erection Loading Case for Typical Spans of 100,6 m (330 ft)

Many cracks developed in the web around the temporary post tensioning blocks and bottom post-tensioning
anchor blocks. Most of the cracks are in the Segments 5 to 12. The crack width of most of the cracks are less
than 0,51 mm (0.002 in.). Fig. 9 shows the crack patterns in Segments 7 to 10. The remaining segments have
similar crack patterns. Most of the cracks around the temporary post-tensioning blocks are nearly horizontal
and those around the bottom tendon anchor blocks are inclined at about 300 to 350 from the horizontal.

Fig. 9. Typical Crack Patterns

3. Theoretical models

3.1 Bridge model

Based on the bridge construction procedures, two basic bridge theoretical models with two different loading
stages were developed. The first model is a cantilever model (see Fig. 10a) which simulates the first loading
stage. The first loading stage consists of the top cantilever tendons, top and bottom temporary post-tensioning
bars and the girder self-weight. The second model is a continuous beam model (see Fig. 10b) which simulates
 Innovative Infrastructures - Toward Human Urbanism 5

the second loading stage. The second loading stage includes bottom post-tensioning tendons, self-weight of the
closure pour, and construction loadings.

(a) (b)
Fig.10. Basic Analytical Model of the bridge, (a) Cantilever Girder, (b) Continuous Girder

The box girder bridge is simulated as a thin-wall structure and divided into a number of quadrilateral shell-plate
elements and analyzed by finite element method. There are 8 elements in each of the cantilever slab, 20
elements in each of the top and bottom slabs of the box, and 12 elements in each of the webs. There are 326
elements in the girder longitudinal direction for the cantilever bridge model. The width and length of each of
the elements are about 305 mm (12 in). The finite elements numbers total over 26000. The continuous bridge
model has 506 elements in bridge longitudinal direction. The total number of the finite elements for the
continuous beam model is over 40000.

It may not always be necessary to consider the stiffness effect of anchorage blisters on entire bridge structures,
however, for the local stress distribution analysis in anchorage zones, it is important to consider the stiffness
effect of the blisters. To simplify the analysis, a refined local stress analytical model was developed for
Segment 8 (see Fig. 11). To eliminate the effect of support conditions on the local stress and limit the number
of finite elements, the girder length is assumed to be 33,5 m (110 ft) long. The dimensions of the girder are the
same as those of Segment E9-D8. There are 38 elements in the top flange, 16 elements in bottom flange, 16
elements in each of the web, and 132 elements in girder longitudinal direction. The width and length of each of
the elements are about 191 x 254 mm (7.5x10 in). The finite elements total over 11000.

Fig. 11. Refined Local Stress Analytical Model

3.2 Post-tensioning force model

Post-tensioning forces are simulated as a number of concentrated forces and moments applied at the half depth
of the related elements (refer to Fig. 12). From Fig. 12, it can be seen that the top tendon is simulated as one
horizontal force and one vertical force applied at the anchorage location (PTT,VTT) and at the location where the
direction of the post tension force changes (PTTΔ,VTTΔ). The top temporary post tension bars are modeled as
three concentrated forces (PTBi) and three concentrated moments (MTBi) along girder vertical direction at their
anchorage locations. The bottom post tensions are simulated as five horizontal forces (PBTi), five vertical forces
(VBTi), and five concentrated moments (MBTi) applied at its anchorage location. For simplicity, only one
concentrated horizontal force (PBTΔ) and one vertical force (VBTΔ) are simulated at the location where the
direction of the post tensioning force is changed.

4. Analytical results

Take Segment E9-D8 for an example to illustrate the web tensile stress distribution under different loading cases
and stages: In Loading Stage I, the unit weight of the reinforced concrete is taken as 235,6 kN/m3 (0.15
kips/ft3). In Loading Stage II, the construction loadings are assumed as two concentrated forces applied at the
top of the webs. In both Loading Stage I and II, the post tensioning forces in Segment E9-D8 were removed and
considered in the analysis of the local stress in the refined local analytical model (Fig. 11). In analyzing the web
tensile stress distribution due to top tendons and temporary bars for Segment E9-D8, the refined local model
6 18TH CONGRESS OF IABSE, SEOUL, 2012

shown in Fig. 11 is assumed to be a cantilever beam supported at the center span of the beam. The top tendons
and bars are anchored in the end segment at the locations shown in Fig. 4 and 7. In analyzing the web tensile
stress distribution due to the bottom post tendons for Segment E9-D8, the refined local model is assumed to be
simply supported at the beam ends and the bottom tendons are assumed to be anchored at mid-span.

For an easy discussion of the tensile stress distributions, the finite element meshes of Segment E9-D8 with the
locations of the anchor blisters are illustrated in Fig. 13. In this figure, Hi represents the horizontal grid lines
and the Vi represents the vertical grid lines. Some of the key nodes are designated as circled numbers.

Fig. 12. Post Tensioning Force Model Fig.13. Web Meshes in Segment E9-D8

Fig.14 (a) illustrates the web shear stress contours due to top temporary bars and Fig 14 (b) shows the
vertical shear stress distributions in the anchorage zone on the inside face of the web along three typical
vertical sections. From this figure, we can see the temporary bars can cause a significant high shear stress. The
shear stress at Node 2 is over 6,9 Mpa (1.0 ksi). Fig. 15 (a) presents the vertical tensile stress contours due to
the bottom post tension tendons and Fig.15 (b) shows the vertical tensile/compression stress distributions on
both inside and outside faces of the web along three typical horizontal sections. From these figures, it can be
seen that the bottom tendons can induce high vertical tensile stress in the inside face of the web. The high
vertical tensile stress is due mostly to the web bending along the girder vertical direction.

(a) (b)
Fig. 14. Vertical Shear Stress Distribution due to Top Post Tension Bars, (a) Contours, (b) Typical Shear Stress
Distributions.

(a) (b)
Fig. 15 Vertical Tensile Stress Distribution due to Bottom Post Tension Tendons, (a) Contours, (b) Typical
Vertical Tensile Stress Distributions (Unit=Mpa)
 Innovative Infrastructures - Toward Human Urbanism 7

As the stress distribution status in the web around the anchorage zone is very complex, it may be easier to
discuss the maximum tensile stress distribution to identify the causes of the cracks. Fig. 16 (a) shows the
maximum tensile stress distribution due to top temporary post-tensioning bars. From this figure, we can see that
the post tension bars can cause very high tensile stresses around the anchor blisters. From the tensile stress
contours shown in Fig. 16 (b), it can be observed that the maximum tensile stress around the observed crack area
is over 3,8 Mpa (550 psi). The maximum tensile stress at Node 2 is about 3,0 Mpa (433 psi) and the direction of
the tensile stress is about 550 with the longitudinal axis. However, just the tensile stress due to the top post
tension bars alone could not cause the observed cracks.

(a) (b)

Fig. 16. Maximum Tensile Stress Distribution due to Top Temporary Post Tension Bars, (a) Tensile Stress
Distribution in the Webs, (b) Tensile Stress Contours

The maximum tensile stress distribution in the inside web due to the bottom post tensioning tendons are shown
in Fig. 17. Significant high tensile stresses in the webs around anchorage zones can be observed. The
maximum tensile stress at Nodes 3 to 7 are over 7,6 Mpa (1.1 ksi). The maximum tensile stress at node 6 is over
13,8 Mpa (2.0 ksi) and its stress direction is about 330 from the longitudinal axis which well matches the crack
patterns observed.

(a) (b)

Fig.17. Maximum Tensile Stress Distribution due to Bottom Post Tension Tendons, (a) Tensile Stress
Distribution in the Webs, (b) Tensile Stress Contours

The bottom post tension tendons also cause high tensile stress around the top temporary anchorage zone. The
maximum tensile stress at Node 2 is about 2,9 Mpa (420 psi) and tensile stress direction is about -720 from the
longitudinal axis. Under the combined action of both top post tension bars and the bottom post tension tendons
the vertical tensile stress near the top bar anchorage zone in the web at Note 2 is about 5,2 Mpa (754 psi) which
is a main factor causing the nearly horizontal cracks.

The maximum tensile stresses in both inside and outside faces at some typical nodes in Loading Stages I and II
are summarized in Table 2. The maximum effect on the vertical tensile stress in Segment E9-D8 in Loading
Stage II was determined based on a critical loading case shown in Fig. 8. From Table 2, we can see that the top
tendons will cause web bending. However, the effect of the top tendons on web tensile stress is comparatively
small. From this table, it also can be observed that the effect of Loading Stages I and II without the local post-
tension tendons on the web tensile stress is comparatively small and that the high web tensile stresses are mainly
8 18TH CONGRESS OF IABSE, SEOUL, 2012

caused by web bending due to bottom tendons. Properly increasing the web thickness can be expected to
effectively reduce the web tensile stresses.

Table 2. Web Maximum Tensile Stresses in Segment E9-D8

Model or Loading Stage Location

Node 2 Node 5
Stress (MN/m2 /psi) Angle Stress (psi) Angle
Inside Outside (Degree) Inside Outside (Degree)
Web Web
Local Top Bar 433 316 550 161 -48 -690
Top Tendons 38 -43 750 273 46 650
Bottom Tendons 420 -59 -720 2198 -52 570
Stage I -20 104 -720 56 142 -680
Stage II 148 66 730 15 102 880

5. Conclusions and recommendations

This paper presents analytical models for analyzing the local and general stress distributions for Hathaway
Concrete Segmental Bridge under different loading stages. Analytical results well match the crack patterns
observed. It was found that the cracks observed are caused by local stress due to the post tension bars and
tendons. The main cause for the cracks near the top temporary post tension bars is the combined action of the
top post tension bars and bottom post tendons. The resulting vertical tensile stress due to this combined action is
over 5,17 Mpa (750 psi) which is the principle reason for the horizontal cracks. The bottom tendons can cause a
significant high tensile stress in the web around the anchorage zone. The maximum tensile stress around the
cracked area ranges from 7,6 Mpa to 15,2 Mpa (1.1 to 2.2 ksi) which is the main factor contributing to the
observed cracks. The post tension bars and tendons can induce a significant bending moment in the web. An
effective way reduce cracking is to increase the vertical reinforcement in the web inside face and to properly
increase the web thickness in such large box segments.

6. Acknowledgements

The authors would like to express their sincere appreciation to Mr. John Previte for his valuable comments and
support received during the research. The opinions and conclusions expressed in this paper are those of the author
and are not necessarily those of the Florida Department of Transportation.

7. References

1. HUANG, D.Z. and SHAHAWY, M, “Analysis of Tensile Stresses in Transfer Zone of Prestressed Concrete U-
Beams”, TRB, Journal of the Transportation Research Board, No. 1928. Transportation Research Board of the
National Academies. Washington. D.C., pp. 134-141, 2005.
2. HUANG, D. Z., ARNOLD, S., and HU, B., ”Cracking Analysis of Florida Barge Canal Spliced I-girder Bridge”,
28th International Bridge Conference, Pittsburg, USA, June 5-8, 2011.
3. The New Hathaway Bridge, Panama City, Florida – Segmental Concrete bridge Project Designed Under
AASHTO-LRFD Specification, TRB, Journal of the Transportation Research Board, CD 11-S. Transportation
Research Board of the National Academies. Washington. D.C., 2005, pp. 109-112.
4. AASHTO Load and Resistance Factor Design (LRFD) Bridge Design Specifications, 5th Edition, AASHTO,
Washington, D.C., 2010.
5. MARSHALL, W. T., and MATTOCK, A. H., “Control of Horizontal Cracking in the Ends of Prestressed
Concrete Girders”, Journal of the Prestressed Concrete Institute, Vol. 7, 1962, pp. 56-74 (1962)
6. GILBERT, R.I., “Shrinkage, Cracking and Deflection – The Serviceability of Concrete Structures”, Electronic
Journal of Structural Engineering, EJSE International, Vol.1., No.1, pp 2-14 (2001).