Analytical and Experimental Studies on Torsion Behavior of

Hybrid Truss Bridge Girder with Various Connection Joints

Ji-Hun Choi, Kwang-Hoe Jung, Tae-Kyun Kim, Jang-Ho Jay Kim
Journal of Advanced Concrete Technology, volume 12 ( 2014 ), pp. 478-495

Inelastic Performance of High-Strength Concrete Bridge Columns under Earthquake

DaiJeong Seong, TaeHoon Kim, MyungSeok Oh, HyunMock Shin
Journal of Advanced Concrete Technology, volume 9 ( 2011 ), pp. 205-220

Development and Evaluation of New Connection Systems for Hybrid Truss Bridges
Kwang-Hoe Jung , Jang-Ho Jay Kim, Jong-Wo Yi , Sang-Hyu Lee
Journal of Advanced Concrete Technology, volume 11 ( 2013 ), pp. 61-79
Journal of Advanced Concrete Technology Vol. 12, 478-495, November 2014 / Copyright © 2014 Japan Concrete Institute 478

Scientific paper

Analytical and Experimental Studies on Torsional Behavior of Hybrid

Truss Bridge Girders with Various Connection Joints
Ji-Hun Choi1, Kwang-Hoe Jung2, Tae-Kyun Kim3 and Jang-Ho Jay Kim4*

Received 10 July 2014, accepted 5 November 2014 doi:10.3151/jact.12.478

Abstract
A hybrid truss bridge (HTB) is a bridge constructed by replacing the concrete webs of a pre-stressed box girder bridge
with steel truss webs. With its open web section, HTB offers advantages of high structural efficiency and excellent aes-
thetic appearance. The core technology of HTB is the connection system between the steel trusses and the concrete deck.
Since the open web section of a HTB decreases the torsional capacity, precise analysis of the torsional behavior of the
open web section with respect to the connection joint type is needed to ensure the safety of HTBs. In this study, the tor-
sional capacity of HTB was calculated using the torsion formula derived based on the steel plate thickness conversion
equation for corrugated steel web plates and the space truss theory. The torsion formula was verified by performing a
parametric study using a commercial finite element program. Next, the torsion analysis results of the HTB girder were
compared to experimental results to evaluate their accuracy. Finally, the simulation, experimental, and torsion formula
results were compared to evaluate the feasibility of using the formula in design.

1. Introduction 2002; Aoki et al. 2005; Fujiwara et al. 2005)

Hybrid truss bridge (HTB) refers to a bridge constructed The core technology of HTB is the connection system
by replacing the solid concrete web sections of a pre- between the steel truss member and the concrete deck.
stressed box girder bridge with open steel truss web In HTB, the joint types embedded in the top and bottom
sections. As shown in Fig. 1, replacement of concrete concrete decks can either be the same or different, de-
webs with relatively light steel truss members would pending on design requirements. An appropriate joint
reduce the self-weight of the super-structure by ap- type is selected based on the code requirements and the
proximately 20% to increase the span length and de- construction cost. As a result of the configuration of
crease the required sub-structure member size. Further- web trusses, stresses are concentrated at the joint with a
more, since the open web section has an excellent aes- complex stress distribution profile. Therefore, the joint
thetic appearance that can harmonizes the bridge with performance must be fully guaranteed to ensure the
the surrounding environment, HTBs are becoming safety of the HTB. The joint dictates not only the local
popular for medium-span bridges (e.g., span length of behavior but also the global behavior of the girder (e.g.,
40-60m) constructed in urban settings. (Minami et al. overall deflection and torsional rotation). Various re-
searchers and engineers have studied, developed, and
verified many different types of joint for HTB. (Miwa et
1
Doctoral Student, School of Civil and Environmental al. 1998; Tsujimura et al. 2002; Shim et al. 2007a, b)
Engineering, Yonsei University, Concrete Structural However, because of the open web section, the torsional
performance of a HTB is weaker than that of a pre-
Engineering Laboratory, Engineering Building #A, 134
stressed concrete (PSC) box girder bridge with solid
Shinchon-dong, Seodaemun-gu, Seoul 120-749, South
web sections. Because of uncertainty of the torsional
Korea.
2
Chief Research Engineer, Research and Development capacity, the open web section has rarely been used in
Division, Hyundai Engineering and Construction Co. eccentrically loaded or curved PSC box girder bridges.
Ltd., 102-4, Mabuk-dong, Giheung-gu, Yongin-si,
Gyeonggi-do, 446-716, South Korea.
3
Doctoral Student, School of Civil and Environmental
Engineering, Yonsei University, Concrete Structural
Engineering Laboratory, Engineering Building #A, 134
Shinchon-dong, Seodaemun-gu, Seoul 120-749, South
Korea.
4
Associate Professor, School of Civil and Environ-
mental Engineering, Yonsei University, Concrete Struc-
tural Engineering Laboratory, Engineering Building #A,
134 Shinchon-dong, Seodaemun-gu, Seoul 120-749,
South Korea.
*Corresponding author, E-mail: jjhkim@yonsei.ac.kr Fig. 1 Hybrid Truss Brige (HTB).
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 479

(Jung et al. 2007a, b, 2010a, b, 2011a, b, 2013a; Choi et centroid

al. 2011) In this study, space truss theory was used to
derive and analyze a torsional moment formula for a e=0 Shear
HTB girder. The commercial finite element (FE) pro- Force
gram MIDAS FEA was then used to simulate a torsion
test of HTB. Finally, the formula, simulation and ex-
perimental results of HTB girder under torsional loading
were compared to evaluate the accuracy of the formula.
Compression Tension
2. Load transfer mechanisms at the
connection joint (a) EHT: Embedded hinge type Hybrid Truss
Local
For this study, three different types of connection sys- Moment
centroid
tem, an embedded hinge type (EHT), continuous stud-
e
flange plate type (FHT), and discontinuous stud-gusset
Shear
plate type (GHT) were used in HTB girder specimens. Force
The load transfer mechanism of the three joint types is
Flange Plate
shown in Fig. 2. (Jung et al. 2013b) In EHT, the center
of the concrete deck is in alignment with the centerline
of the steel truss member such that the eccentric mo-
Compression Tension
ment would be minimized as shown in Fig. 2(a). In
FHT and GHT, shown in Fig. 2(b) and 2(c), respectively, (b) FHT: Flange plate type Hybrid Truss
the top or bottom truss ends are connected to a single Local Expected
continuous plate for all joints or a discontinuous gusset Moment failure line

plate for each joint, respectively. For both FHT and centroid

GHT, shear studs are welded on the surface in contact e

with concrete for shear connection and the center of the Shear
concrete deck does not coincide with the centerline of Force
the truss members. Therefore, the eccentricity between base plate

the two centerlines causes eccentric moment at the joint. gusset plate

HTB girder specimens 4 m in length were constructed

with the three joint types and tested in torsion to evalu-
ate and compare their torsional and load carrying ca-
pacities.
3. Derivation of analytical torsion formula
of HTB
3.1 Space truss theory
At the present time there are no specific analytical for-
mulas for calculation of torsional capacity of HTB.
Therefore, we formulated an analytical formula for tor-
sion by applying previous study results of a corrugated
steel plate web girder to the HTB girder. For torsional
behavior of the reinforced concrete (RC) girder, the
space truss theory is most commonly used by bridge
engineers. The space truss theory was first introduced
by Rausch and further developed by Lampert. (Rausch
1929; Lampert 1968) The design concepts of the space
truss theory are based on the assumption that steel re-
bars are embedded in the concrete. The girder cross-
section resists the torsion, because space truss members
run in the direction of crack development such that tor-
sion-resistant capacity is obtained from the concrete
surrounding the rebars. However, in the design of the
space truss theory, the torsional strength is usually over-
estimated compared to the actual torsional strength.
Therefore, Collins modified the conventional space
Compression Tension
(c) GHT: Gusset plate type Hybrid Truss
Fig. 2 Load transfer mechanism at the connection joint.
truss theory by making an assumption that the centerline
of the shear flow coincided with the centerline of the
concrete section, and expressed the nominal torsion
strength as a function of the area enclosed by the center-
line of the shear flow and the cross-sectional area of the
structure as shown in Equation 1: (Collins et al. 1980)
A0 At f y
Tn = 2 cot α (1)
s
where Tn is the nominal torsion; A0 is the area en-
closed by the centerline of the shear flow; At is the
cross section of the lateral steel rebar; f y is the yield
strength of lateral steel rebar; s is the stirrup spacing;
and α is the inclination angle.
3.2 Analytical torsion model of HTB girder
Mo et al. (2000) assumed that when torsion is applied to
a corrugated steel plate web girder bridge with top and
the bottom PSC decks, the area enclosing the shear flow
traveling through the centerline in the concrete deck A0
is determined by only using the concrete decks as
shown in Fig. 3. They proposed that the area of the
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 480

flange should be half of the entire area. (Mo et al. 2000,

2006) Based on this proposal, the torsion acting on the
top and bottom concrete decks is calculated using Equa-
tion 2:
1
A0 = (b − tb )(h − td ) (2)
2
where tb is the thickness of the corrugated web steel
plate and td is the effective thickness of the concrete
deck. It is assumed that the cross-section of the box
girder under torsion is a rigid body rotating along the
longitudinal axis of the girder and that the torsional an-
gle of the cross-section of the box girder under torsion is Fig. 3 Concrete box-girder bridge with corrugated steel
uniformly distributed on the cross-section. The torsion web sections.
resisted by the web can be calculated as a function of
the shear force acting on the web and the distance to the
corrugated steel plate web as shown in Equation 3:
Tw = τ w Aw (b − tb ) (3)

where Tw is the torsion acting on the web steel plate;

τ w is the shear stress; Aw is the web cross-sectional
area. The total torsion acting on the corrugated steel
plate web bridge is calculated as the sum of the torsion
acting on the top and bottom concrete decks Tn and the
torsion acting on the corrugated steel plate web Tw as
shown in Equation 4:
(a) Side view
T = Tn + Tw (4)

3.3 Web plate thickness conversion

The cross-section in a HTB girder is not constant, simi-
lar to that in a corrugated steel plate web girder. There-
fore, the formula for the torsional capacity of corrugated
steel plate web girder can be applied to the HTB girder
with minor modifications. In this study, the web truss
members are assumed to be arranged in series with con-
stant spacing and attached to the top and bottom con-
crete decks as shown in Fig. 4. If the cross-sectional
area for the torsion is calculated based on the assumed
configuration, the converted plate thickness of the web (b) Sectional view
truss member assumed as a concrete web can be calcu- Fig. 4 Calculated web plate thickness from the conver-
lated using Equation 5: (JPCEA 2005) sion.
⎛ ⎞
⎜ ⎟ crete ( Es / Ec ) ; Es is elastic modulus of the steel truss;
E ah
t* = c ⎜ 3 ⎟ (5) Asd is the cross-sectional area of the steel truss; and
Gc ⎜ d a3 ⎛ 1 1 ⎞⎟ Acu and Acl are the concrete area of the top and bot-
⎜ ⎟ ⎟⎟
⎜n A + 3 ⎜A +A tom connection joints of the HTB girder, respectively.
⎝ e sd ⎝ cu cl ⎠⎠
Equation 5 was derived for an enclosed cross-section in
where t * is the plate thickness calculated by converting which the torsional shear force is transported to the top
the steel truss member into a concrete web thickness and bottom concrete decks as a shear flow traveling
that is continuous in the longitudinal direction; Ec is through the truss members and connection joints. In this
the elastic modulus of concrete; Gc is the shear study, it was assumed that the converted thickness t * is
modulus of concrete; a is 1/2 of the horizontal dis- dependent on Acu and Acl , due to the difference in
tance between the connection joints of the steel truss shear flow transfer mechanism of various joint systems.
members; h is the height of the torsion resisting en- Therefore, t * would replace the corrugated steel web
closed cross-section; d is the length of the steel truss; thickness tb used in Equation 2.
ne is the ratio of the elastic modulus of steel and con-
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 481

3.4 Torsion formula evaluations 1.5mm.

To verify the effect of the three connection joints (EHT,
FHT, and GHT) on the torsional capacity of HTB girder,
the analytical formula was evaluated for the three dif-
ferent joints. The converted plate thicknesses of EHT,
FHT, and GHT calculated by the plate thickness conver-
sion formula of Equation 5 are shown in Table 1. The
calculation was performed using MathCAD Version
14.0. As stated before, the centerline of the concrete
deck coincides with the centerline of the steel truss in
EHT, whereas they do not coincide in FHT and GHT.
Therefore, when 1/2 of the horizontal distance from the
concrete deck to the steel truss connection joint (e.g.,
“ a ” in Fig. 4(a)) is taken as 450 mm, the truss angle
becomes 62° in EHT and 64.2° in FHT and GHT. Hence,
the distance of the steel truss (e.g., “ d ” in Fig. 4(a))
becomes 962mm in EHT and 1,034mm in FHT and
GHT. The effective thickness of the concrete deck td
is 150mm in both FHT and GHT, connection systems
without a haunch. For EHT, a connection system with a
haunch, the haunch area is divided by the member width
to yield a value of 40 mm, which is then added to the
deck thickness to obtain an effective thickness of 190
mm. On the other hand, the concrete areas of the top
and bottom connection joints ( Acu and Acl , respec-
tively), are 50,000 mm2 in EHT and 45,000 mm2 in FHT
and GHT. As a result, the converted thickness tb of the
steel truss calculated by Equation 5 is 13.0mm in EHT
and 11.5mm in FHT and GHT, showing a difference of
The torsion calculated from Equations 1 to 4 and the
corresponding torsional loads are shown in Table 2. The
effective cross-sectional area A0 corresponding to the
area enclosed by the centerline of the shear flow is
339,000 mm2 in EHT and 356,400 mm2 in FHT and
GHT; this difference is due to the difference in the ef-
fective thickness described above. Because of the dif-
ferences in the truss angle, the length and converted
thickness of the steel truss, and the existence of the
haunch, the calculated torsion is 1,361 kN·m in EHT
and 1,305 kN·m in FHT and GHT, showing a difference
of approximately 4.3%. The calculated ultimate load
considering the loading beam is 454 kN in EHT and 435
kN in FHT and GHT.
4. Torsion experiment of HTB girder
4.1 Details of torsion experiment
For this study, the torsion capacity of the HTB girders
with the three connection joints was verified by a tor-
sion experiment as shown in Fig. 2 and 5. Tables 3 and
4 show the specifications and the material properties,
respectively, of the test specimens. The three test speci-
mens were prepared to have a square cross-section with
a height and width of 1,000 mm and a length of 4,300
mm. The top and bottom concrete deck thickness was
150 mm, but a 100-mm thick concrete haunch was in-
stalled in EHT to provide sufficient concrete to fully

Table 1 Equivalent plate thickness.

Index
a Truss angle d td Acu Acl tb
(mm) (degree) (mm) (mm) (mm2) (mm2) (mm)
EHT 450 62.1 962 190 50,000 50,000 13.0
FHT 450 64.2 1,034 150 45,000 45,000 11.5
GHT 450 64.2 1,034 150 45,000 45,000 11.5

Table 2 Results of the torsion formula.

Index τ w (MPa) A0 (mm ) 2
Aw (mm2) Tn (kN․m) Tw (kN․m) T (kN․m) Pu (kN)
EHT 230.94 339,000 526,200 343.6 1,017 1,361 454
FHT 230.94 356,400 487,500 361.2 944.1 1,305 435
GHT 230.94 356,400 487,500 361.2 944.1 1,305 435

Table 3 Dimension of the specimens [unit : mm].

Index Width Height Length Slab thickness Haunch height
EHT 1,000 1,000 4,300 150 100
FHT 1,000 1,000 4,300 150 -
GHT 1,000 1,000 4,300 150 -

Table 4 Material properties in the simulation [unit : MPa].

Index Concrete Steel Truss / plate Rebar
Type OPC SS440 SD400
Allowable Stress - f a =140 f a =180
Strength f ck = 40 f y =240 f y =400
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 482

(a) Front and right side view(EHT)

(b) Front and right side view(FHT)

(c) Front and right side view(GHT)

Fig. 5 Design details of the test specimens.

embed the hinge joint. To fix the specimen to a strong
floor and apply torsional load, both ends of the speci-
men were extended with 800-mm concrete sections to
attach the necessary apparatus. Therefore, the total
length of the specimen was 4,300 mm with the truss
web included in a 2,700-mm region in the middle.
In FHT and GHT, the steel truss joints were attached
to a longitudinal flange plate and gusset plate, respec-
tively, and were integrated with the top and bottom con-
crete decks using the shear studs. The diameter and
thickness of the steel trusses were 114 mm and 6 mm,
respectively, for all specimens. In addition, the yield
strength of the EHT connection hinge plate, the FHT
flange plate, and the FHT gusset plate was 240 MPa,
equivalent to that of the steel truss. SS400 steel with
allowable strength of 140 MPa and thickness of 6 mm
was used for the truss members in all specimens. The
studs used for FHT and GHT had the same specifica-
tions as those used for general hybrid bridges with a
diameter of 19 mm and a height of 120 mm, and were
arranged at an interval of 150 mm. All of the longitudi-
nal steel reinforcing bars were SD400 high-strength
steel with a diameter of 13 mm and yield strength of
400 MPa.
4.2 Torsion test setup
The manufacturing process of the test specimens is
shown in Fig. 6. The steel pipes and plates were assem-
bled in the factory by fillet welding according to the
design configurations. The studs were attached using a
stud gun. After placing the assembled trusses in a steel
mold at a predetermined position, the steel rebars at-
tached to strain gauges were added before concrete cast-
ing. After concrete casting, the specimens were cured
for 28 days to achieve a concrete compressive strength
of 40 MPa. When the specimen was ready for testing,
the right end of the specimen was tightly fixed onto the
strong floor using four anchor bolts as shown in Fig. 7.
A steel pipe was attached to the left end in a longitudi-
nal direction for use as a rotational axis. In addition, a 3-
m loading beam attached to an actuator with 2,000 kN
capacity was installed at the left end for torsion loading
by applying a vertical load.
4.3 Data measurement positions and loading
The location of installed strain and displacement gauges
are shown in Fig. 8. A Rosette gauge was installed on
the diagonal truss to measure the strains occurring on
the diagonal truss. A concrete Rosette gauge was also
installed on the top surface of the concrete deck to
measure the crack strain and the principle strain direc-
tion on the concrete surface. In addition, six displace-
ment gauges were installed at critical locations to accu-
rately measure the displacement and torsional angle
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 483

(a) Plate and Steel pipe (b) Studs

(c) Assembly (d) Strain Gauge

(e) Concrete pour (f) Test setup

Fig. 6 Manufacturing process of the test specimens.

Fig. 7 Torsion test setup.

from torsional loading. A displacement controlled load
was applied with a rate of 0.02 mm/sec. For initial set-
ting of the specimen a load of ± 50 kN was first applied
before adding incremental loading of 50 kN.
5. Analysis of torsion experiment results
5.1 Deflection and torsional rotation
The torsional moment-displacement curves drawn using
the displacement measurements from LVDT 2 and
LVDT 3, which were installed at the rotational end of
the specimens, are shown in Fig. 9. In all three speci-
mens, symmetrical displacement profiles and ultimate
failure due to a shear failure of the top deck were ob-
served. The torsional rigidity did not change until the
moment exceeded 900 kN ∙ m; at this point the dis-
placement of FHT and GHT significantly increased and
the rigidity decreased, whereas those of EHT only
slightly increased and decreased, respectively. The tor-
sional moment-rotation curves drawn using the relative
displacement of LVDT 2 and LVDT 3 are shown in Fig.
10. In all three specimens, a linearly increasing curve
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 484

Fig. 9 Torsional moment-displacement relations.

(a) Gauge location

(b) LVDT location Fig. 10 Moment-torsional rotation angle relations.

Fig. 8 Strain and displacement gauge locations.
was observed up to approximately 700 kN∙m at which
point the torsional rotation of FHT and GHT drastically
increased, whereas that of EHT increased at a constant
rate until decreasing at 900 kN∙m.
In Fig. 10, ⓔ, ⓕ, and ⓖ curves are torsional rota-
tion calculated by assuming the specimen to be an elas-
tic body. ⓔ curve shows the rotation calculated in EHT,
which behaves similarly to the rotation obtained from
the test up to 900 kN∙m. ⓕ and ⓖ curves are the rota-
tions calculated in FHT and GHT, respectively, where a
single curve represents both FHT and GHT since they
both have same cross-section. The measured rotation
was smaller than the calculated rotation up to approxi-
mately 700 kN ∙ m. However, beyond this point the
measured rotation increased drastically, exceeding the
calculated rotation at a load of approximately 900 kN∙m.
The calculated and measured torsional rotation of the
specimens for moment steps of 300, 600, 900, 1,050,
1,200, and 1,350 kN∙m are shown in Table 5. The cal-
culated rotation of EHT is smaller than that of FHT and
GHT, because the torsional rigidity of EHT is greater
than that of FHT and GHT. The measured rotations of
the three specimens at 300 and 600kN∙m were almost
the same, but beyond 600 kN∙m the measured rotation
of FHT and GHT was greater than that of EHT. In
summary, the elastic limit of EHT was approximately
600 kN ∙ m and that of FHT and GHT was approxi-
mately 750 kN∙m, indicating that the torsional rigidity
of EHT was greater than that of FHT and GHT. As
shown in Table 5, the torsional rotations were not sig-
nificantly different, but the torsional rigidities and fail-
ure behaviors were different depending on the connec-
tion system. FHT and GHT showed almost the same
torsional behavior, indicating that continuity of the lon-
gitudinal flange plate did not have a great effect on the
torsional behavior.

-3
Table 5 Torsional lotation results [unit : 10 radian].
Applied Moment ① Calculated Torsional Lotation ② Measured Torsional Lotation ② / ①
(kN) EHT FHT GHT EHT FHT GHT EHT FHT GHT
300 1.24 1.52 1.52 1.15 1.17 1.57 0.9 0.8 1.0
600 2.48 3.05 3.05 2.18 2.17 2.60 0.9 0.7 0.9
900 3.72 4.57 4.57 3.74 5.16 5.42 1.0 1.1 1.2
1,050 4.34 5.33 5.33 4.82 8.97 8.00 1.1 1.7 1.5
1,200 4.96 6.09 6.09 5.81 - - 1.2 - -
1,350 5.58 6.85 6.85 6.90 - - 1.2 - -
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 485

5.2 Stress acting on steel truss web

Strains of the trusses located nearest to the rotational
axis (L_T1 and L_T2) among the six pairs of trusses
installed on both sides of the specimens are shown in
Fig. 11. As shown in the figure, strains in the diagonal
trusses for FHT and GHT were almost the same, but the
strain of the trusses of EHT was approximately one-
third of that of FHT and GHT. We can postulate that the
truss strain of EHT was small, because the torsional
load was resisted mainly by the concrete decks in EHT,
whereas in FHT and GHT the stress delivered to the
concrete decks was directly transferred to the truss
through the flange and gusset plates, respectively.
The measured strains from the six pairs of trusses Fig. 11 Torsional moment-strain relations.
were converted to stresses and are shown in Table 6 to 8.
The truss stress of EHT showed a maximum stress of 1,074 kN∙m and 1,092 kN∙m, respectively, some of the
91.8 MPa from the maximum torsional moment of web trusses yielded at torsional stresses of -270 MPa
1,398 kN∙m as shown in Table 6; this is below the al- and -286 MPa, respectively, when the yield strength of
lowable stress limit of the SS400 steel of 140 MPa and the SS400 steel of 240 MPa was exceeded.
within the linear elasticity range. On the other hand, the
truss stresses of FHT and GHT showed that the torsional 5.3 Crack patterns
stresses exceeded the allowable stress limit of the SS400 The crack patterns of the top concrete deck from the
steel of 140 MPa at loading of 900 kN∙m as shown in torsional load are shown in Fig. 12. Vertical and hori-
Tables 7 and 8, respectively. In addition, when the zontal lines are drawn on the concrete surface at an in-
maximum torsional moments of FHT and GHT were terval of 100 mm as reference axes to clearly indicate

Table 6 Axial stress distribution of EHT.

Truss Axial Stress (MPa)
Applied Moment
L_T1 L_T2 L_T3 L_T4 L_T5 L_T6
(kN·m)
R_T1 R_T2 R_T3 R_T4 R_T5 R_T6
-17.1 27.7 -15.7 17.6 -21.8 20.1
300
-1.9 27.4 -15.8 30.0 -16.6 40.2
-37.0 46.4 -45.3 42.1 -54.0 44.7
600
-21.8 42.9 -32.9 40.2 -32.9 50.4
-45.5 59.3 -59.3 53.5 -70.9 70.0
900
-24.9 54.7 -55.1 56.6 -51.8 60.7
-45.9 62.6 -62.2 55.5 -74.7 77.7
1,050
-26.3 55.1 -61.6 60.7 -55.1 63.8
-48.7 64.6 -63.3 59.0 -74.9 83.0
1,200
-28.8 54.1 -66.9 60.9 -56.0 63.4
-51.2 66.7 -64.8 60.7 -78.3 89.3
1,350
-30.3 53.7 -70.5 63.0 -56.0 65.1
-51.9 67.5 -64.6 60.9 -78.9 91.8
1,398
-30.7 53.7 -72.1 63.6 -56.2 65.3

Table 7 Axial stress distribution of FHT.

Truss Axial Stress (MPa)
Applied Moment
L_T1 L_T2 L_T3 L_T4 L_T5 L_T6
(kN·m)
R_T1 R_T2 R_T3 R_T4 R_T5 R_T6
-51.6 53.5 -60.9 46.3 -57.6 49.5
300
-37.7 40.4 -47.3 36.1 -44.6 36.3
-99.7 103.7 -118.3 89.5 -113.2 98.0
600
-74.2 81.0 -93.1 76.5 -88.1 78.3
-149.4 146.9 -170.0 124.9 -162.5 152.6
900
-135.7 122.9 -165.4 119.4 -165.2 139.7
-171.9 161.5 -191.7 121.9 -168.0 186.9
1,050
-202.1 139.3 -230.9 138.4 -230.3 180.9
-175.5 164.2 -194.5 116.8 -178.6 204.2
1,074
-222.8 139.3 -270.0 141.8 -268.2 193.8
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 486

Table 8 Axial stress distribution of GHT,

Truss Axial Stress (MPa)

Applied Moment
(kN·m) L_T1 L_T2 L_T3 L_T4 L_T5 L_T6
R_T1 R_T2 R_T3 R_T4 R_T5 R_T6
-56.7 69.8 -72.6 73.6 -81.5 58.2
300
-37.1 60.9 -61.6 55.3 -60.7 42.9
-103.3 125.1 -134.4 134.4 -147.1 101.2
600
-70.0 110.0 -115.9 103.0 -113.8 80.0
-170.8 163.8 -212.5 201.3 -251.9 151.5
900
-123.3 156.3 -204.3 154.8 -215.3 126.0
-208.0 184.3 -242.1 230.0 -300.9 176.3
1,050
-154.2 181.3 -243.5 180.3 -265.3 157.9
-223.1 185.2 -242.7 236.2 -314.5 179.5
1,092
-167.9 184.5 -258.6 182.4 -286.0 165.4

(a) EHT

(b) FHT

(c) GHT
Fig. 12 Crack patterns.

the crack pattern. As shown in Fig. 12, the cracks propa-
gated toward the top right corner at approximately 45°
in all three specimens, representing a typical torsional
shear failure mode and crack pattern. In FHT and GHT
specimens, cracks were formed at approximately 60 kN
and then rapidly propagated at a load of 250 kN. How-
ever, in EHT the cracks were formed at approximately
90 kN and then propagated in a stable manner until the
maximum load was reached.
6. HTB girder torsion simulation
A 3-D nonlinear FE simulation was performed for the
torsional experiment to validate the analytical formula
derived using the space truss theory and converted plate
thickness. A commercial nonlinear FE analysis software
Midas FEA was used for the simulation. Nonlinear con-
stitutive models were used in the simulation. (Midas
2008)
6.1 Concrete and steel reinforcing bar model
To properly represent the nonlinear constitutive behav-
ior of steel, an appropriate plastic model is required. In
this study, the von Mises plastic model was used. To
implement the isotropic stress effect in the simulation, a
proper compressive stress-strain relation must be used.
We used the Thorenfeldt hardening curve with softening
characteristic applicable to a compression region, as
shown in Fig. 13 and Equation 6: (Thorenfeldt et al.
1987)
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 487

⎛ ⎞
⎜ ⎟
αi ⎜ n ⎟
(6)
f = fp ⎜ ⎟
αp ⎜ ⎛ αi ⎞
nk
⎟
⎜⎜ n − 1 + ⎜⎜ ⎟⎟ ⎟⎟
⎝ ⎝αp ⎠ ⎠

⎧1 if 0 > α i > α p
f cc ⎪
where n = 0.80 + , k=⎨ f cc
17 ⎪⎩0.67 + 62 if α i ≤ α p

where α p is strain at peak stress (negative value) and

f p is peak stress (positive value). Fig. 13 Thorenfeldt compression curve.
For the tensile model, a brittle model with a cap tensile
strength was used as shown in Fig. 14. The fracture en-
ergy as a function of peak strain is shown in Equation 7:
1
Gf = ft ε nnpeak h (7)
2
where G f is fracture energy; ε nnpeak is the peak strain;
h is crack band width; ft is tensile strength.

6.2 Torsion simulation details

The specimen and test details of the simulation were as
same as those for the torsion experiment. Eight-node
and six-node cubic elements were used for the concrete Fig. 14 Brittle tension curve.
and web truss, respectively, as shown in Fig. 15. The
steel rebar was modeled using a bar element as shown in
Fig. 16. The material properties obtained from the tor-
sion test were used in the simulation. The elastic
modulus of concrete calculated from a prediction equa-
tion was used. The material properties used in the simu-
lation are provided in Table 9. The design strength of
the concrete was 40 MPa. SS400 steel with allowable
and yield strength of 140 and 240 MPa, respectively,
was used for the steel truss and plates. SD400 steel re- ① 8 node element ② 6 node element
bar with a diameter of 13 mm and yield strength of 240 Fig. 15 Solid element types.
MPa was used for the longitudinal and lateral rebar.

6.3 Load and boundary condition details

The load and boundary conditions used in the simula-
tion are shown in Fig. 17. A distributed load of equal
magnitude along the length of the loading grip was ap-
plied to the four edges around the rotational axis to ap-
ply torsional load. A load was applied to the specimen
with 100-kN increments until the specimen failed. The (a) Bar model (b) Bar element in solid element
top and bottom surfaces of the right end were fixed to Fig. 16 Bar element modeling.
prevent displacement or rotation.

Table 9 Material properties of the test specimens.

Index Concrete Steel Truss / Plate Rebar
Material OPC SS400 SD400
Strength [MPa] f ck = 40 f y =240 f y =400
Elastic Modulus [MPa] 31618 2.0×105 2.0×105
Weight Density [kN/m3] 24.5 76.9 76.9
Poisson's ratio 0.18 0.26 0.3
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 488

the maximum load was 2,800 and 2,600 kN for EHT

Fig. 17 Boundary conditions.
Fig. 18 Torsional moment-displacement relations.
7. Result analysis
7.1 Torsional load and behavior
In the experiment, torsional load was applied to the
specimen by eccentrically applying a vertical load using
a beam attached to the top surface of the specimen on
the loading end. In addition, a steel pipe was embedded
at the center of the specimen on the left end to measure
torsional rotation. In contrast, in the simulation, the
loads were directly applied to the 4 edges on the loading
end of the specimen. The simulation results showed that
and FHT/GHT, respectively, whereas the experimental
results for the maximum load were 466, 358, and 364
kN for EHT, FHT and GHT, respectively. Due to the
difference in the load modeling in the experiment and
simulation, it is necessary to compare load results with
respect to the applied torsion according to the load ap-
plication setup. If the vertical load from the experiment
is converted to the 4 edge loads of the simulation, the
maximum load becomes 467.18 and 433.17 kN for EHT
and FHT/GHT, respectively. The maximum load of
EHT from the simulation would be approximately 7.7%
larger than that of FHT/GHT.
In the simulation, all three specimens showed linear
elastic torsion behavior until a load of 250 kN. However,
the torsional rotation change beyond the load of 250 kN
was much more rapid in FHT/GHT than in EHT, show-
ing that EHT is better at resisting torsional load. The
torsional behaviors of the experimental and simulation
results are compared in Fig. 18. The calculation results
of the analytical formula based on space truss theory
and converted plate thickness are given in Table 10. The
analytical formula calculation of torsional load and
moment were 453.58 kN and 1,361 kN·m respectively
for EHT, and 435.12 kN and 1,305 kN·m for FHT/GHT.
As shown in Table 10, the torsional moment and the
maximum load results of “①/③” and “②/③”, respec-
tively, obtained from the analytical formula and the
simulation showed approximately 98% similarity. Both
the experimental and simulation results for EHT were
similar to the formula results. However, the experimen-
tal results of FHT/GHT showed approximately 80%
similarity to the formula and simulation results. This
difference in the similarity is due to differences in the
calculation of an effective cross-sectional area used to
resist torsion in the formula and simulation compared to
the experiment. These result comparisons showed that it
may be possible to apply the analytical formula to an
actual HTB girder analysis, but a sufficient safety factor
of 20% or higher should be implemented if the formula
is to be used for HTB girder design.

Table 10 Torsional moment and maximum load results.

Index EHT FHT GHT
① Calculated Moment 1,361kN•m 1,305kN•m 1,305kN•m
② Measured Moment 1,398kN•m 1,074kN•m 1,092kN•m
Torsional
Moment ③ Analyzed Moment 1,400kN•m 1,300kN•m 1,300kN•m
①/③ 0.97 1.00 1.00
②/③ 0.99 0.82 0.82
① Calculated Ultimate Load 453.58kN 435.12kN 435.12kN
Load ② Measured Max. Load 466kN 358kN 364kN
③ Analyzed Max. Load 467.18kN 433.17kN 433.17kN
① Measured Max. Displacement 17.39mm 29.74mm 18.8mm
Displacement
② Analyzed Max. Displacement 16.95mm 24.99mm 25.73mm
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014
(a) EHT
(c) GHT
Fig. 19 von Mises stresses of the trusses of the specimens.
7.2 von Mises stresses
The von Mises stresses of the trusses for the specimens
are compared in Fig. 19. The stress on EHT exceeded
the allowable stress of the SS400 steel of 140 MPa at
the moment of 800 kN·m, and reached the maximum
stress of 252.8 MPa at the moment of 1,401 kN·m. The
stress on FHT exceeded the allowable stress of the
SS400 steel at 800 kN·m and reached the maximum
stress at 264.3 MPa at the moment of 1,301 kN·m. The
stress on GHT exceeded the allowable stress of the
SS400 steel at 600 kN·m and reached the maximum
stress of 298.9 MPa at the moment of 1,301 kN·m. The
stresses of FHT and GHT were similar to the experi-
mental results. However, in the EHT experiment, the
stress was mainly resisted by the deck because of the
existence of the haunch. Since the boundary conditions
of the simulation assumed that the haunch and truss are
perfectly fixed, the stress increased because of efficient
stress transfer from the truss to the haunch. When the
stresses in all three experimental models exceeded the
allowable stress of the SS400 steel, the torsional stress
of EHT was larger than that of FHT and GHT at the
maximum load. The load was mainly resisted by the
concrete deck in EHT, whereas the continuous longitu-
dinal flange plate in FHT distributed the stress in the
deck to the web trusses. In contrast, in GHT the load
delivered to the deck was directly transferred to the
truss through a discontinuous gusset plate. The stress
489
(b) FHT
profiles obtained from the FE simulation results of the
three specimens are shown in Fig. 20.
7.3 Crack patterns
The crack patterns obtained from the simulation are
shown in Fig. 21 to 23. All three specimens showed a
typical torsional shear failure crack pattern. In EHT, the
initial crack was generated at a load of 116 kN at a loca-
tion between the bottom deck haunch and the embedded
connection joint, and then propagated toward the top
surface of the deck with an inclined angle of 45° to the
top right corner at a load of approximately 200 kN.
More cracks were generated as the torsional load in-
creased. In FHT and GHT, a crack was generated and
propagated in a similar manner as in EHT. The initial
crack appeared around the joint region at a load of 83
kN and then propagated toward the top surface of the
deck at approximately 150 kN; however, this was fol-
lowed by drastic crack propagation at a load of 183 kN
unlike the gradual crack propagation until the maximum
load was reached in EHT. In all three test specimens the
cracks were concentrated at the location between the
connection system and the deck, showing an effective
resistance to the torsional load. The crack pattern and
crack propagation stability showed that EHT has advan-
tages over FHT and GHT in terms of serviceability is-
sues.
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 490

(a) EHT

(b) FHT

(c) GHT
Fig. 20 von Mises stresses.
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 491

(a) 700kN

(b) 1,200kN

(c) 2,800kN
Fig. 21 Crack patterns of EHT.
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 492

(a) 600kN

(b) 1,200kN

(c) 2,600kN
Fig. 22 Crack patterns of FHT.
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014 493

(a) 600kN

(b) 1,100kN

(c) 2,600kN
Fig. 23 Crack patterns of GHT.
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014
Table 13 Cracking loads [unit : kN].
Index ① Calculated Cracking Load ② Measured Cracking Load
EHT 108
FHT 124
GHT 124
In Table 13, the cracking torsion formula calculation
of Equation 8 is compared to the simulation and ex-
perimental results (KCI 2007)
1 Acp2
Tcr = f ck
3 pco
where Acp denotes a cross-sectional area enclosed by
the shear flow of the torsional load and pco is a
circumferential length of the shear flow. In the experi-
ment, crack load was calculated from strains obtained
from the Rosette gauge installed on the concrete surface.
In EHT, experimental and calculated cracking loads
were the same and equal to approximately 116 kN.
However, in FHT and GHT, the experimental cracking
load was approximately 67% of the calculated cracking
load, indicating that FHT and GHT are more vulnerable
to cracking and serviceability problems than EHT. In
addition, the cracking load results showed that the HTB
girder with open web sections must be carefully studied
before its implementation in design. Comparison of the
simulation and formula results showed that EHT with a
haunch section resisted torsional loading better than
FHT/GHT without a haunch section. Therefore, based
on the comparison study we concluded that the hinge
embedded connection system of EHT is a more optimal
joint selection than FHT and GHT with respect to tor-
sional rigidity and usability.
8. Conclusions
In summary, the HTB torsion strength formula derived
from the space truss theory and corrugated steel plate
web thickness conversion showed good agreement with
the experimental and simulation results. Also, evalua-
tion of experimental and simulation verification of the
three different joint systems of EHT, FHT and GHT
showed that the haunch section of EHT conferred supe-
rior torsion resisting capacity compared to FHT/GHT
without a haunch section.
Comparison of the formula and simulation results
showed 99% similarity for all three specimens as shown
in Table 10. Additionally, comparison of the formula
and experimental results of EHT and FHT/GHT showed
a similarity of 97% and 80%, respectively. Based on
these study results, further detailed parametric studies of
the torsion formula are required for design usage.
Acknowledgements
This work was partially supported by the National Re-
search Foundation of Korea (NRF) grant funded by the
Korea government (MSIP) (No.2011-0030040). This
494
③ Analyzed Cracking Load ② / ① ③ / ①
79 116.7 0.73 1.08
52 83.3 0.42 0.67
56 83.3 0.42 0.67
work was also partially supported by R&D Policy Infra-
structure Technology Commercialization Project by the
Ministry of Land, Transport, and Maritime Affairs of the
Korea government. The authors wish to express their
gratitude for this financial support. The opinions, find-
(8)
ings, and conclusions of the paper are the authors’ and
do not necessarily reflect the views of the sponsors.
References
Aoki, K., Takatoku, Y., Notoya, H., Uehira, Y., Kato, T.
and Yamaguchi, T., (2005). “Design and construction
of Sarutagawa bridge and Tomoegawa bridge.”
Bridge and Foundation, 39(5), 2-11. (in Japanese)
Choi, J. H., Jung, K. H., Ha, J. H. and Kim, J. J. H.,
(2011). “A study on the torsional stiffness of the
hybrid truss bridge.” Proceedings Journal of the
Korea Concrete Institute, 23(1), 165-166.
Collins, M. P. and Mitchell, D., (1980). “Shear and
torsion design of prestressed and non-prestressed
concrete beams.” Journal of the Precast/Prestressed
Concrete Institute, 25(5), 32-100.
Fujiwara, H., Shoji, A., Sakata, H., Ushiroshoji, S.,
Kabaki, Y. and Noro, T., (2005). “Design and
construction of Shitsumi Ohashi bridge.” Bridge and
Foundation, 39(11), 2-11. (in Japanese)
JPCEA (Japan Prestressed Concrete Engineering
Association), (2005). “Design and construction
standards for composite bridges.” Gihodo Shuppan
Co. Ltd. (in Japanese).
Jung, K. H., Kim, K. S., Chung, C. H. and Shim, C. S.
(2007a). “An experimental study on the horizontal
shear strength of composite truss joint according to
the structural connection system.” Civil Expo 2007,
Korean Society of Civil Engineers, Daegu, 130-133.
Jung, K. H., Kim, K. S. and Chung, W. S., (2007b).
“Flexural behavior of prestressed concrete hybrid
girder with steel webs.” In: Proceedings of 4th
International Conference on The Conceptual Ap-
proach to Structural Design, Venice, Italy, 297-305.
Jung, K., Yi, J. and Kim, J. J. (2010a). “Structural safety
of the newly developed connection system of the
prestressed concrete hybrid girder with truss web
section.” In: Proceedings of 8th Short and Medium
Bridge Conference, CSCE, Niagara Falls, Canada,
210(1-8).
Jung, K., Yi, J. and Kim, J. J., (2010b). “Structural
safety and serviceability evaluations of prestressed
concrete hybrid bridge girders with corrugated or
steel truss web members.” Engineering Structures,
32(12), 3866-3878.
Jung, K. H., Lee, S. H., Choi, J. H., Choi, J. H. and Kim,
J-H. Choi, K-H. Jung, T-K. Kim and J-H. J. Kim / Journal of Advanced Concrete Technology Vol. 12, 478-495, 2014
J. J. H., (2011a). “Study on the torsional behavior of
the hybrid truss bridges according to the connection
systems.” Proceedings of the Korea Concrete Insti-
tute, 23(2), 79-80.
Jung, K. H., Yi, J. W., Lee, S. H. and Kim, J. J. H.,
(2011b). “Fatigue capacity evaluation of hinge type
connection system for a hybrid truss bridge.” Journal
of the Korea Concrete Institute, 23(3), 303-310.
Jung, K. H., Kim, J. J. H., Yi, J. W. and Lee, S. H.,
(2013a). “Development and evaluation of new con-
nection systems for hybrid truss bridges.” Journal of
Advanced Concrete Technology, 11(1), 61-79.
Jung, K. H., Yi, J. W., Lee, S. H., Choi, J. H. and Kim, J.
J. H., (2013b). “Torsional behavior of hybrid truss
bridge according to connection systems.” Journal of
the Korea Concrete Institute, 25(1), 63-72.
KCI (Korea Concrete Institute), (2007). “Structural
concrete design code.” Korea.
Lampert, P. and Thürlimann, B., (1968). “Torsion tests
of reinforced concrete Beams (Torsionversuche an
Stahlbetonba lken).” Report No. 6506-2, 101.
Midas Information Technology Co., Ltd., (2008).
“Midas User's Manual, Analysis & Algorithm.”
Korea.
Minami, H., Yamamura, M., Taira, Y. and Furuichi, K.,
(2002). “Design of the Kinokawa viaduct composite
truss bridge.” In: Proceedings of the 1st JSCE
Congress, Composite Structures, Osaka, Japan, 371-
380.
Miwa, H., Nagasawa, T., Yoda, T., Suzuki, T. and
Kumagai, Y., (1998). “Experimental study on the
mechanical behavior of panel Joints in PC hybrid
495
truss bridges.” Journal of Structural Engineering,
JSCE, 44A, 1475-1484.
Mo, Y. L., Jeng, C. H. and Chang, Y. S., (2000).
“Torsional behavior of prestressed concrete box-
girder bridges with corrugated steel webs.” ACI
Structural Journal, 97(6), 849-859.
Mo, Y. L. and Fan, Y. L. (2006), “Torsional design of
hybrid concrete box girders.” Journal of Bridge
Engineering, ASCE, 11(3), 329-339.
Rausch, E., (1929). “Design of reinforced concrete in
torsion (Berechnung des Eisen betons gegen
Verdrehung).” PhD thesis, Technische Hochschule,
Berlin, 53.
Shim, C. S., Park, J. S. and Kim, K. S., (2007a). “An
experimental study on joint structures ofcomposite
truss bridges.” Journal of Korean Society of Steel
Construction, KSSC, 19(3), 303-312.
Shim, C. S., Park, J. S., Chung, C. H. and Kim, K. S.,
(2007b). “Design and experiments on connection of
composite truss bridges.” In: Proceedings of 6th
International Conference Steel and Aluminium Struc-
tures, Oxford, 963-970.
Tsujimura, T., Shoji, A., Noro, T. and Muroi, S., (2002).
“Experimental study on a joint in prestressed concrete
bridge with steel truss web.” In: Proceedings of the
1st fib Congress, Osaka, Japan, Composite Structures,
347-352.
Thorenfeldt, E., Tomaszewicz, A. and Jenson, J. J.,
(1987). “Mechanical properties of high-strength
concrete and applications in design.” In: Proc. Symp.
Utilization of High-Strength Concrete, Tapir.