You are on page 1of 14

Journal of Constructional Steel Research 161 (2019) 355–368

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Finite element analysis of headed studs embedded in thin UHPC


Junhui Cao, Xudong Shao ⁎
College of Civil Engineering & key Laboratory for Wind and Bridge Engineering of Hunan Province, Hunan University, Changsha 410082, China

a r t i c l e i n f o a b s t r a c t

Article history: With an increasing use of ultra-high performance concrete (UHPC) in civil engineering, steel-concrete composite
Received 11 May 2018 structures continue to progress. However, studies focusing on headed studs embedded in UHPC are limited,
Received in revised form 20 September 2018 which hinders the design of steel-UHPC composite structures. This paper investigates the shear-resisting behav-
Accepted 12 March 2019
ior of headed studs embedded in thin UHPC by using the finite-element (FE) analysis technique. A FE model was
Available online xxxx
established for a push-out test, aiming to reflect the plastic behavior of the specimens and to evaluate the damage
Keywords:
process in the headed studs and UHPC. The analysis results were compared to test results. It was revealed that the
Headed stud load-slip curve, shear strength, and failure mode all agreed well with the experimental observations. A paramet-
Ultra-high performance concrete (UHPC) ric analysis was performed based on the validated FE model. The influence of stud diameter, stud height, and
Push-out test compressive strength of UHPC on the shear behavior of the headed studs was revealed. Further, the shear
Nonlinear FE model strengths obtained in the FE analysis were compared to the results predicted through theoretical equations.
Parametric analysis The comparison indicates that for headed studs embedded in UHPC, the contribution of the weld collar should
be considered to evaluate the shear strength.
© 2019 Elsevier Ltd. All rights reserved.

1. Introduction to 25 mm. Cao et al. [8] tested headed studs embedded in UHPC with
an even lower aspect ratio: 2.7. The test results implied that the shear
Headed stud is a connector commonly used in steel-concrete com- strength of the headed studs was not affected under the extremely
posite structures. Extensive studies have been accomplished to reveal low aspect ratio.
the behavior of headed studs embedded in both normal concrete and Experimental tests for composite structures are costly. Fortunately,
high performance concrete, including shear strength [1], shear stiffness this drawback can be compensated by using finite-element (FE) analy-
[2], fatigue life [3,4], etc. sis technique. A delicate FE model can simulate the damage process and
In recent years, composite structures continue to progress with the failure mode occurring in a test, and can be used to explore the behavior
application of ultra-high performance concrete (UHPC). UHPC is an in- of headed studs via parametric analyses. Nguyen et al. [9] developed an
novative cement-based material with excellent mechanical properties accurate FE model for the push-out test and the FE results agreed well
and durability [5]. Applications of steel-UHPC composite structures with the experimental results. Based on the validated FE model, a para-
have been found in some real construction projects in Canada, Germany, metric study was performed on stud diameter and concrete strength to
United States, China, etc. For example, Shao et al. [6] proposed a understand their influence on shear strength of headed studs. Luo et al.
new bridge deck — the steel-UHPC lightweight composite deck [10] constructed a FE model for headed studs embedded in UHPC. In the
(LWCD), aiming to reduce fatigue cracking risks in orthotropic steel analysis, the influence of stud spacing on shear strength was revealed
decks. In a typical LWCD, the UHPC layer is only 45–60 mm thick, and formulas considering the group stud effect were proposed. In addi-
which is connected with a steel bridge deck through Φ13 × 35 mm tion, FE models were developed in other studies to discuss the behavior
headed studs. Currently, this LWCD system has been used in 30 real of headed studs embedded in normal concrete [11,12].
bridges in China. However, the above analyses are mainly related to normal steel-con-
However, up to now, very few studies have been conducted and are crete composite structures, and few focuses on short headed studs in
available for shear connectors embedded in UHPC. Kim et al. [7] uncov- UHPC. The research work in this paper highlights the behavior of
ered the behavior of headed studs in UHPC through push-out tests. headed studs in LWCD, an innovative steel-UHPC composite deck. A del-
According to the test results, the shear strength in studs was not influ- icate 3D FE model was constructed to investigate the failure mode and
enced when the aspect ratio (i.e., the stud height-to-diameter ratio) shear strength of short headed studs embedded in UHPC. The FE
was reduced from 4 to 3.1 and the cover was decreased from 50 mm model was validated using experimental results on push-out speci-
mens. Based on the FE model, a parametric analysis was performed
⁎ Corresponding author. and the influences of stud diameter, stud height, and concrete strength
E-mail addresses: caojunhui@hnu.edu.cn (J. Cao), shaoxd@hnu.edu.cn (X. Shao). on shear strength of headed studs were discussed. Further, the shear

https://doi.org/10.1016/j.jcsr.2019.03.016
0143-974X/© 2019 Elsevier Ltd. All rights reserved.
356 J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368

strengths derived from the FE analysis were compared to the results


predicted by theoretical equations.

2. Push-out test for headed studs in UHPC

Behaviors of headed studs can be obtained using either push-out


specimens or steel-concrete composite beam specimens. Slutter and
Fisher [13] conducted a comprehensive comparison study between
the above two methods. It was found that the shear strength derived
from push-out specimens is generally lower than that from composite
beam specimens. This implies that push-out tests can result in more
conservative results. Thus, the test presented in this paper is a push-
out test.
Fig. 1 shows the configuration of the push-out specimens. Each spec-
imen consisted of an I-shaped steel part and two UHPC layers. The UHPC
layer was 50 mm thick and was embedded with Φ10 mm steel rebars, Fig. 2. Photographs of the test setup.
transversely and longitudinally. Four headed studs were welded to the
steel flanges at the steel-UHPC interface. The studs had a diameter (d)
of 13 mm and an overall height (h) of 35 mm, resulting in a low aspect and dynamic problems. FE models constructed in Abaqus Explicit are
ratio of h/d = 2.7. solved using the central difference method and hence iterations are
The specimen was tested by a compression testing machine with a not necessary for solving nonlinear problems. Consequently, Abaqus
load capacity of 2000 kN. The setup is shown in Fig. 2. The load was Explicit is a powerful tool to deal with complex nonlinear problems
exerted from the top of the specimen, and dial indicators were mounted such as contact, damage, failure, etc., and the computing cost can be sig-
to the sides to record relative slips at the steel-UHPC interfaces. A de- nificantly reduced. Thus, the current FE analysis was accomplished in
tailed description of the test setup is presented in reference [8]. Abaqus Explicit.
Three push-out specimens were fabricated and tested. The test All of the components in the push-out specimen were modelled, in-
results indicated that the failure modes of the three specimens were cluding the steel plates, the UHPC layer, the headed studs, and the steel
identical, i.e., the headed studs fractured from the roots, resulting in a meshes. Details of the FE analysis are presented in the following
separation between the UHPC layer and the steel part. While the subsections.
UHPC layer only developed marginal damage around the root of the
headed studs. These observations imply that although the headed 3.2. Element type
studs had an extremely low aspect ratio, their shear strengths could
be fully developed. The shear strength and load-slip response were cap- Only a quarter of the specimen was modelled in the FE analysis con-
tured for the short headed studs, and details can be referred to in [8]. sidering structural symmetry, as shown in Fig. 3. The steel plates, the
UHPC layer, and the headed studs were all simulated through C3D8R el-
3. Establishment of 3D FE model ements. An element C3D8R is a brick element with eight nodes and each
node has three translational degrees of freedom, which is commonly
3.1. Basic information used for solid structures. While the steel meshes were constructed via
T3D2 elements. An element T3D2 is a truss element possessing two
The FE analysis was performed in Abaqus, a commercial FE software nodes and each node has three translational degrees of freedom. The
[14]. Abaqus provides two optional modes, i.e., Abaqus Standard and FE model was meshed using elements with dimensions not N8 mm.
Abaqus Explicit, and the latter is more suitable for nonlinear quasi-static It should be addressed that the weld collar at the root of the headed
studs was also modelled. The reason is that for headed studs embedded
UHPC, the weld collar contributes to a significant portion of the shear
strength and should be taken into account in the analysis [10].

Fig. 1. Configuration of push-out specimen (unit: mm). Fig. 3. Finite element model of push-out specimen.
J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368 357

Fig. 4. Simulation of the weld slope in the comparative FE model.

In the analysis, a comparative study was performed to reveal the in- Specifically, two cases were considered. In one case, the cross sec-
fluence of the weld shape on the analysis results. Headed studs are gen- tion of the weld had a shape of rectangular, as shown in Fig. 3. While
erally welded using the electric arc stud welding. A ceramic ferrule is in the other case, a 1 mm × 1 mm slope was simulated at the top
positioned around the root of the headed stud, which serves as a corner of the weld, as shown in Fig. 4. It should be addressed that
mould for the molten metal to harden. Thus, the shape of the weld is extra efforts are required in the latter FE model to develop the circu-
identical to the inner shape of a ceramic ferrule. lar slope.

Fig. 5. Interaction between steel and UHPC.

Fig. 6. Stress-strain curves of steel plate and steel rebar.


358 J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368

3.4. Constitutional laws of materials

3.4.1. Steel plate and steel rebars


The push-out tests [8] showed that the steel plates were not
damaged. On the other hand, the UHPC layers only exhibited marginal
damage around the stud roots. This implies that the steel rebars embed-
ded in UHPC should also develop no damage. Consequently, the steel
plates and steel rebars were assumed as ideal elastic-to-plastic mate-
rials in the FE analysis.
The steel plates were fabricated from Q345, a steel grade with nom-
inal yield strength of 345 MPa, which is commonly used for steel bridges
in China. The steel rebars had a grade of HRB400, a type of steel bar with
nominal yield strength of 400 MPa. The constitutional laws of the two
materials are demonstrated in Fig. 6.

3.4.2. Headed stud


According to the test results [8], the headed studs broke from the
roots and consequently the UHPC layer was separated from the steel
plate. Thus, the headed studs experienced serious damage and failure
Fig. 7. Stress-strain curve of headed studs.
in the test, and these observations should be reflected in the FE analysis.
Stress-strain relationship models available for headed studs are rich.
While the most commonly used one is the tri-line stress-strain model
[9,10,15,16]. This kind of constitutional law can reflect not only the
yield stage but also the strengthening stage of a headed stud. Thus, a
tri-line stress-strain relationship was defined for the headed studs in
this study, as shown in Fig. 7. The yield strength (fy) and ultimate
3.3. Interaction and boundary conditions strength (fp) of were defined as 375 MPa and 450 MPa, respectively.
The values are not appreciably different to those reported by other
To properly consider the interaction effect, a surface-to-surface con- scholars [9,10,15,16].
tact was established at the steel-UHPC interface. The contact acts on two In Abaqus, the damage of a material is defined using the index D.
surfaces, i.e., the inner surface of the UHPC layer, and the outer surface of Material damage is a gradual process, which is accompanied with the
the steel part, including the steel plate and the headed studs welded development of tiny cracks and cavities. Eq. (1) provides a definition
(Fig. 5). of the damage index D. According to the definition, D = 0 corresponds
Contact properties were defined along both the normal and tangen- to no damage and D = 1 represents a full damage.
tial directions. Regarding the normal direction, a default method of
“hard contact” was defined. While for the tangential direction, the pen- Ed
alty function method was employed. According to PCI committee, the D ¼ 1− ð1Þ
E0
friction coefficient between steel and concrete is generally 0.4 [10].
Thus, this value was considered in the tangential direction. where, E0 = the initial elastic modulus of a material; Ed = the elastic
The boundary condition of the FE model is as follows. The nodes at modulus of a material with damage.
the bottom of the UHPC layer were restrained from all of the transla- In the FE model, the fracture strain of the headed studs was defined
tional degrees of freedom. as the plastic strain at the ultimate state, i.e., εp − εy = 0.00418. It
should be addressed that Abaqus could automatically delete elements
with a serious damage. The default threshold value is Dmax = 0.99. If
the damage index value D at any integration point of an element ex-
ceeds Dmax, the element will be removed from the FE model.

3.4.3. Weld
As mentioned earlier, headed studs are generally welded by using
electric arc stud welding. Theoretically, the strength of a weld should
be higher than that of the headed studs and steel plate. However, if
the welding parameters such as the lift height, electric current, welding
time are disturbed, the weld quality must be influenced. Thus, the weld
strength is not always higher than that of the headed studs and steel
plate. This statement is supported by the push-out test accomplished
by the authors [8]. According to the test results, there were two failure
patterns for the headed studs embedded in UHPC, i.e., the headed
studs failed from either the weld root or the stud shank. It was observed
that the two failure patterns could exist simultaneously in a push-out
specimen. Thus, the weld strength should be carefully defined to avoid
either an overestimation or an underestimation to the shear strength.
A comparative study was performed to evaluate the influence of the
weld strength on the analysis results. However, studies on the stress-
strain models for the weld of headed studs are currently unavailable.
According to studies by Kunihiko [17], if the strength of the weld is no
Fig. 8. Stress-strain curve for weld. b80% of the strength to a matrix plate, the welded structure is expected
J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368 359

Fig. 9. Stress-strain curves of UHPC.

to fail from the matrix plate rather than the weld. Thus, 80% was consid-
ered as a reference value to define the stress-strain curve for the weld.
Three cases were considered. In the basic case, the weld strength
was assumed to be equal to that of the headed studs (i.e., fweld =
fstud). While in the other two cases, the weld strength was defined to
be either higher or lower than that of the headed studs, i.e., fweld =
fstud/0.8 and fweld = fstud × 0.8. Thus, based on the stress-strain curve
for the headed studs, the constitutional law could be defined for the
weld, as shown in Fig. 8.

3.4.4. UHPC
According to the test results, the UHPC layer also developed damage,
although the damaged area was very locally. Thus, the concrete damage
plasticity (CDP) model was used for UHPC. CDP is commonly used to
simulate the damage and failure process in concrete structures.
For concrete, the stress-strain curve (or the damage index D) under
tension and compression must be defined independently. Yang and
Fang [18] conducted experimental tests for UHPC specimens under uni-
axial compression and proposed an equation to calculate the stress-

Fig. 10. Definition of damage indexes in UHPC.

Fig. 11. Positions to extract relative slip. Fig. 12. Influence of the weld strength on the load-slip curve.
360 J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368

Fig. 13. Influence of the weld strength on the yield strength and ultimate strength of a single headed stud.

strain law in compression (Eq. (2)). Based on the aforementioned references, the full-range stress-strain
curves were defined for UHPC, as shown in Fig. 9.
8 2
>
> nξ−ξ To simulate the damage evolution process, the damage indexes
>
< fc ε ≤ ε0
1 þ ðn−2Þξ should also be defined. According to relevant studies, concrete damage
σc ¼ ð2Þ index can be calculated using Eq. (3).
>
> ξ
>
: fc 2
εN ε0
2ðξ−1Þ þ ξ rffiffiffiffiffiffiffiffi
σ
D ¼ 1− ð3Þ
where, fc = the compressive strength of UHPC cubic (100 mm × 100 E0 ε
mm × 100 mm), whose value is 135.9 MPa; ξ = the strain ratio, which
is defined as ξ ¼ εε0 ; ε0 = the strain corresponding to the peak stress, By substituting the stress-strain curves into Eq. (3), the damage in-
whose value is 0.0035; n = the ratio of elastic modulus, which is defined dexes were obtained (Fig. 10). In the figure, the abscissa represents
as n ¼ EEsec0 ; E0 = the initial elastic modulus, whose value is 42.6 GPa; Esec the inelastic strain, in which the elastic strain (Eσ0 ) was deducted.
= the secant modulus calculated at the peak stress point on the stress-
strain curve. 3.4.5. Solving method
On the other hand, Zhang et al. [19] obtained the constitutional law In the push-out test, the load was exerted slowly enough to form a
for UHPC under direct tensile loads. It was revealed that the UHPC ex- quasi-static loading process. Thus, Abaqus Explicit should be more suit-
hibited a strain-hardening behavior, accompanied with multiple find able for simulating the testing process.
and tiny cracks. This is beneficial for UHPC to resist tensile or flexural According to the test results, the three specimens had an averaged
stresses. failure load of 480.8 kN. Considering that the top of the specimen had
a contacting area of 284 × 150 = 42,600 mm2, the pressure load
under failure was calculated as 11.3 MPa. A slightly higher pressure
load of 14 MPa was exerted to the FE model. Further, the time period
was defined as 6000 s, a value corresponding to the averaged time dura-
tion of the test. The loading rate was hence calculated as 0.0023 MPa/s.

4. Verification of FE results

4.1. Influence of weld strength

To reveal the influence of the weld strength on the FE results, a de-


tailed comparison was made between the predicted results and the
test results. As mentioned earlier, three cases were considered for the
weld strength, i.e., fweld = fstud, fweld = fstud/0.8, and fweld = fstud × 0.8.

4.1.1. Load-slip curve


In the test, the slips were recorded by eight dial indicators installed
to the stud positions of the specimen. Thus, the slips should be captured
from the same positions in the FE model. In Fig. 11, P1 and P2 represent
two points at which the relative slips were captured. On the other hand,
the reaction force was calculated from the nodes at the bottom of the
Fig. 14. Influence of the weld shape on the load-slip curve. UHPC layer.
J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368 361

Fig. 15. Influence of the weld shape on the yield strength and ultimate strength of a single headed stud.

Fig. 12 shows the load-slip curves obtained from both the FE analysis speaking, the FE model which assumes that the weld and headed
and the experimental test. Generally, the load-slip curves exhibit two studs had an identical strength could yield a satisfactory agreement
stages, i.e., the elastic stage and the yield stage. In the elastic stage, the with the test results.
slip arises linearly with the increasing load and the values of slip are It should be addressed that the three load-slip curves derived from
generally small within this stage. While in the yield stage, the slip in- the experimental test seem to terminate suddenly. The fact is that
creases much faster than the load, indicating that the headed stud al- when failure was near, an increasing snapping sound could be heard.
ready yielded. A distinct turning point could be seen from a load slip Meanwhile, the indexes on the dial indicators rotated much faster and
curve and this point could be deemed as the beginning of the yield the slip values became unstable. To avoid being hurt by a sudden failure
stage. of the specimen, the testers were kept away from the specimen. Thus, it
The figure also indicates that the predicted load-slip curve generally was impossible to read the slip data at the final stage of the test. How-
agrees well with the test results when the weld strength is no less than ever, the load was still monitored using a vibrating-wire force indicator.
that of the headed studs (i.e., fweld = fstud or fweld = fstud/0.8). However, Consequently, the termination of the load-slip curves does not mean
when the weld has a lower strength than the headed studs (i.e., fweld = that the test was stopped suddenly. It only means that the full-range
fstud × 0.8), the test results are underestimated significantly. Generally load-slip curves could not be depicted because of a lack of the slip values
at the final phase of testing.

4.1.2. Yield strength and ultimate strength


The yield load is defined as the distinct turning point on the load-slip
curve, and the ultimate load is understood as the peak point on the load-
slip curve.
As shown in Fig. 13, when the weld strength is higher than that of
the headed studs (i.e., fweld = fstud/0.8), the FE model overestimates
the yield strength and ultimate strength to a headed stud by 11% and
9%, respectively. While in the case of a weaker weld strength (i.e., fweld
= fstud × 0.8), the yield strength and ultimate strength were
underestimated by 15–20%. Thus, the FE model assuming a moderate
weld strength (i.e., fweld = fstud) could gain a satisfactory agreement rel-
ative to the test results. In this case, the difference between the pre-
dicted strength and the test values is 2–7%.

4.2. Influence of weld shape

As mentioned earlier, two cases were considered to reveal the influ-


ence of the weld shape on the analysis results. In one case, the cross sec-
tion of the weld was assumed to a rectangular. While in the other case, a
1 mm × 1 mm slope was simulated at the upper corner of the weld. Fur-
ther, the above discussions indicate that when the weld and headed
studs have equal strength, the FE results are close enough to the test re-
Fig. 16. Position of the slicing plane. sults. Thus, to save the computing cost, the FE analysis in the following
362 J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368

Fig. 17. Equivalent plastic strain contour at the cutting plane.

sections assumes that the weld strength is equal to that of the headed equivalent plastic strain is denoted as PEEQ, which is an important pa-
studs (i.e., fweld = fstud). rameter to reflect the plastic status of a structure. When the value of
PEEQ to an element is greater than zero, it is indicated that the element
4.2.1. Load-slip curve begins to yield.
Fig. 14 compares the theoretical load-slip curves to those obtained in The equivalent plastic strain was captured basing on the stress slic-
the test. It can be seen that the two load-slip curves derived from the FE ing technique in Abaqus. The slicing path is shown in Fig. 16, which rep-
analysis are close to the experimental curves. This implies that weld resents a plane connecting through the center of the two headed studs.
shape only has a minor influence on the load-slip curve. In addition, The equivalent plastic strain countour is projected on the slicing
the FE model with a rectangular weld seems to be a better choice, as plane (Fig. 17). According to Fig. 17(a), plastic strain initiated in UHPC,
the load-slip curve based on this model yields a better agreement from the corner of the roots of the stud weld. Then the plastic strain de-
with the test results. veloped more rapidly in the headed studs than in the UHPC layer. With
the load increasing, the plastic strain continuously develops in the
headed studs, and a slight separation could be seen at the front surface
4.2.2. Yield strength and ultimate strength
of the headed studs. Finally, the two headed studs failed, indicating a
According to Fig. 15a, the yield strength of a single headed does not
failure of the push-out specimen. The 1# headed stud failed from the
show a significant difference, implying that the effect of the weld shape
weld, and the 2# headed stud broke from the stud shank.
on the yield strength should be ignorable. Regarding the ultimate
Fig. 17(b) exhibits the damage process of the two headed stud. As
strength (Fig. 15b), the FE model without the slope provides a closer re-
shown in the figure, the plastic strain initiated from the corner of the
sult (58.9 kN) relative to the test result (60.1 kN), with the relative dif-
stud and propagated through the stud weld. An obvious deformation
ference being only 2%. Thus, there is no necessity to consider a weld
could be seen in both the weld collar and the stud shank. The elements
slope in the FE analysis.
with severe damage were removed from the FE model automatically
when failure occurred.
4.3. Damage evolution and failure mode The damage evolution is illustrated for UHPC in Fig. 18, by using the
damage index (D). Generally, the most portion of UHPC was intact, and
Based on the above discussions, the weld strength is recommended damage was only observed at localized zones around the headed studs.
to be identical as that of the headed studs. Further, it is unnecessary to Tensile damage mainly develops at the rear of the headed studs, while
consider a slope at the corner of the weld. The FE models built under compression damage is obvious at the front of the headed studs.
these considerations could yield a good result. The FE analysis implies that the failure of the specimen was caused
To investigate the damage process of the push-out specimens, the by the fracture of the headed studs (Fig. 19a). While the UHPC did not
equivalent plastic strain was observed in the FE model. In Abaqus, the
J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368 363

T=1800s

T=3900s

T=4500s

T=5400s

(a) Tensile damage (b) Compressive damage


Fig. 18. Damage evolution in UHPC.

develop appreciable damage, except for local zones near the stud roots i.e., break of stud shank and failure of concrete, whichever occurs first.
(Fig. 19b). These observations agree well with the test results. Thus, the Eq. (4) shows a theoretical equation to calculate the shear strength of
FE model could reflect the damage process and failure mode of a push- a headed stud embedded in concrete.
out testing. qffiffiffiffiffiffiffiffiffi
0
Q u ¼ ϕ0:5Asc f c Ec ≤ϕAsc f ð4Þ
5. Parametric analysis and discussion
where, Asc = sectional area of headed studs (mm2); f'c = compressive
5.1. General information
strength of concrete cylinders (MPa); Ec = modulus of elasticity of con-
crete (MPa); f = tensile strength of headed studs (MPa); and ϕ = resis-
Push-out tests on steel-UHPC composite structures are costly. Thus,
tance reduction factor, whose value is 0.85.
to further reveal the behavior of the headed studs embedded in UHPC,
Eq. (5) presents an equation specified in Eurocode 4 (ECS 2005). On
a parametric analysis was performed basing on the validated FE model.
the left part of the equation, the aspect ratio of the headed studs is con-
Three parameters are included, i.e., the stud diameter, stud height,
sidered.
and compressive strength of UHPC. Fig. 20 shows the notations of
main parameters. A series of FE models were established following the qffiffiffiffiffiffiffiffiffi
2 0
methods described previously. Q u ¼ 0:29αd f c Ec =γ v ≤0:8Asc f =γ v ð5Þ

5.2. Theoretical equations for calculation of shear strength where α = the aspect ratio coefficient, whose value is α = 0.2(h/d + 1)
≤ 1.0 when h/d ≥ 3, in which h and d are the height and diameter of
As stated in AASHTO LRFD (AASHTO 2012), the shear strength for headed studs, respectively; f'c = compressive strength of concrete cylin-
headed studs in normal concrete is governed by two failure criteria ders (MPa); Ec = elastic modulus of concrete (MPa); Asc = cross-
364 J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368

Fig. 19. Comparison of FE observations to test results.

sectional area of headed studs (mm2); f = tensile strength of headed Table 1


studs (MPa); and γv = material partial factor, whose value is 1.25. Parameters of headed studs embedded in UHPC.
The above equations are suitable for headed studs embedded in nor-
Serial No. Sizes of headed stud (mm) Thickness of
mal concrete. While for headed studs in UHPC, only limited studies are UHPC (mm)
available. Luo et al. (2015b) proposed an equation to predict shear
d dwc dk h lwc k H c
strength of headed studs in UHPC, as shown in Eq. (6). Note that in
this equation, only one failure mode (i.e., stud failure) is considered. 1 13 17 22 80 3 8 95 15
2 16 21 29 80 4.5 8 95 15
This should be attributed to the fact that UHPC has extremely high
3 19 23 32 80 6 10 95 15
strengths and is consequently difficult to fail in a push-out test. In addi- 4 22 29 35 80 6 10 95 15
tion, the equation considers the contribution of the weld collar on the

UHPC
8+3&ቲlayer
c

dk
k

d
Weld
h

Headed stud
ṃ䪹
collar
✺㕍
lwc
lw

ddwcw
Steel
䫒ᶯplate

Fig. 20. Schematic drawing of the parametric analysis. Fig. 21. Influence of stud diameter on load-slip curve.
J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368 365

Fig. 22. Influence of stud diameter on shear strength of headed studs in UHPC.
Fig. 24. Influence of stud height on load-slip curve.

shear strength.
 0

Q u ¼ 0:85Asc f þ η f c dwc lwc =γ v ð6Þ

where, Asc = cross-sectional area of headed stud (mm2); f = tensile


strength of headed stud (MPa); η = improvement coefficient of shear
strength produced by weld collar, whose value is 2.5; f'c = compressive
strength of concrete cylinders (MPa); dwc =diameter of weld collar
(mm); lwc = height of weld collar (mm); γv = material partial factor,
whose value is 1.25.
The UHPC investigated in this paper had a compressive strength of
fcu = 135.9 MPa and a modulus of elasticity of Ec = 42.6 GPa. Further,
experimental tests accomplished by Graybeal [20] indicated that the
compressive strength of UHPC derived from cylinders and cubes did
not exhibit significant differences. Thus, the compressive strength of
UHPC cylinders could be calculated as f'c = 0.95fcu = 129.1 MPa. Headed
studs have a nominal tensile strength of f = 400 MPa [21].

5.3. Definition of shear stiffness

Shear stiffness is an important parameter for headed studs in con- Fig. 25. Influence of stud height on shear capacity of headed studs in UHPC.
crete. The calculation of shear stiffness for headed studs is generally

Fig. 23. Influence of stud diameter on the yield strength and shear stiffness for a single headed stud.
366 J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368

Fig. 26. Influence of stud height on the yield strength and shear stiffness for a single headed stud.

Fig. 27. Constitutional laws of different grades of UHPC in compression.

based on the load-slip curve. The shear stiffness is defined as the slope of
a secant line on the load-slip curve. The lower point of the secant line is
the original point, and the upper point is different according to different
methods, which normally corresponds to 0.3–0.7Qu [8].
In this study, the method recommended by Eurocode 4 is used.
The definition of the shear stiffness of a single headed stud is shown
in Eq. (7).

k ¼ 0:7Q u =s ð7Þ

where, k = the shear stiffness of a headed stud; Qu = the ultimate


strength of the headed stud; s= the relative slip corresponding to a
load of 0.7Qu on the load-slip curve.

5.4. Parameter 1: stud diameter

According to the Chinese design specification [22], headed stud with


standard sizes has a diameter of 10 to 25 mm, with an increment of 3
mm. Considering that the diameters of 10 mm and 25 mm are not fre-
quently used in composite structures, only four diameter categories Fig. 28. Influence of UHPC strength on load-slip curve.
J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368 367

by Eq. (6) exhibit a good agreement with the FE results. While the re-
sults based on the other two equations seem to be too conservative.
This can be explained as that Eq. (4) and Eq. (5) are suitable for headed
studs embedded in normal concrete. In the two equations, the contribu-
tion of weld collar to the shear strength is not included. According to Fig.
19, the failure of the headed studs was accompanied with cracking and
crushing in UHPC closely around the stud root positions. This implies
that for the headed studs in UHPC, the contribution of the weld collar
should not be ignored. Otherwise, the calculation results may be too
conservative.
The influence of stud diameter on the yield strength and shear stiff-
ness is also analyzed. As shown in Fig. 23, for the headed studs with di-
ameters of 13–22 mm, the yield strength is about 47–116 kN,
accounting for 75–82% of the ultimate strength. On the other hand,
the shear stiffness per headed stud is 258–420 kN/mm. The figure also
indicates that a larger headed stud produces higher yield strength and
shear stiffness.

5.5. Parameter 2: stud height


Fig. 29. Influence of UHPC strength on shear strength.
In this subsection, the influence of stud height is analyzed. The
headed studs had a fixed diameter of d = 13 mm. Five heights were
were considered in the parametric analysis. The sizes of each diameter considered, i.e., h = 20 mm, 35 mm, 50 mm, 65 mm, 80 mm. The
category can be referred to in the Chinese specification [22]. cover was also set to a fixed value of c = 15 mm.
Table 1 presents the dimensions of the headed studs involved in the Fig. 24 shows the load-slip curves derived from the FE analysis. From
parametric analysis, as well as the thickness of UHPC. To eliminate the the figure, the linear and yield stages are distinct in all curves. Further, it
influence of stud height, a fixed value should be assigned for the height. is clear that the load-slip curves are close to each other regardless of the
Considering that the minimum height-to-diameter ratio of a headed stud height, and this phenomenon is especially true within the linear
stud is about 4, the stud height was set as 80 mm in all of the four cases. stage. It can be concluded from the figure that for headed studs embed-
Four push-out FE models were built basing on the configurations ded in UHPC, the load-slip curves are nearly unaffected by the parame-
shown in Table 1. The method used to establish the FE models was iden- ter of stud height.
tical to that described previously. The shear strengths of the headed studs with different heights are
The load-slip curves are presented in Fig. 21. It can be observed that depicted in Fig. 25. The figure confirms that for headed studs in UHPC,
all of the load-slip curves are characterized by two stages, i.e., the elastic the stud height has negligible influence on the shear strength. This
stage and the yield stage, regardless of the stud diameter. In the elastic should be attributed to the fact that stress concentration constantly de-
stage, the interfacial slip increases linearly with the increasing load. At veloped from the stud root, regardless of the stud height. Hence, the
the end of this stage, the curve gradually deviates from its original headed studs failed from the root zones in all of the five cases. In addi-
path, indicating that the specimen steps into the yield stage. The figure tion, the figure shows that the shear strength calculated by Eq. (6) is
shows that the larger the stud diameter, the higher the yielding load. In closer to the FE result. This implies that for headed studs in UHPC, the
addition, the ultimate load is also enhanced with a larger stud diameter. contribution of the weld collar should be considered to evaluate the
Thus, the load-slip curve is significantly affected by the stud diameter. shear strength.
Fig. 22 compares the shear strength predicted in the FE analysis to According to Fig. 26(a), the yield strength nearly remains unchanged
that calculated through Eqs. (4)–(6). The theoretical results calculated with the increasing stud height. This indicates that the yield strength of

Fig. 30. Influence of UHPC strength on the yield strength and shear stiffness for a single headed stud.
368 J. Cao, X. Shao / Journal of Constructional Steel Research 161 (2019) 355–368

the headed studs in UHPC is insensitive to the stud height. It is also of UHPC, is marginal.
found that the yield strength accounts for 78–82% of the ultimate (3) The shear strengths of the headed studs obtained in FE analysis
strength. Further, Fig. 26(b) implies that the influence of the stud height were compared to those predicted by theoretical equations. It
on the shear stiffness is not distinct. Generally, when the stud heights was found that Eq. (6) exhibited a better agreement with the
are 20–80 mm, the shear stiffness is within a range of 194–257 kN/mm. FE results than the other two equations. The agreement implies
that the contribution of the weld collar to the shear strength
5.6. Parameter 3: compressive strength of UHPC should be included for the headed studs in UHPC.

Headed studs may be embedded in UHPC slabs with different grades.


Thus, it is necessary to reveal the influence of the compressive strength
on the behavior of the headed studs. Acknowledgement
The diameter of headed studs (d) is 13 mm and the stud height (h) is
80 mm. The compressive strength of UHPC (fc) was considered as 110 The authors would like to thank the following funders for their sup-
MPa, 140 MPa, and 170 MPa. port to the studies in this paper: (1) the National Natural Science Foun-
Constitutional laws were established for the UHPC with different dation of China (grant Nos. 51708200 and 51778223); (2) China
grades. Regarding the stress state of compression, the stress–strain Postdoctoral Science Foundation (grant No. 2017 M610497); (3) the
curve was calculated using Eq. (2). The stress-strain curves for different Science and Technology Agency in Hunan Province (grant No.
grades of concrete and UHPC are presented in Fig. 27(a), and the dam- 2018JJ3052); (4) the Major Program of Science and Technology of Hu-
age index-inelastic strain curves are illustrated in Fig. 27(b). While re- nan Province (grant No. 2017SK1010).
garding the stress state of tension, the corresponding subfigures in
Figs. 9 and 10 are used. References
Fig. 28 shows the load-slip curves in the FE analysis. The figure dem- [1] L. Pallarés, J.F. Hajjar, Headed steel stud anchors in composite structures, part I:
onstrates that in the elastic stage, the three curves nearly coincide with shear, J. Constr. Steel Res. 66 (2) (2010) 198–212.
[2] D.J. Oehlers, C.G. Coughlan, The shear stiffness of stud shear connections in compos-
each other. In the yield stage, the curves exhibit a slight difference, while
ite beams, J. Constr. Steel Res. 6 (1986) 273–284.
the trend of the three curves is the same. Hence, the compressive [3] E. Xie, Fatigue strength of shear connectors, University of Minho, Guimarães, 2011
strength of UHPC has a marginal effect on the load-slip curve. 1–58.
Fig. 29 compares the shear strength in FE analysis to that obtained in [4] J. Hildebrand, H. Soltanzadeh, A review on assessment of fatigue strength in welded
studs, Int. J. Steel Struct. 14 (2) (2014) 421–438.
Eq. (4)-Eq.(6). The FE analysis indicates that when the compressive [5] E. Fehling, M. Schmidt, J. Walraven, T. Leutbecher, S. Fröhlich, Ultra-High Perfor-
strength of UHPC is increased, the ultimate strength of the headed mance Concrete UHPC, Fundamentals – Design – Examples, Wiley Ernst & Sohn,
studs improve slightly. This confirms that the compressive strength Germany, 2013.
[6] X.D. Shao, D.T. Yi, Z.Y. Huang, H. Zhao, B. Chen, M.L. Liu, Basic performance of the
should have a slight influence on the shear strength for the headed composite deck system composed of orthotropic steel deck and ultrathin RPC
studs. Again, the shear strength predicted by Eq. (6) is closer to the FE layer, J. Bridg. Eng. 18 (5) (2013) 417–428.
result, implying that the contribution of the weld collar to the shear [7] J.S. Kim, J. Kwark, C. Joh, S.W. Yoo, K.C. Lee, Headed stud shear connector for thin ul-
trahigh-performance concrete bridge deck, J. Constr. Steel Res. 108 (2015) 23–30.
strength should not be ignored. [8] J.H. Cao, X.D. Shao, L. Deng, Y.D. Gan, Static and fatigue behavior of short headed
Fig. 30 implies that the influence of the compressive strength of studs embedded in thin UHPC layer, J. Bridg. Eng. 22 (5) (2017).
UHPC on the yield strength and shear stiffness of the headed studs is [9] H.T. Nguyen, S.E. Kim, Finite element modeling of push-out tests for large stud shear
connectors, J. Constr. Steel Res. 65 (2009) 1909–1920.
negligible. Generally, with the compressive strength of UHPC being in-
[10] Y. Luo, K. Hoki, K. Hayashi, M. Nakashima, Behavior and strength of headed stud –
creased from 110 MPa to 170 MPa, the yield strength of the headed SFRCC shear connection, II: strength evaluation, J. Struct. Eng. 142 (2) (2015)
studs is about 46 kN, 78–86% to the ultimate load. On the other hand, https://doi.org/10.1061/(ASCE)ST.1943-541X.0001372.
[11] S. Guezouli, A. Lachal, Numerical analysis of frictional contact effects in push-out
the shear stiffness per headed stud is 237–256 kN/mm.
tests, Eng. Struct. 40 (2012) 39–50.
[12] D. Lam, E. El-Lobody, Behavior of headed stud shear connectors in composite beam,
6. Conclusions J. Struct. Eng. 131 (1) (2005) 96–107.
[13] R.G. Slutter, J.W. Fisher, Fatigue strength of shear connectors, Highw. Res. Rec. 147
(1966) 65–88.
This paper investigates the behavior of the headed studs embedded [14] Dassault Systèmes Simulia, ABAQUS Analysis User's Manual, Version 6.12, Hibbitt,
in thin UHPC. A 3D FE model was built to evaluate the damage process Karlsson, and Sorenson Inc., Providence, 2012.
and failure mode for the push-out specimens. The FE model was vali- [15] S.E. Kim, H.T. Nguyen, Finite element modeling and analysis of a hybrid steel-PSC
beam connection, Eng. Struct. 32 (2010) 2557–2569.
dated basing on the test results and hence a parametric study was per- [16] X.F. Ding, M. Ni, Y.Z. Gong, Z.W. Yu, Z. Zhou, L.C. Zhou, Experimental study on slip
formed. Based on the current study, the following conclusions can be behavior and calculation of shear bearing capacity for shear stud connectors, J.
drawn: Build. Struct. 35 (9) (2014) 98–106 (In Chinese).
[17] M.S. Kunihiko, Welding engineering, Tokyo 1979.
(1) Material nonlinearity was carefully defined for different parts of [18] J. Yang, Z. Fang, Research on stress-strain relation of ultra high performance con-
crete, Concrete 7 (2008) 11–15.
the push-out specimen in the FE analysis. The analysis indicated [19] Z. Zhang, X.D. Shao, W.G. Li, P. Zhu, H. Chen, Axial tensile behavior test of ultra high
that the load-slip curve and shear strength were highly in accor- performance concrete, China J. Highway Transport 28 (8) (2015) 50–58.
dance with the test results. In addition, the failure mode ob- [20] B.A. Graybeal, Characterization of the Behavior of Ultra-High Performance Concrete,
Department of Civil and Environmental Engineering, University of Maryland College
served in the test could be adequately reflected by the FE model.
Park, Maryland, 2005 36–64.
(2) Based on the validated FE model, a parametric analysis was con- [21] MOHURDC, Code for design of steel and concrete composite bridges (GB 50917-
ducted. The analysis results imply that the parameter of stud di- 2013), Ministry of Housing and Urban-Rural Development of China, Beijing, China,
ameter exhibits a significant effect on the response of the headed China Planning Press, 2013 1–118 (In Chinese).
[22] GAQSIQOC, Cheese head studs for arc stud welding (GB/T 10433-2002), General Ad-
studs, including the load-slip curve, yield strength, ultimate ministration of Quality Supervision, Inspection and Quarantine of China, Beijing,
strength, and shear stiffness. While the influence of the other China, 2002 1–2 (In Chinese).
two parameters, i.e., the stud height and compressive strength

You might also like