Steel-Yielding Demountable Shear Connector for

Composite Floors with Precast Hollow-Core Slab Units

Eliza Feidaki 1; George Vasdravellis 2; Jun He 3; and Sihao Wang 4

Abstract: A demountable shear connector for use in composite floors with precast hollow-core slab units is proposed. The proposed shear
connector consists of a steel square hollow tube welded on a steel plate that is bolted on the steel section’s top flange using four high-strength
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

bolts. Concrete is cast only around the connector, allowing for an easy separation of the precast slab units from the steel sections during the
deconstruction phase, with operations required only from the top of the composite floor. The shape of the hollow tube promotes a ductile
elastic-plastic behavior under the longitudinal shear flow in a composite beam. Ten push tests using a horizontal testing arrangement were
carried out to assess the structural performance of the novel shear connector. The experimental results show that the shear connector has a
stiffness comparable to that of a welded shear stud, strength that can be adjusted to achieve the desired degree of shear connection in a
composite beam, and slip capacity that is much higher than the requirements of the current codes of practice. A finite-element model
was also calibrated against the tests and found capable of accurately reproducing the experimental behavior. The finite-element model
was then used to conduct a number of parametric studies to safely generalize the experimental results. Based on the experimental and numeri-
cal results, a design model is proposed to predict the strength of the demountable shear connector based on either the yielding of the steel tube
or the failure of the precast slab. DOI: 10.1061/(ASCE)ST.1943-541X.0002356. © 2019 American Society of Civil Engineers.

Introduction (Webster and Costello 2005; Densley Tingley and Davison 2011;
Brambilla et al. 2019).
The construction sector plays an important role in achieving future A number of recent research efforts have been devoted to devel-
sustainability targets set by many countries, since the production of oping novel methods that allow for the separation of slabs from
steel and cement, the two materials that are exclusively used in con- steel sections in a composite floor and the reuse of components
struction, accounts for 15% of the total global anthropogenic CO2 using demountable shear connectors. Lam and Saveri (2012) and
emissions (IEA 2007). Moreover, it is foreseen that the material Dai et al. (2015) proposed the use of threaded Nelson shear studs
demand will double globally by 2050, indicating that more respon- that are bolted (instead of welded) on the top flange of steel sec-
sible consumption of natural resources should be adopted. In ad- tions. They investigated the performance of bolted studs in pushout
dition, the construction and demolition sectors are responsible for a tests and performed a parametric study using a finite-element
large amount of waste sent to landfills (Burka 1993). Thus, high method (FEM). They found that the bolted studs had comparable
sustainability standards in construction can be achieved by devel- strength, higher slip capacity, and approximately half the initial
oping structural systems that offer the advantage of deconstruction stiffness of the welded studs due to the unavoidable clearance be-
at the end of their service life, as opposed to demolition, and the tween the stud collar and the flange holes. Rehman et al. (2016)
reuse of structural components. conducted 12 pushout tests using threaded shear studs in slabs with
The so-called design for deconstruction principle, however, is profiled metal decking and found a behavior similar to that in solid
difficult to apply to steel–concrete composite floors, which are slabs but with a slightly reduced ultimate strength. Pavlovic et al.
widely used in multistory buildings worldwide, because the headed (2013) carried out four pushout tests and numerical simulations to
shear studs are commonly welded on the steel sections and em- investigate the performance of high-strength bolts (M16, Grade
bedded in concrete. For this reason, recent research works recom- 8.8) as shear connectors in precast slabs with cast-on-site openings.
mend avoiding this type of construction when designing for reuse Compared to welded shear studs, the high-strength bolts exhibited
similar strength and half the initial stiffness, and their slip capacity
1
was limited to 4–5 mm. Moynihan and Allwood (2014) conducted
Ph.D. Candidate, School of Energy, Geoscience, Infrastructure and three composite beam tests using M20 bolts as shear connectors.
Society, Institute for Infrastructure and Environment, Heriot-Watt Univ.,
Edinburgh EH14 4AS, UK.
They found that the flexural response of specific beams with
2
Associate Professor, School of Energy, Geoscience, Infrastructure and demountable connectors was comparable to that of previously
Society, Institute for Infrastructure and Environment, Heriot-Watt Univ., tested beams with welded shear studs in terms of ultimate strength
Edinburgh EH14 4AS, UK (corresponding author). ORCID: https://orcid and ductility. Liu et al. (2015) proposed the use of high-strength
.org/0000-0002-7910-8190. Email: [email protected] friction-grip bolts as shear connectors in conjunction with precast
3
Associate Professor, School of Civil Engineering, Changsha Univ. of geopolymer concrete slabs and evaluated them through pushout
Science and Technology, Changsha Shi, Hunan Sheng 410205, China. tests. The friction bolts exhibit a rigid initial behavior due to pre-
4
Ph.D. Candidate, Dept. of Bridge Engineering, Tongji Univ., Shanghai tension, a sudden slip once the friction force is overcome, followed
Shi 200072, China. ORCID: https://orcid.org/0000-0002-6281-4937.
by a plastic response and subsequent bolt shear failure. Further tests
Note. This manuscript was submitted on May 21, 2018; approved on
December 17, 2018; published online on May 31, 2019. Discussion period on composite beams using high-strength friction-grip bolts and
open until October 31, 2019; separate discussions must be submitted for geopolymer concrete precast slabs (Ataei et al. 2016; Liu et al.
individual papers. This paper is part of the Journal of Structural Engineer- 2017) showed good ductility and adequate ultimate strength.
ing, © ASCE, ISSN 0733-9445. Pathirana et al. (2015, 2016a) proposed the use of two blind bolt

© ASCE 04019076-1 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 for the placement of rebars for the effective connection between
adjacent units. Shear studs are welded on the steel section, and in
situ concrete is poured to fill the open cores and to connect the units
Precast Rebars with the studs. Fig. 1 shows a composite beam section using HCUs.
units Since the welded shear studs are fully embedded in the concrete,
the deconstruction of such composite floor systems is still problem-
In situ
atic; however, demountable shear connectors for use with HCUs are
concrete not yet proposed in the literature.
In this paper, a demountable steel-yielding shear connector is
proposed for use in composite floors with HCUs. The proposed
shear connector consists of a steel square hollow tube welded
on a steel plate that is bolted on the steel section’s top flange using
Fig. 1. Steel-concrete composite section with HCUs. four high-strength bolts, and it has the following unique features:
(1) it is not fully embedded in concrete, so operations only from the
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

top of the floor are required to dismantle the composite floor system
during deconstruction; (2) it has a shape that promotes a ductile and
types to retrofit composite beams and compared their behavior with predictable force-slip behavior under the longitudinal shear force in
that of welded shear studs through pushout tests. The blind bolts in a composite beam; (3) it has high initial stiffness, strength that can
the pushout tests had a higher ultimate capacity than the welded be adjusted to required levels, and slip capacity much higher than
studs, but smaller stiffness. They also tested the blind bolts in the minimum requirements of current codes of practice, a feature
full-scale composite beams, which showed comparable behavior that enables the designer to use large spans with partial shear con-
to beams with welded studs in terms of strength, stiffness, and nection; and (4) it eliminates the issue of tolerances in the reuse
ductility (Pathirana et al. 2016b). Blind bolts as an alternative to phase, as opposed to the case of bolted connectors that protrude
welded studs were successfully implemented in composite beams from slabs and, therefore, may not fit precisely in the predrilled
under sustained loads (Ban et al. 2015) and under dynamic loading holes of steel sections. The proposed shear connector was exper-
(Henderson et al. 2015a, b, 2017). Suwaed and Karavasilis (2017) imentally evaluated through 10 push tests using a horizontal testing
proposed a locked-nut shear connector for use in composite bridges arrangement. A detailed numerical model based on the FEM was
with precast decks. They used a cone-shaped nut that “locks” into then calibrated against the experimental results and found to be
counter-shaped corresponding holes on the steel section flange to capable of accurately simulating the experimental behavior. The
eliminate the initial slip due to tolerances. The resulting force-slip validated FEM model was used to carry out a number of parametric
response was characterized by high initial stiffness (comparable to studies to safely generalize the experimental findings. Based on the
that of a welded stud) and strength and slip capacities that were experimental and numerical results, a design model to predict the
significantly higher than those of a welded stud. Note that the strength of the proposed connector based on either the yielding of
minimum slip capacity required by Eurocode 4 (BSI 2004) in order the steel element or the failure of the precast slab is proposed.
for a shear connector to be characterized as ductile is 6 mm.
A composite floor system using precast concrete slab units
acting compositely with steel sections offer the advantages of Steel–Concrete Composite Floor with Hollow-Core
off-site quality-controlled fabrication, flexibility in the geometric Slabs and a Demountable Steel-Yielding Shear
characteristics of the precast units, possibility for custom-made Connection Mechanism
geometries, durability, and fewer operations on site (Couchman
2014). The use of precast hollow-core units (HCUs) offers the addi-
tional advantage of large spans without the need for secondary Description
beams, enabling even greater economy in steel and fewer opera- The proposed composite floor system uses precast HCUs and a novel
tions on site. Typically, HCUs are 1.2 m wide, and alternate cores demountable shear connection mechanism consisting of a steel-
are left open to the top for a specific length (usually 0.5 m) to allow yielding device, referred to as the yielding pocket (YP). Fig. 2(a)

YP Slotted hole and rebar In situ concrete

Foam
HCU 1

HCU 1 HCU 2

Edge
4xBolts
cut out Polythene foam
Elongated holes
HCU 2 Steel section

Deformed YP:

Rebars
Steel section
(a) (b)

Fig. 2. (a) 3D view of proposed demountable shear connector; and (b) longitudinal view and deformed YP.

© ASCE 04019076-2 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 Tooth Tooth
Plastic hinges

h1
h1 h

Steel plate Steel strips

w w w 20 mm
(a) (b) (c)

Fig. 3. (a) Geometric properties of a YP’s vertical strips; (b) plan view of YP showing vertical wall sections (hatched regions); and (c) assumed plastic
mechanism of YP under longitudinal shear force.
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

shows a three-dimensional (3D) view of a beam segment with two and the thickness of the hollow section, t. The foam around the YP
HCUs, one YP installed, and three open cores per HCU. One or two is placed in such a way that the bottom side of the concrete tooth is
(depending on the degree of shear connection required, partial or in line with the top end of the vertical strips. Under the longitudinal
full) cutouts are made through the depth of the HCUs at the edges shear force in a composite beam, the YP deforms as shown in
parallel to the beam axis in order to accommodate the installation of Fig. 3(c). The resistance, therefore, is provided by the plastic bend-
YPs. A YP consists of a steel square hollow section having a steel ing of the vertical strips and the vertical walls. The section of the
plate welded at the bottom and total length equal to the slab depth. vertical walls is shown in Fig. 3(b) as hatched regions. Assuming
Vertical elongated holes are cut on the sides of the YP that are par- fixed boundary conditions, the vertical strips are expected to de-
allel to the beam axis in order to form vertical steel strips. Aligned velop two plastic hinges at their ends, as shown in Fig. 3(c). The
horizontally slotted holes are also opened on the same sides of the distance between the top and bottom plastic hinges is h1 , which is
YP. A 3D view of a YP is shown in Fig. 2(a). less than h. h1 is taken as the distance between the sections of the
Four high-strength bolts are used to clamp a YP on the top strip with width w that are just before the semicircular part of the
flange of the steel section, as shown in the longitudinal section elongated hole [Fig. 3(c)]. The plastic moment of resistance of a
of Fig. 2(b). Rebars are placed in the open cores. A rebar is inserted vertical strip section is given by
through the horizontally slotted holes of the YP and placed in the
tw2
middle open core, aiming to prevent uplift of the slab during M p;strip ¼ f ð1Þ
the bending of the beam. A 50-mm gap is left between the YP 4 ys
and the HCUs, and in situ concrete is poured to fill the open cores where fys = yield strength of the vertical strip material. The force
and the gaps around the YP. Before pouring the in situ concrete, provided by all the vertical strips is given by
polythene foam is used as formwork around the YP to disconnect
the bottom part of the concrete from the YP. In this way, the lon- M p;strip
Fp;strips ¼ 2 n ð2Þ
gitudinal shear force resultant is moved slightly above the base of h1
the YP and resisted by the bending of the vertical steel strips and the
walls of the YP. The inset of Fig. 2(b) shows the expected deformed where n = total number of strips.
shape of the YP under the longitudinal shear force. Thus, a ductile Likewise, the vertical walls are expected to develop plastic
failure of the YP is promoted, as opposed to a brittle failure due hinges at their top and bottom ends. Using plastic sectional analy-
to bolt shearing that would occur if the concrete was in full-depth sis, the plastic moment of resistance of the wall section [hatched
contact with the YP. region in Fig. 3(b)] is given by
To deconstruct the system, the concrete “tooth” around the YP M p;wall ¼ Zp f yw ð3Þ
[Fig. 2(b)] is removed using a standard concrete-cutting tool, the
rebar passing through the YP is cut, and the YPs are unbolted where Zp = plastic section modulus of the wall section; and
and removed, which is an operation performed from the top of fyw = yield strength of the material of the vertical walls. The force
the beam. The slab can then be removed and recycled using stan- provided by the two vertical walls is
dard methods, while both the steel sections and the YPs can be
reused directly in new building construction projects. Another pos- M p;wall
Fp;walls ¼ 4 ð4Þ
sibility that this system offers is to reuse not only the steel sections h1
and YPs but also the precast slabs. This would require removing the
concrete strip above the steel section and cutting all the rebars in the Thus, the total strength of the YP is given by
middle. The HCUs will have rebars protruding out of the concrete FYP ¼ Fp;strips þ Fp;walls ð5Þ
(since only the concrete strip above the steel section is removed
during deconstruction, as described previously). To reuse them To allow for the aforementioned plastic shear-resisting mecha-
in a new building and avoid reopening the cores, couplers can nism to develop, concrete failure should be prevented before the
be used to connect the protruding rebars, as indicated in Fig. 2(a). force FYP has been reached. Possible failure modes of a slab
due to dowel bearing forces include (Oehlers and Bradford 1995)
concrete shear failure with the formation of diagonal cracks, con-
Mechanics-Based Strength of Demountable Shear
crete splitting, and concrete ripping. Of these failure mechanisms,
Connector
concrete shear failure was found to be the most critical. The con-
Figs. 3(a and b) show the geometric properties of a YP, which are crete shear strength was taken as the shear strength of a cracked
the width of the vertical strips, w, the height of the vertical strips, h, plane, according to Eurocode 2 (BSI 2002b):

© ASCE 04019076-3 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 Fconc ¼ 2ðcf ct Ac þ μf yr Atr κÞ ð6Þ Prototype Steel-Concrete Composite Floor System
Based on the foregoing design procedure, a YP’s strength can be
where fct = lowest tensile strength between in situ concrete and adjusted by selecting the width and height of the elongated holes
HCU concrete; Ac = area of shear plane; fyr and Atr = yield stress and the thickness of the hollow section. Fig. 4 shows a 3D view of
and area of transverse reinforcement, respectively; κ = reduction part of a composite floor system having 12-m-long main composite
factor to account for insufficient anchorage between reinforcement beams using the proposed structural details. The main beams are
and concrete, taken as 0.5, as recommended in fib (2010); and placed at 7.5-m intervals, and 200-mm-deep HCUs are used to span
c and μ = cohesion and friction coefficients equal to 0.5 and this distance without the need for secondary beams. One cutout is
0.9, respectively, as recommended in Eurocode 2 (BSI 2002b). made in the midlength of each 1.2-m-long HCU and a half cutout at
Eq. (6) represents the contribution of adhesion, aggregate interlock, both ends. In this way, YPs are placed at 600-mm center-to-center
and friction mechanism along a shear crack. The coefficient 2 intervals on the steel section. Using a typical office building loading
accounts for the fact that there are two shear critical planes, one (BSI 2002a), and assuming simply supported boundary conditions,
on each side of the YP. the design of a main composite beam requires an IPE550 steel sec-
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

tion and 24 YPs with strength equal to 280 kN each to achieve full
Steel section
shear connection. This prototype composite floor system served to
L=12 m select the specimen geometry for the experimental program.

Experimental Program

HCU 200
Test Setup
mm To study the structural behavior of the proposed demountable shear
connector, ten push tests were conducted using the horizontal test-
ing arrangement shown in Fig. 5. The reason for using an horizontal
test setup instead of the standard vertical setup prescribed in
Eurocode 4 (BSI 2004) is practical, i.e., the size of the HCUs would
make the fabrication and handling of the specimens problematic
in the Lab environment. Other reasons for conducting push tests in
an horizontal arrangement when using precast HCUs are given in
Lam (2007).
An hydraulic actuator of 1,000 kN force capacity and 500 mm
Fig. 4. Design example of composite floor system using proposed de-
displacement capacity with attached load cells was used to push the
mountable shear connector.
HCUs against the steel section through a strong spreader beam,

Steel rods M28 for

floor connection
Spreader beam
UC305X305X13 Specimen
Reaction frame
ribs 40 mm

Actuator
HCU 1
2m

HCU 2

1m 3m 4.30 m
6.30 m

Wooden
Reaction frame Spreader beam part
Actuator
UC305X305X137
1.20 m Specimen
Mortar
tooth
HCU

Strong floor

Threaded Support Support

rods M28

Fig. 5. Test setup.

© ASCE 04019076-4 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 indicated in Fig. 5. The actuator was attached on a strong reaction
HCU 1
frame bolted to the strong floor through eight M28 high-strength

800 mm

1670 mm
bolts. To fix the steel section of each specimen to the strong floor,
four 30-mm-thick steel plates were welded to the bottom flange and Rebar 12
perpendicular to the section and then bolted to the strong floor us- In situ concrete
ing M28 steel rods, as shown in Fig. 5. The bolts and rods were Foam
pretensioned to their proof load to create a strong friction connec-
Steel section
tion, thereby eliminating relative slip. The slab protruded from the

500 mm
UB533x210x92
steel section by 100 mm to allow the movement of the spreader YP
beam against the steel section. The centroid of the actuator was 4xbolts
in line with the middepth of the slab. The spreader beam was sup- HCU 2
ported by a steel base. The top surface of the base was greased to
minimize the friction force during the sliding of the actuator on it.
(a)
Two configurations were employed to transfer the load to the
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

specimens, shown in Fig. 5(b). The first configuration used wooden YP

parts of a rectangular section between the HCUs and the flange of In situ concrete 12
Foam
the spreader beam. The second configuration used a mortar tooth 12 200 mm
cast before testing between the flange of the spreader beam and the
HCU
HCUs. All the specimens were tested using the first configuration 100 mm

533 mm
apart from Specimens SP1A and SP6, for which the second con- 4xBolts
figuration was used. Steel section 1s 2h 3s 4s 5s 6h 7s 8h 9s

UB533X210X92

Specimen Description and Design Parameters 2000 mm

(b)
A plan view and a longitudinal section of a typical push specimen
are shown in Figs. 6(a and b), respectively. The specimens’ dimen- Fig. 6. Specimen details.
sions were selected based on the prototype composite floor
design. Each specimen consisted of two HCUs connected to a
UB533 × 210 × 92 (equivalent to an IPE550 or W530 × 101)
steel section using a single YP. The HCUs used for all specimens The geometric parameters of the YP were changed to produce a
had dimensions (depth × nominal width × length) equal to 200 × different FYP according to Eq. (5). The specimens in Groups
1,200 × 800 mm, respectively, nine cores, and provided by Bison SP1–SP4 were constructed using a relatively high-strength in situ
Precast Limited (Swadlincote, UK). They had a compressive cube concrete mix, with cube strength ranging from 38 to 68 N=mm2 , in
strength equal to 55 N=mm2 . Five cores were left open at the top combination with a relatively low YP yield strength. The YP
for the placement of transverse reinforcement, i.e., Cores 1, 3, 5, 7, strength gradually increased from Group SP1 to SP4. FYP accord-
and 9 shown in the inset of Fig. 6(b). The edge cutout in each HCU ing to Eq. (5) had values ranging from 150 kN in Group SP1 to
was 300 mm long (in the direction parallel to the steel section) and 263 kN in Group SP4. Specimens SP5 and SP6 were constructed
70 mm wide. Polythene foams were placed around the YP to form using a normal-strength in situ concrete mix, with cube strengths
the concrete tooth before pouring the in situ concrete. The five open equal to 30 and 35 N=mm2 in SP5 and SP6, respectively, and
cores were filled with in situ concrete after the 12 mm-diameter stronger YPs. FYP was 345 kN in Group SP5 and 310 kN in Group
rebars were placed inside the cores. SP6 according to Eq. (5). The ratio FYP =Fconc is given in Table 1.
Based on Eq. (6) and one trial test, it was concluded that the core This ratio was low in Specimens SP1–SP4, assuming values in the
located just before the YP should be filled with concrete as well, range 0.34–0.73, and high in Specimens SP5 and SP6, with values
because otherwise the slab would fail prematurely due to the large of 1.17 and 1.11, respectively. Thus, failure in Specimens SP1–SP4
stress field concentrated in that region before an effective shear was expected to occur due to YP yielding, whereas in Specimens
transfer took place. This result was confirmed using the FEM model, SP5 and SP6 the concrete slab was expected to be involved in the
as will be described later. For this reason, the Core 4s in Fig. 6 was failure modes, considering also the hardening behavior of the YP
that would actually produce ultimate strengths higher than FYP .
left open at the top and filled with in situ concrete in all specimens.
Each YP was manufactured from a square SHS180 × 180 × 8 or
SHS180 × 180 × 10 hollow tube of S355 steel grade. The steel Material Property Tests
plate welded at the bottom of a YP had dimensions 200 × 200 mm, Both concrete and steel material tests were performed prior to test-
was 20 mm thick, and was made of S275 steel grade for all spec- ing. Concrete tests were conducted on the in situ concrete mix of
imens. The depth of the horizontally slotted hole of the YP was each specimen. They consisted of standard cube compressive tests
designed to allow the placement of a 12-mm-diameter rebar with and splitting tensile tests (BSI 2009). The latter aimed to determine
minimum tolerance. the tensile strength of the concrete. Three cube compressive tests
The parameters that varied in the tests were the YP geometry were carried out on the same day as the corresponding push
and the in situ concrete strength. A summary of the test matrix test. The concrete cubes were cast on the same day as the corre-
is provided in Table 1. The specimens are designated by the prefix sponding specimen using the same concrete mix. The cubes were
SP followed by a number and a letter. The number indicates a spe- 100 × 100 × 100 mm, while the splitting tensile tests were per-
cific YP geometry group, and the letters indicate different tests with formed on 100 × 200 mm standard cylinders as per BSI (2009).
identical YP geometries. When there is no letter, only one test was Table 2 summarizes the average compressive concrete strength
conducted with the specific YP geometry. Note that the in situ of three cubes and the average splitting tensile strength of two cyl-
concrete cube strength was different in all tests. inders on the test day. The maximum aggregate size used for the

© ASCE 04019076-5 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 Table 1. Test matrix
YP geometry (mm)
Concrete cube
Specimen t w h strength (N=mm2 ) FYP [kN, Eq. (5)] Fconc [kN, Eq. (6)] FYP =Fconc
SP1A 8 17 75 61 150 440 0.34
SP1B 8 17 75 52 150 388 0.39
SP1C 8 17 75 50 150 370 0.41
SP2A 8 13 55 47 186 370 0.50
SP2B 8 13 55 56 186 434 0.43
SP3 8 17 55 38 215 333 0.65
SP4A 10 14 75 68 263 440 0.6
SP4B 10 14 75 47 263 360 0.73
SP5 10 18 65 35 345 295 1.17
SP6 10 14 65 30 310 280 1.11
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

Table 2. In situ concrete material properties LVDTs connected a point of the bottom steel section flange to a
Age at testing Compressive Splitting tensile point of the bottom side of the HCU. Two more LVDTs were used
Specimen (days) strength (N=mm2 ) strength (N=mm2 ) in SP1A and SP6 to measure the differential horizontal displace-
ment between the top and bottom flanges of the steel section
SP1A 27 61 4.3
and, thus, to check whether significant second-order bending
SP1B 17 52 3.1
SP1C 11 50 2.9
was introduced in the beam due to the eccentric application of
SP2A 7 47 2.9 the loading with respect to the beam’s centroid in the specific test
SP2B 17 56 3.6 setup. The horizontal load was measured by the built-in load cell of
SP3 12 38 2.5 the hydraulic actuator. Strain gauges were installed on the steel
SP4A 11 68 4.5 strips and vertical walls of the YP and on the concrete slab around
SP4B 35 47 2.80 the YP to monitor the development of strains. Four strain gauges
SP5 10 35 2.1 were also attached to the transverse rebars of Specimen SP6. Fig. 7
SP6 12 30 1.9 shows the instrumentation used for Specimens SP1A and SP6. In
the rest of the specimens, only the relative horizontal slip in the
HCU–steel section interface, the strains developed at the five points
on the concrete slab shown in Fig. 7, and the strains developed in
concrete mix was 10 mm. A medium- to high-workability concrete
the YP were measured.
was produced using a polycarboxylate high-range water-reducing
The load was initially applied under force control up to 40%
admixture.
of the expected yield load and then cycled 25 times between
The material properties of the YP were obtained from standard
5% and 40% of the expected yield load. The rate of loading during
tensile coupon tests (ASTM 2011) taken from the flat and corner
the cyclic loading was around 1 kN=s. After the cyclic loading,
regions of the SHS180 × 180 × 8 and SHS180 × 180 × 10 tubes
the load was applied under displacement control at a rate of
used to manufacture the YPs. The properties obtained are the
0.2 mm=min up to failure.
modulus of elasticity, the 0.2% proof stress, the ultimate stress,
and the tensile strain corresponding to the ultimate stress. The re-
sults are given in Table 3 as average values of three coupon tests. Experimental Results

Instrumentation and Loading Procedure Failure Modes and Observations

A number of linear variable differential transformers (LVDTs) and Table 4 summarizes the failure modes of the specimens. Specimens
strain gauges were used to monitor the experimental behavior of the in Groups SP1–SP4 (low YP strength—high-strength in situ con-
push specimens. The relative slip between the steel section and crete mix) failed in a ductile mode with plastic deformations con-
the HCUs was measured using two LVDTs horizontally placed near centrated in the YPs, Specimen SP5 failed due to concrete slab
the YP, on each side of the steel section. The base of each of these cracking without significant deformations in the YP, and Specimen
LVDTs was attached to the bottom side of the HCU and the tip on SP6 failed in a mixed mode that included both the YP plastic bend-
the top flange of the steel section. The vertical separation between ing and the failure of the concrete slab at later stages of the loading.
the slab and the steel section, or uplift, was measured during the Specimen SP1A was loaded until all possible failure modes
tests of Specimens SP1A and SP6 using two LVDTs vertically were met. The response was ductile due to plastic bending of the
placed near the YP on each side of the steel section. Each of these YP up to a slip of 47 mm corresponding to a force of 357 kN.

Table 3. YP’s steel material properties (average values of three coupon tests)
Modulus of elasticity 0.2% proof stress Tensile stress Strain at
Coupon region (N=mm2 ) (N=mm2 ) (N=mm2 ) ultimate stress
Flat region SHS180 × 8 193,600 390 460 0.168
Corner region SHS180 × 8 221,000 390 425 0.02
Flat region SHS180 × 10 202,850 500 565 0.087
Corner region SHS180 × 10 215,500 525 590 0.028

© ASCE 04019076-6 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 2 1
Spreader beam Push specimen
3
3 4
4
5
12

Steel beam Slip and uplift

2 3 recordings recordings
(both sides)
1 4
Concrete strain gauges LVDTs
Steel strain gauges
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Instrumentation on Specimens SP1A and SP6.

Table 4. Experimental strength and failure modes

Slip at maximum
Specimen Fy;exp (kN) Fmax;exp (kN) force (mm) Fy;exp =FYP Fmax;exp =Fy;exp Failure mode
SP1A 170 357 47 1.13 2.10 Ductile
SP1B 170 325 34 1.13 1.91 Ductile
SP1C 160 250 30 1.07 2.23 Ductile
SP2A 200 356 35 1.08 1.78 Ductile
SP2B 200 339 35 1.08 1.79 Ductile
SP3 230 359 36 1.07 1.56 Ductile
SP4A 260 378 33 0.99 1.37 Ductile
SP4B 260 375 33 0.99 1.44 Ductile
SP5 N/A 321 8.3 N/A N/A Slab shear failure
SP6 320 427 19.6 1.03 1.33 Mixed

The plastic deformations were concentrated in the YP’s vertical happened at a slip of 58 mm and is shown in Fig. 8(c). Although the
strips and walls and no signs of cracks in the slab were observed slab ultimately failed in a brittle manner due to the sudden trans-
up to that point. Fig. 8(a) shows the deformed YP at the end of the verse cracking, the response of Specimen SP1A is characterized as
test; due to the very large inelastic deformations, cracks formed at ductile because the main source of plasticity up to an excessive slip
the ends of the vertical strips at the positions of the plastic hinges, as of 47 mm was the plastic bending of the YP. Note that slip demands
indicated by an arrow in the same figure. At a slip of 47 mm, the of more than 20–22 mm can be met in composite beams used in
deformed transverse wall of the YP came into contact with the buildings, for example, in design cases where long spans and partial
row of bolts close to the loaded side of the specimen, as shown shear connection are used (Zona and Ranzi 2014). The proposed
in Fig. 8(b). As the specimen was further loaded, the slab units connector with the large ductile slip capacity can be safely used
started moving upward and sliding over the YP due to the large in such cases where force redistribution is needed.
inclination of the YP’s walls. This new deformed position created Specimen SP5, which had a combination of high-strength YP
a secondary hogging moment, which finally led to the flexural fail- and normal-strength in situ concrete, failed due to concrete
ure of the slab by the sudden formation of a large crack through the cracking in the pattern shown in Fig. 9(a). A transverse crack
width of the slab units at a position just before the YP. This failure started to form on the surface of the slab at a load of 130 kN

Fig. 8. Failure modes of Specimen SP1A: (a) deformed YP after test with developed plastic mechanism and signs of crack initiation (arrow); (b) walls
in contact with first row of bolts; and (c) ultimate failure mode of slab at slip of 56 mm.

© ASCE 04019076-7 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Failure mode of Specimen SP5: (a) cracking pattern of slab; and (b) YP after test without evidence of significant inelastic deformation.

and corresponding slip of 0.9 mm. The crack was formed in the Specimen SP6 failed in a mixed mode. Up to 19 mm of slip, the
interface between the fourth core of the HCU and the in situ con- response was ductile due to YP plastic bending, while there were no
crete on one side of the YP. A second transverse crack was formed signs of cracks on the slab. At a force equal to 427 kN and corre-
on the surface of the specimen between the fifth core of the HCU sponding slip of 19.6 mm, diagonal shear cracks started to form.
and the in situ concrete at a load of 200 kN and corresponding slip The cracks were symmetric with respect to the beam’s centerline.
of 2.2 mm. Shear diagonal cracks started forming at the corners of Further loading of the specimen caused the force to drop gradually
the YP at a load of 314 kN and 7.2 mm of slip, propagating diago- to 380 kN, and the test was stopped. Fig. 10(a) shows the failure
nally toward the spreader beam at an angle of approximately 45° to mode of the slab at the end of the tests, and Fig. 10(b) shows the
the beam axis. Further increase of the load caused more shear deformed YP after the test. The YP plastically deformed but much
cracks to form, and the specimen ultimately failed at 321 kN and less than in test SP1A [Fig. 8(a)].
corresponding slip of 8.3 mm, followed by a gradual drop in the In Specimen SP1C, an inadequate amount of in situ concrete
load to 286 kN, at which point the test was stopped. After disman- was placed adjacent to the sides of the YP that were parallel to
tling the specimen, very little deformation was observed in the YP. the beam axis. This resulted in the out-of-plane bending of the
Fig. 9(b) shows the YP at the end of the test. Note that, although the device and the inability to develop the full plastic strength accord-
shear failure in this specimen was predicted by Eq. (6) (since ing to Eq. (5). The deformed shape of the YP in this test is shown
FYP =Fconc ¼ 1.17 in Table 1), the transverse (ripping) cracks were in Fig. 11.
unexpected, and they are attributed to eccentricities that were in- A common secondary failure mode observed in all tests is the
troduced in the application of the loading due to the imperfections local crushing of the bottom part of the concrete tooth due to
on the surface of the specific HCUs used in this test and, in turn, the concentration of compressive stresses in that region as the
resulted in secondary moments and torsional effects, which were YP deforms. This local failure is shown in Fig. 12.
extremely difficult to quantify. The diagonal cracks were not sym-
metric with respect to the beam’s centerline, as can be seen from
Experimental versus Theoretical Strength
Fig. 9(a), due to torsional eccentricity during the test. The torsional
eccentricity was recorded by the LVDTs used to measure the hori- Table 4 summarizes the experimental yield (Fy;exp ) and maximum
zontal slip and monitored higher slip values on the side where (Fmax;exp ) strengths achieved in each specimen; the interface slip
cracks were more prominent. corresponding to maximum force; the ratio of the experimental

Fig. 10. Failure mode of Specimen SP6: (a) diagonal shear cracking of slab; and (b) deformed YP after test.

© ASCE 04019076-8 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 recorded by the two LVDTs on the right- and left-hand sides of
the YP. The force corresponds to the feedback output given by
the load cell of the actuator. Fig. 13 shows the response of spec-
imens in the SP1 group. The sudden increase of the force in SP1A
that occurs at 47 mm corresponds to the bearing force that was de-
veloped after the YP came into contact with the bolts (as described
in the previous section). As can be seen from the graph, the force
increases suddenly from 357 to 520 kN. The maximum force cor-
responds to a slip of 52 mm. The response after this point starts to
soften due to the gradual failure of the slab, and at a slip of 56 mm
the curve drops suddenly due to the complete failure of the slab
[Fig. 8(c)].
Following the test of SP1A that was brought to complete failure,
the remaining specimens that were designed to fail in a ductile
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

mode, i.e., the specimens in Groups SP1–SP4, were tested up to

approximately 30 mm of slip. At this slip value the tests were
Fig. 11. Out-of-plane bending of YP in Specimen SP1C due to inade- stopped for the following reasons: (1) in all of those tests, there
quate lateral restraint. were no signs of cracks in the slab while the YP was plastically
deforming, and therefore it was assumed that the failure mode
would be similar to the one observed in SP1A if the slab was further
pushed against the steel section; (2) to avoid damaging the steel
to theoretical yield strength of the YP, Fy;exp =FYP ; and the ratio sections and slab units so that they could be reused in subsequent
of the maximum experimental force to the experimental yield tests and assess the deconstructability of the system; and (3) because
strength, Fmax;exp =Fy;exp . In Specimen SP1A, the maximum force 30 mm of slip is already a much higher value of slip than that
and corresponding slip are taken to be those recorded just before expected in typical composite beams.
the bearing of the YP walls on the bolts, i.e., 357 kN and 47 mm, As can be seen from the force-slip graphs in Fig. 13, Specimen
respectively. The yield force of Specimen SP5 is not defined be- SP1B follows exactly the same force-slip path as SP1A up to
cause there was no significant deformation of the YP in that test. 34 mm of slip. Specimen SP1C has a significantly reduced initial
The Fy;exp =FYP ratio has a mean value of 1.06 with standard stiffness and force compared to the other two tests of the series
deviation 0.05, which indicates that Eq. (5) based on simple me- due to the inadequate amount of in situ concrete placed around
chanics predicted the yield force of the YP in the specimens with no the YP that resulted in its out-of-plane bending, as described in
early failure of the concrete slab with good accuracy. Based on the
the previous section. The force is reduced by 40 kN, and the initial
failure modes of the specimens, it is concluded that the FYP =Fconc
stiffness is reduced by 20%. This behavior is both undesirable and
ratio, reported in Table 1, which was used for the initial design of
unpredictable, and therefore the YP should be fully restrained by in
the specimens, predicted accurately the slab failure in Specimens
situ concrete.
SP5 and SP6. These results will be generalized using the FEM
Specimen SP2A was deconstructed after the test by cutting and
model described later in the paper.
removing the in situ concrete surrounding the YP, and both the steel
The maximum experimental strength is higher than the
sections and the slab units were reused for the test on SP2B with
yield strength due to the hardening behavior of the YP. The
Fmax;exp =Fy;exp ratio is approximately 2.0 in the SP1 group, 1.8 new in situ concrete cast around the YP. Fig. 14 shows that the
in the SP2 group, 1.6 in Specimen SP3, 1.4 in the SP4 group, original and reused specimens have equal stiffness and a slight
and 1.33 in Specimen SP6, indicating the considerable hardening difference in strength, i.e., SP2B is approximately 10 kN weaker
behavior of the YP. This hardening behavior of the proposed con- than SP2A. The test of Specimen SP3 was a standalone one, while
nector is different than the typical force-slip curve of a welded stud, Fig. 14 shows that Specimens SP4A and SP4B had identical
which has a well-defined plateau before the ultimate failure. responses.
Fig. 15 shows the force-slip responses of Specimens SP5 and 6.
It can be seen that, despite the higher YP strength of Specimen SP5
Force-Slip Responses (345 kN), Specimen SP6, with FYP ¼ 310 kN, achieved a higher
Figs. 13–15 show the force versus slip responses of the specimens. stiffness and strength. The reason for this is the early concrete
The slip history plotted in these graphs is the average value failure in SP5, for the reasons explained in the previous section.

Local crushing of concrete

F,

Fig. 12. Local concrete crushing at bottom of concrete tooth at large imposed displacement, observed in all specimens.

© ASCE 04019076-9 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 200
SP3
150 Welded stud

Force (kN)
100

50

0
0 0.5 1 1.5 2
Slip (mm)

Fig. 16. Comparison of initial stiffness between Specimen SP3 and a

welded stud tested in Lam et al. (1998).
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Force versus average slip responses of specimens in SP1

group.
fully consistent, since the welded stud was tested in a 150-mm-
thick HCU and the in situ concrete had different properties in
400
the study of Lam et al. (1998), it shows in a qualitative way the
significantly higher strength and slip capacity that is achieved using
the YP. In Fig. 16, the initial stiffness of Specimen SP3, including
300 the initial 25 cycles between 5% and 40% of the expected yield
SP2A strength, is compared with the same welded shear stud in HCUs
Force (kN)

SP2B
(Lam et al. 1998), and it demonstrates that the initial stiffness of
200 SP3
SP4A the YP is comparable to that of a welded stud.
SP4B The LVDTs measuring the uplift showed that the vertical sep-
100 Welded stud aration between the slab and the steel section in Specimens SP1A
and SP6 was at any given time much less than half of the horizontal
slip, and thus it is considered acceptable as per Eurocode 4
0 (BSI 2004).
0 10 20 30 40
Slip (mm)
Strain Recordings
Fig. 14. Force versus average slip responses of specimens in SP2, SP3,
The microstrain (με) values recorded by the strain gauge attached at
and SP4 groups, and force-slip response of welded stud tested in Lam
the midwidth of the concrete tooth, i.e., Gauge 3 in Fig. 7, indicate
et al. (1998).
almost zero strain. This indicates that the compressive stress on the
face of the YP is not significant and can be explained considering
the fact that the YP deforms in such a way that the top side of the
YP and the top fibers of the concrete tooth disconnect from each
other. The strain gauges placed closer to the corners of the YP,
i.e., Gauges 2 and 4, recorded small compressive strains. This in-
dicates a compressive stress field developed from the introduction
of the load from the spreader beam to the YP. The recordings of the
strain gauges installed closer to the sides of the YP, i.e., Gauges 1
and 5, indicate larger tensile strains up to 3,000 με. This justifies
the shear crack pattern observed in Specimen SP6.
Fig. 17(a) shows the force versus microstrain graphs as recorded
by the strain gauges attached on the steel rebars of Specimen SP6.
The results show that the strain of the rebars started increasing after
the load reached approximately 350 kN and corresponding relative
slip of 6.5 mm. The recorded strains increased rapidly from below
Fig. 15. Force versus average slip responses of SP5 and SP6 70 μe to 400 με as the load approached the failure load of the slab
specimens. due to shear failure at 400 kN applied load. The strains developed
in the rebars correspond to a stress of approximately 100 N=mm2,
while the yield stress of a rebar is 500 N=mm2 . This result indicates
that the transverse rebars contributed to the shear transfer mecha-
The slip at a maximum force of SP5 is 8.3 mm, which is the small- nism, although they did not yield.
est of all tests, but still higher than the 6 mm of the minimum slip The results from the strain gauges attached to the YP’s steel
requirement in Eurocode 4. In SP6, the response starts plateauing at strips and walls are shown in Fig. 17(b). The yield strain of the
a slip of 19.6 mm, where the first visible diagonal crack on the slab YP material is approximately 1,950 με, based on the material test
was formed. results of Table 3 for the flat regions. Large strains up to 6,000 με
In Fig. 14, a comparison is made between the force-slip behav- developed in the top of the YP’s steel strips and walls for Specimen
ior of the proposed connector and that of a welded shear stud in SP1A. The YP of Specimen SP5 appeared to have limited defor-
HCUs tested in Lam et al. (1998). Although the comparison is not mation at the end of the test. The recordings of the strain gauges

© ASCE 04019076-10 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 17. Force versus microstrain graphs from strain gauges: (a) attached to rebars of Specimen SP6; and (b) on walls of YP in Specimen SP1A.

installed on this YP indicated minor yielding of the steel strips and components. Tie constraints were imposed between the HCUs
onset of yielding at the transverse wall. However, these results are and the in situ concrete filled cores. The transverse rebars were em-
sensitive to the precise position of the strain gauges along the width bedded in the in situ concrete cores of the HCUs, and therefore a
of the steel strip and wall. perfect bond between the rebars and the surrounding concrete was
assumed in the simulations.

Finite-Element Model Material Modeling

The material model for steel was defined using an elastic-plastic
Model Description
constitutive law. The stress-strain relationship of the YP’s steel
The finite-element analysis program Abaqus (Dassault Systémes material followed the uniaxial tensile coupon tests, which can
2014) was used to perform numerical simulations of push tests. be simplified as bilinear curves. A bilinear elastic–perfectly plastic
Due to symmetric geometry, only half of a push test specimen stress-strain relationship was used for the rebars, bolts, and base
was built, as shown in Fig. 18. The concrete slab, YP, bolts, plate of the YP using nominal material values.
and steel section were modeled using the eight-node reduced- The uniaxial compressive stress-strain behavior of concrete was
integration brick finite-element C3D8R. The transverse rebars were defined according to the relationship provided in fib (2010), includ-
modeled using the two-node linear truss element T3D2. The mesh ing the descending (softening) branch after the peak compressive
of the YP and the concrete slab around the YP was refined to give strength, fc0 , was reached. For uncracked concrete subject to ten-
accurate predictions of the failure modes and force-slip responses. sion, a linear stress-strain relationship was adopted when the stress
Fully fixed boundary conditions were assigned to the bottom of the is less than the tensile stress, f t ¼ 1.4ðf c0 =10Þ2=3 . The Young’s
steel beam. modulus in tension was assumed to be the same as that in compres-
Displacement was imposed on the “loading” surface shown sion. The postfailure behavior for direct straining across cracks was
in Fig. 18. The displacement was applied slowly by a smooth am- modeled using the tension stiffening option and determining a lin-
plitude function, which allows arbitrary time variations of dis- ear reduction until stress is zero at a strain value of 10 times the
placement to be given throughout the analysis. The solution was strain at f t . The damaged plasticity model was used to define
obtained using the dynamic explicit method since it is very efficient the plastic behavior of concrete (Dassault Systémes 2014). The
at modeling complex nonlinear problems. Hard contact and penalty parameters of the damaged plasticity model were set to their rec-
friction formulations were used to describe the normal and the tan- ommended values, i.e., dilation angle ψ ¼ 30°, flow potential
gential behavior of concrete interfaces in contact with the YP and of eccentricity ε ¼ 0.1, viscosity parameter μ ¼ 0, ratio of biaxial
the bolts in contact with the welded plate and the flange of the steel to uniaxial compressive strength σb0 =σc0 ¼ 1.16, and ratio of
section. The friction coefficient was chosen as 0.4 in the contact second stress invariant on the tensile meridian to that on the com-
boundary between steel and concrete and 0.25 between steel pressive meridian K ¼ 2=3.

HCU
In situ
concrete
Steel beam

Loading
Uy=0,URx,URy=0
Z Fixed
X YP
Bolt
Y Base plate

Fig. 18. Finite-element model.

© ASCE 04019076-11 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 400 500

400
300

Force (kN)
Force (kN)
300
200
Test FEM 200
SP1A Test FEM
100 SP2A SP5
100
SP3 SP6
SP4A
0 0
0 10 20 30 40 0 10 20 30 40
(a) Slip (mm) (b) Slip (mm)
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 19. Comparison of force versus slip responses between tests and numerical simulations.

Table 5. Comparison of experimental and numerical results

Yield strength (kN) Ultimate strength (kN) Slip at maximum force (mm)
Specimen Test FEM Test/FEM Test FEM Test/FEM Test FEM Test/FEM
SP1A 170 185 0.92 357 350 0.98 47 45 0.96
SP2A 200 190 1.05 356 341 0.96 35 36 1.03
SP3 230 225 1.02 359 357 0.99 36 37 1.03
SP4A 260 267 0.97 378 385 1.02 33 35 1.06
SP5 N/A N/A N/A 321 350 1.09 8.3 9 1.08
SP6 320 300 1.07 427 417 0.98 19.6 21 1.07
Mean 1.0 1.01 1.04
Standard deviation 0.05 0.05 0.05

Fig. 20. Plastic strain distribution in YPs of specimens: (a) SP1a; (b) SP5; and (c) SP6.

FEM Model Validation with a standard deviation of 0.05. The maximum difference be-
tween the experimental and numerical maximum strength is 9%
The force versus slip curves resulting from the numerical analyses
of the push specimens are plotted together with the experimental for Specimen SP5, and it is attributed to eccentricities introduced
curves in Fig. 19. Comparison of the results shows that the exper- in the application of the loading due to the imperfections on the
imentally obtained strength and stiffness of the shear connection surface of the specific HCUs (as described in the previous section)
can be predicted by the numerical solutions with good accuracy. and in turn resulted in secondary moments and torsional effects,
Note that the analysis of Specimen SP1A was stopped at a slip which were extremely difficult to simulate in the FEM analyses.
of 47 mm, i.e., when the transverse wall of the YP came into contact Thus, the proposed finite-element model can reasonably predict
with the row of bolts close to the loaded side of the specimen. the yield strength, ultimate strength, and relative slip of the YPs.
Table 5 summarizes the yield strength, maximum strength, and Fig. 20 shows contour plots of the equivalent plastic strain
slip at maximum force obtained from the tests and numerical analy- (PEEQ) on the deformed shape of the YPs at the end of the analyses
ses. The numerical to experimental yield and maximum strength of Specimens SP1A, SP5, and SP6. Plastic deformation concentra-
ratios have mean values of 1.0 and 1.01 and standard deviations tion occurs at the ends of the steel strips and at the ends of the YP’s
of 0.05 and 0.05, respectively, while the ratio of numerical to ex- walls for SP1A; the base plate also developed some plastic defor-
perimental relative slip at maximum force has a mean value of 1.04 mation. Minimum plastic deformation was observed in the YP of

© ASCE 04019076-12 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 Fig. 21. Tensile strain distribution in slabs of specimens: (a) SP1a; (b) SP5; and (c) SP6, at failure (darker regions: larger plastic strain).
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

400 500

400
300
Force (kN)

Force (kN)
300
200
W-4s W/O-4s
200
SP1
SP2 W-4s W/O-4s
100
SP3 100 SP5
SP4 SP6
0 0
0 10 20 30 40 50 0 10 20 30 40
(a) Slip(mm) (b) Slip (mm)

Fig. 22. Force-slip responses of specimens with or without in situ concrete in core before YP.

Specimen SP5, and plastic deformation occurred for Specimen SP6

but much less than in the test on Specimen SP1A. These observa-
tions are consistent with the experimental evidence and the strain
recordings of the gauges placed on the YPs.
Fig. 21 shows the principal tensile strain distribution in the slabs
of Specimens SP1A, SP5, and SP6. In Specimen SP1A, no plastic
strains were observed in the concrete slab until a slip of 47 mm was
reached, indicating that the concrete slab did not fail up to that
point, as observed in the corresponding push test. The relative mag-
nitude and orientation of the plastic strains of SP5 and SP6 indicate
that the slab failure zone in the model is consistent with the con-
crete failure modes observed during the tests, i.e., in the corners of
the YPs toward the spreader beam.
From the foregoing comparisons between the numerical and ex-
perimental results, it is concluded that the FEM model is reliable
and can be used for further parametric studies.

Parametric Study

Effect of In Situ Concrete in Core before YP

To investigate the effect of in situ concrete in the core just before the
YP, i.e., Core 4s in Fig. 6, on the strength and failure modes of
the specimens, the in situ concrete was removed from Core 4s and
the analyses rerun, keeping the other parameters identical.
Fig. 22 shows a comparison of force-slip curves for all test spec-
Fig. 23. Effect of in situ concrete in core before YP on failure modes
imens with or without in situ concrete in Core 4s, denoted W-4s and
of slab in Specimens SP4 and SP5: (a) SP4 with concrete in core 4s;
W/O-4s, respectively. The in situ concrete in Core 4s does not affect
(b) SP4 without concrete in core 4s; (c) SP5 with concrete in core 4s; and
the stiffness of the shear connection in the elastic stage; however,
(d) SP5 without concrete in core 4s (darker areas: larger plastic strain).
it improves the strength and ductility of the shear connection.

© ASCE 04019076-13 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 500 500
SP1 SP2
fc=25MPa fc=50MPa fc=25MPa fc=50MPa
400 fc=30MPa fc=60MPa 400 fc=30MPa fc=60MPa
fc=40MPa fc=40MPa

Force (kN)

Force (kN)
300 300
350

200 200
300 350

100 250
100
300
20 30 40 50 30 40 50
0 0
0 10 20 30 40 50 0 10 20 30 40 50
(a) Slip (mm) (b) Slip (mm)

500 500 SP4

SP3
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

fc=25MPa fc=50MPa fc=25MPa fc=30MPa fc=40MPa

400 fc=30MPa fc=60MPa 400 fc=50MPa fc=60MPa
fc=40MPa
300 300
Force (kN)

Force (kN)
400
400
200 200
350

100 100
350
300
30 40 50 30 40 50 60
0 0
0 10 20 30 40 50 0 10 20 30 40 50 60
(c) Slip (mm) (d) Slip (mm)

SP6
SP5
fc=25MPa fc=30MPa fc=40MPa
500 fc=25MPa fc=30MPa fc=40MPa 500
fc=50MPa fc=60MPa
fc=50MPa fc=60MPa
400 400
Force (kN)

Force (kN)

300 450 300 450

200 400
200 400

350
100 350 100
300
10 20 30 10 20 30
0 0
0 10 20 30 40 0 10 20 30 40
(e) Slip (mm) (f) Slip (mm)

Fig. 24. Effect of in situ concrete strength on force-slip responses of specimens.

The ratios of ultimate strengths of Specimens SP1–SP6 with in situ at the local area of the “concrete tooth” between the YP and Core 4s
concrete in Core 4s to those without are 1.10, 1.21, 1.29, 1.04, 1.33, when the in situ concrete was removed from Core 4s.
and 1.12, respectively. From the graphs of Fig. 22(a) (SP1–SP4), it
can be seen that there is significant improvement in the postelastic
response of specimens, and particularly in Specimens SP3 (with in Effect of In Situ Concrete Strength
situ concrete strength 38 N=mm2 ) and SP2 (with in situ concrete To investigate the effect of in situ concrete strength on the response
strength 47 N=mm2 ). Notably, Specimen SP3 without in situ con- of the proposed demountable shear connector, the compressive
crete in Core 4s failed immediately after the end of the elastic stage, strength of in situ concrete in the cores was varied from 25 to
at a slip of around 6 mm, showing no ductile behavior. The effect is 60 N=mm2 , i.e., fc ¼ 25, 30, 40, 50, and 60 N=mm2 for all the
also significant in Specimens SP5 and SP6, as shown in Fig. 22(b), specimens (SP1–SP6). Except for the in situ concrete strength,
with SP5 showing a much weaker and less ductile response when all other parameters were the same as with the tested specimens.
there is no in situ concrete in Core 4s. Fig. 24 shows the force versus slip responses of all specimens
The placement of in situ concrete in the core before the YP af- with different in situ concrete strengths. The in situ concrete
fected the failure modes of the specimens. Fig. 23 shows the failure strength does not affect the initial stiffness of the shear connection;
modes of Specimens SP4 and SP5 with or without in situ concrete however, it slightly affects the ultimate shear strength and signifi-
in Core 4s, in terms of maximum principal plastic strain distribu- cantly affects the ductility of the specimen. The curves in Fig. 24(a)
tion. In Specimen SP4, the failure mode changes from concrete indicate that Specimen SP1 always fails due to YP yielding,
shear failure to coupled shear and ripping failure when in situ con- i.e., without any type of concrete failure involved, regardless of
crete is removed from Core 4s. In Specimen SP5, the failure mode the concrete strength used. In Specimens SP2, SP3, and SP4
changes from shear failure of concrete slab to compressive crushing [Figs. 24(b–d)], concrete failure occurs earlier when f c ¼ 25 or

© ASCE 04019076-14 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 400 500

400
300

Force (kN)
Force (kN)
300
200
W/O-rebar W-rebar
200
SP1-25MPa
100 SP2-25MPa W/O-rebar W-rebar
SP3-25MPa 100 SP5
SP4-25MPa SP6
0 0
0 10 20 30 40 50 0 10 20 30 40
(a) Slip/mm (b) Slip (mm)

Fig. 25. Effect of additional rebar in core 4s.

Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

400 500
FYP =FY,Expor FY,FEM
Experiment

FConc,Exp or FConc,FEM (kN)

Mean value (FYP,mean )
FEA
FY,Exp or FY,FEM (kN)

Standard deviation 400

300 (FYP,mean

300

200
200 FConc =FConc,Expor FConc,FEM
Experiment Mean value (FConc,mean )
FEM Standard deviation(FConc,mean
100 100
100 200 300 400 100 200 300 400 500
(a) FYP (kN) (b) FConc (kN)

Fig. 26. Comparison of proposed design equations with experimental and numerical results.

30 N=mm2 , while for the remaining in situ concrete grades the Evaluation of Design Equations
responses are almost identical to the experimental response.
There is a much more significant effect of the in situ concrete The push tests and the FEM parametric analyses showed that two
strength on Specimens SP5 and SP6. As can be seen from main failure modes control the ultimate strength of the proposed
Figs. 24(e and f), increasing the in situ concrete strength delays the demountable shear connector: the ductile failure of the YP and
failure of concrete. For example, if in situ concrete with fc ¼ the failure of the concrete slab. The correlation of FYP from
60 N=mm2 were used, then both SP5 and SP6 would be able to reach Eq. (5) with the experimental and numerical yield strength
more than 30 mm of slip and more than 400 kN of ultimate strength. (FY;exp or FY;FEM ) is shown in Fig. 26(a). The mean value of
FYP =ðFYP;FEM or FYP;exp ) is 0.95 with a standard deviation
of 0.06. The correlation of Fconc from Eq. (6) with the experimental
Effect of Additional Transverse Rebar in
and numerical strength of specimens that failed due to shear crack-
Core before YP
ing in the slab (Fconc;exp or Fconc;FEM ) is shown in Fig. 26(b). The
The effect of placing an additional 12-mm-diameter transverse re- mean value of Fconc =ðFconc;exp or Fconc;FEM ) is 0.91 with stan-
bar in Core 4s was investigated. For this purpose, the specimens dard deviation 0.12. The preceding experimental and numerical
that experienced shear failure of the concrete slab, i.e., Specimens agreement with the design equations show that Eq. (5) can be used
SP2–SP4 with in situ concrete strength 25 N=mm2 , SP5 and SP6, to predict the yield strength of the YP with good accuracy, while
were reanalyzed with the additional rebar installed. Eq. (6) can be conservatively used as an upper bound for concrete
Fig. 25 plots the force versus slip responses of specimens with shear failure, since it slightly underestimates the concrete failure. In
or without additional rebar in Core 4s. The additional rebar con- addition, the condition FYP =Fconc < 1 should be used to ensure a
tributes slightly to the ultimate strength of the specimens, and it ductile failure mode and at least 6 mm of slip of the proposed
increases the slip capacity (and hence the ductility) of the shear demountable shear connector, provided that the cores before and
connection by delaying the concrete failure. For example, the slip after the YP are filled with in situ concrete.
at ultimate force for SP6 increased from 21 to 26.2 mm (about 25%
increase) when adding an additional 12-mm-diameter transverse
rebar in Core 4s. Conclusions
The FEM model showed that the additional rebar redistributed
the load to the cores. For example, the normal stresses in rebars A steel-yielding demountable shear connector, referred to as a
in Cores 1s and 3s (Fig. 6) for Specimen SP6 are 251 and yielding pocket or YP, was proposed for use in composite floors
135 N=mm2 , respectively, but reduce to 227 and 110 N=mm2 after with precast hollow core slab units (HCUs). The proposed connec-
adding the rebar in core 4s, while the normal stress in the additional tor was experimentally validated through ten full-scale push tests
rebar is 210 N=mm2 . that were conducted in a horizontal testing arrangement and using

© ASCE 04019076-15 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 sections extracted from a prototype composite floor. A numerical Brambilla, G., M. Lavagna, G. Vasdravellis, and C. A. Castiglioni. 2019.
model was calibrated to the experimental results and used to carry “Environmental benefits arising from demountable steel-concrete
out a number of parametric studies. Based on the findings of the composite floor systems in buildings.” Res. Conserv. Recycl.
experimental and numerical study, the following conclusions can 141 (Feb): 133–142. https://doi.org/10.1016/j.resconrec.2018.10.014.
BSI (British Standards Institution). 2002a. Eurocode 1: Actions on struc-
be drawn:
tures. Part 1-1: General actions—Densities, self—Weight, imposed
• A YP can provide an adequate shear connection between the loads for buildings. Brussels, Belgium: BSI.
steel section and the HCUs if properly designed. The experi- BSI (British Standards Institution). 2002b. Eurocode 2. Design of
mental results show that a YP can be designed to have high concrete structures. Part 1–1. General rules and rules for buildings.
strength, high initial stiffness, and very high slip capacity. EN 1992-1-1. Brussels, Belgium: BSI.
• The YPs provided longitudinal shear resistance in the range of BSI (British Standards Institution). 2004. Eurocode 4. Design of composite
170–320 kN in the tests reported herein. Specimens designed to steel and concrete structures. Part 1–1. General rules and rules for
fail in a ductile mode had at least a 30-mm slip capacity. Speci- buildings. EN 1994-1-1. Brussels, Belgium: BSI.
mens that failed due to concrete shear cracking still achieved BSI (British Standards Institution). 2009. Testing hardened concrete,
a slip capacity much higher than that required by Eurocode 4, Part 3: Compressive strength of test specimens. BS EN 12390-6.
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

i.e., higher than 6 mm. London: BSI.

• Two main failure modes control the strength and ductility of the Burka, L. P. 1993. “A hypertext history of multi-user dimensions.”
Accessed December 5, 1994. http://www.ccs.neu.edu.
proposed shear connection, i.e., the yielding of the YP walls and
Couchman, G. 2014. Design of composite beams using precast concrete
vertical strips and the shear failure of the slab units. slabs in accordance with Eurocode 4: Steel construction institute
• The cores of the HCUs before and after the YP should be filled publication P401. Ascot, UK: Berkshire.
with in situ concrete in order to avoid premature concrete failure Dai, X. H., D. Lam, and E. Saveri. 2015. “Effect of concrete strength and
and to ensure that the longitudinal shear force can be transferred stud collar size to shear capacity of demountable shear connectors.”
from the YP to the concrete component. J. Struct. Eng. 141 (11): 04015025. https://doi.org/10.1061/(ASCE)ST
• The strength of in situ concrete does not affect the stiffness .1943-541X.0001267.
and strength of shear connection, but it significantly affects its Dassault Systémes. 2014. ABAQUS documentation. Providence, RI:
ductility. Dassault Systémes Simulia.
• An additional transverse rebar in the core before a YP has a slight Densley Tingley, D., and B. Davison. 2011. “Design for deconstruction and
effect on the strength and slip capacity of shear connections. material reuse.” Proc. Inst. Civ. Eng. Energy 164 (4): 195–204.
• The YP should be adequately restrained by the in situ concrete fib (Fédération internationale du béton). 2010. “First complete draft.
Vol. 1. Comité Euro-International du Beton.” In Model code 2010,
on all sides, as inadequate restraint will cause out-of-plane
secretariat permanent. Lausanne, Switzerland: fib.
bending that results in lower strength and stiffness. Henderson, I., X. Zhu, B. Uy, and O. Mirza. 2015a. “Dynamic behaviour
• The deconstruction was assessed by reusing the steel section of steel-concrete com- posite beams with different types of shear con-
and HCUs of Specimen SP2A in Specimen SP2B. The disman- nectors. Part I: Experimental study.” Supplement, Eng. Struct. 103 (SC):
tling of the specimen was implemented in a lab environment, 298–307. https://doi.org/10.1016/j.engstruct.2015.08.035.
and the force-slip responses of the two specimens were within Henderson, I., X. Zhu, B. Uy, and O. Mirza. 2015b. “Dynamic behaviour
10% difference. of steel-concrete composite beams with different types of shear
• The strength of a YP can be accurately predicted using a plastic connectors. Part II: Modelling and comparison.” Supplement, Eng.
analysis of the device, and the strength of the concrete slab can Struct. 103 (SC): 308–317. https://doi.org/10.1016/j.engstruct.2015
be conservatively predicted using an equation based on the shear .08.033.
strength of a cracked plane, adopted from current codes of prac- Henderson, I., X. Zhu, B. Uy, and O. Mirza. 2017. “Dynamic behaviour of
steel-concrete composite beams retrofitted with various bolted shear
tice. In addition, the condition that the strength of the YP should
connectors.” Supplement, Eng. Struct. 131 (SC): 115–135. https://doi
be less than that of the shear strength of the concrete slab is a .org/10.1016/j.engstruct.2016.10.021.
reliable predictor of the ductile force-slip response of the shear IEA (International Energy Association). 2007. Tracking industrial energy
connector. efficiency and CO2 emissions. Paris: IEA.
Lam, D. 2007. “Capacities of headed stud shear connectors in composite
steel beams with precast hollow core slabs.” J. Constr. Steel Res. 63 (9):
Acknowledgments 1160–1174. https://doi.org/10.1016/j.jcsr.2006.11.012.
Lam, D., K. S. Elliot, and D. A. Nethercot. 1998. “Push-off tests on shear
The experiments in this study were funded by the Engineering studs with hollow-cored floor slabs.” Struct. Eng. 76 (9): 167–174.
and Physical Science Research Council of the UK, Grant Lam, D., and E. Saveri. 2012. “Shear capacity of demountable shear
EP/P004253/1, awarded to the second author. The authors would connectors.” In Proc., 10th Int. Conf. on Advances in Steel Concrete
like to thank Bison Precast for offering the hollow-core precast slab Composite and Hybrid Structures, 767–774. Singapore: Research
units used in the specimens. Publishing.
Liu, X., M. A. Bradford, and A. Ataei. 2017. “Flexural performance of
innovative sustainable composite steel-concrete beams.” Supplement,
References Eng. Struct. 130 (SC): 282–296. https://doi.org/10.1016/j.engstruct
.2016.10.009.
ASTM. 2011. Standard test methods for tension testing of metallic Liu, X., M. A. Bradford, and M. S. S. Lee. 2015. “Behavior of high-
materials. ASTM E8/E8M. West Conshohocken, PA: ASTM. strength friction-grip bolted shear connectors in sustainable composite
Ataei, A., M. A. Bradford, and X. Liu. 2016. “Experimental study of beams.” J. Struct. Eng. 141 (6): 04014149. https://doi.org/10.1061
composite beams having a precast geopolymer concrete slab and decon- /(ASCE)ST.1943-541X.0001090.
structable bolted shear connectors.” Eng. Struct. 114 (May): 1–13. Moynihan, M. C., and J. M. Allwood. 2014. “Viability and performance of
https://doi.org/10.1016/j.engstruct.2015.10.041. demountable composite connectors.” Supplement, J. Constr. Steel Res.
Ban, H., B. Uy, S. W. Pathirana, I. Henderson, O. Mirza, and X. Zhu. 2015. 99 (SC): 47–56. https://doi.org/10.1016/j.jcsr.2014.03.008.
“Time-dependent behaviour of composite beams with blind bolts under Oehlers, D. J., and M. A. Bradford. 1995. Composite steel and concrete
sustained loads.” J. Constr. Steel Res. 112 (Sep): 196–207. https://doi structural members: Fundamental Behviour. Amsterdam, Netherlands:
.org/10.1016/j.jcsr.2015.05.004. Elsevier.

© ASCE 04019076-16 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076

 Pathirana, S. W., B. Uy, O. Mirza, and X. Zhu. 2015. “Strengthening Rehman, N., D. Lam, X. Dai, and A. Ashour. 2016. “Experimental study
of existing composite steel—Concrete beams utilising bolted shear con- on demountable shear connectors in composite slabs with profiled
nectors and welded studs.” J. Constr. Steel Res. 114 (Nov): 417–430. decking.” Supplement, J. Constr. Steel Res. 122 (SC): 178–189. https://
https://doi.org/10.1016/j.jcsr.2015.09.006. doi.org/10.1016/j.jcsr.2016.03.021.
Pathirana, S. W., B. Uy, O. Mirza, and X. Zhu. 2016a. “Bolted and welded Suwaed, A. S. H., and T. L. Karavasilis. 2017. “Novel demountable
connectors for the rehabilitation of composite beams.” Supplement, shear connector for accelerated disassembly, repair, or replacement of
J. Constr. Steel Res. 125 (SC): 61–73. https://doi.org/10.1016/j.jcsr precast steel-concrete composite bridges.” J. Bridge Eng. 22 (9):
.2016.06.003. 04017052. https://doi.org/10.1061/(ASCE)BE.1943-5592.0001080.
Pathirana, S. W., B. Uy, O. Mirza, and X. Zhu. 2016b. “Flexural behaviour Webster, M., and D. Costello. 2005. “Designing structural systems for
of composite steel—Concrete beams utilising blind bolt shear connec- deconstruction: How to extend a new building’s useful life and prevent
tors.” Supplement, Eng. Struct. 114 (SC): 181–194. https://doi.org/10 it from going to waste when the end finally comes.” In Proc., Green
.1016/j.engstruct.2016.01.057. Building Conf. Atlanta.
Pavlovic, M., Z. Markovic, M. Veljkovic, and D. Bucevac. 2013. Zona, A., and G. Ranzi. 2014. “Shear connection slip demand in com-
“Bolted shear connectors vs. headed studs behaviour in push-out tests.” posite steel-concrete beams with solid slabs.” J. Constr. Steel Res.
Supplement, J. Constr. Steel Res. 88 (SC): 134–149. 102 (Nov): 266–281. https://doi.org/10.1016/j.jcsr.2014.07.018.
Downloaded from ascelibrary.org by Universitas Indonesia on 04/18/21. Copyright ASCE. For personal use only; all rights reserved.

© ASCE 04019076-17 J. Struct. Eng.

J. Struct. Eng., 2019, 145(8): 04019076