Professional Documents
Culture Documents
Abstract
Presented are the results from a finite element model of steel–concrete composite beams with deep decks and a comparison
with various analytical/design methods. Using a deck deeper than 80 mm are becoming popular with a desire for longer
spanning capability and lower concrete volume. However, there are no design rules in either American or European design
codes for using a deck deeper than 80 mm, as both codes limit the deck rib height to 75 and 85 mm, respectively for using
the stud’s capacity formula. Therefore, research is needed to establish the design stud capacity in beams with decks deeper
than 80 mm. After extensive validation, the 3-D FE model is used for a parametric study with tests having decks deeper than
80 mm. The parameters include rib geometries, studs’ layout and concrete slab reinforcements. The FE results showed that
stud capacity with narrow and deep decks (100–150 mm) is about 70% of the conventional decking (60–80 mm deep). The
stud capacities from the numerical results were compared to the predicted strengths from the design/theoretical models. While
the equations from the concrete pull-out failure mode by Johnson and Yuan (Proc Inst Civ Eng Struct Build 128(3):252–263,
1998) gave reasonable predictions with a coefficient of variation as 11%, both EC4 and ANSI/AISC rules provided inac-
curate and inconsistent predicted strengths. A generalised stud capacity formula should be developed in the design codes
for decks deeper than 80 mm.
Keywords Deep steel decking · Headed stud connector · Push test · Finite element modelling · Composite beams
13
Vol.:(0123456789)
Author's personal copy
International Journal of Steel Structures
13
Author's personal copy
International Journal of Steel Structures
both European and American design codes limit the sheet- profiled sheeting. This seminal work allowed the separa-
ing depth to 85 and 75 mm, respectively for using their stud tion between steel deck and concrete slab to be modelled,
capacity formula. The structural designers have no option resulting in accurate shear connector resistance predic-
but to rely on the manufacturer’s specified characteristic stud tions. Due to ineffectiveness of the European code in pre-
strength, developed through experiments with specific stud dicting the strengths, Shin et al. (2013) proposed a new
height and concrete strength in mind. Research is needed design equation using results from FE analysis to esti-
to establish the design stud capacity in composite beams mate the in-plane compressive strengths of HPS deck with
with decks higher than 80 mm. A generalised stud capac- U-shaped ribs. With another numerical study, Shin et al.
ity formula should be developed in the design codes for (2014) proposed new design equation to predict the in-
these deep decks. This study presents a useful insight into plane compressive strengths of HPS box girders with open
understanding the behaviour of composite beams with decks ribs. Rahnavard et al. (2015–2019) presented a variety of
deeper than 80 mm by numerical modelling and provide a methods for modelling concrete and steel interaction using
comparison with available theoretical/design models for the shear connector. In their research, the B31 element type
shear stud capacity. was used for shear connector modelling in composite shear
Using a composite deck deeper than 80 mm has some walls (Rahnavard and Hassanipour 2015; Rahnavard et al.
benefits, including 20–30% reduction in concrete volume 2016), composite connections (Rahnavard et al. 2017a, b),
and increase in the slab’s spanning capability to as much composite steel–concrete beam (Rahnavard et al. 2017a,
as 6 m (Kingspan 2011). However, lack of design guide- b), and buckling of restrained braces (Rahnavard et al.
lines for stud capacity remains the single most challenge 2018; Naghavi et al. 2019).
in the design of the composite beams with decks deeper Chen et al. (2015) investigated the behaviour of headed
than 80 mm. Research on the behaviour of composite beams studs at elevated temperatures through an experimental and
with deep decks is very limited. Rambo-Roddenberry (2002) numerical study. Nellinger et al. (2017) concluded that the
studied the behaviour and strength of welded stud connec- embedment depth of stud in the concrete above the sheeting
tors in various depths of profiled decking up to 150 mm. rib had significant influence on the load-slip behaviour. For
However, the shape of ribs of profiled decking, deeper than push tests with studs that satisfied the minimum embedment
80 mm was rectangular, which is neither common in com- depth of two diameters, which is required by the EN 1994-
posite construction, nor economical. Although Adany et al. 1-1 (2004), the double curvature deformation of stud and
(2013) carried out experimental research on 153 mm deep large slip capacity were observed. While single curvature
trapezoidal sheeting, the research solely focused on test- deformation of stud and small slip capacity were witnessed
ing bare steel decking, without concrete and shear studs, to for tests that did not satisfy the minimum stud embedment
determine its bending resistance. depth of 2 diameters. A FE model was developed by Katwal
Several numerical and experimental studies were con- et al. (2018) for full-scale composite beams with profiled
ducted previously to study the structural behaviour of steel sheeting. However, the profiled decking depth consid-
composite beams with different types and geometries of ered in all studies did not exceed 80 mm. A very limited
ribs. Qureshi et al. (2011a) were the first to numerically research is available to address the impact of steel decks
model the post-failure behaviour of composite beams with deeper than 80 mm on composite beams.
13
Author's personal copy
International Journal of Steel Structures
Lack of design rules and no past research on the behav- resistance, the change in the slab depth from 140 to 215 mm
iour of composite beams with steel decks deeper than 80 mm increased the shear resistance about 30%.
is a major knowledge gap. This paper aims to bridge this gap
by investigating various theoretical models for stud resist- 2.2 Finite Element Mesh and Type
ance and comparing them with a validated nonlinear 3-D
finite element model using ABAQUS. A parametric study is Figure 3 shows the half-symmetric FE model of the push
performed on composite beams with 100 and 146 mm deep test conducted by Smith and Couchman (2010) with sym-
decks, after extensive validation against experiments with metry along the centreline of the beam web. It consists of
steel decks ranging from 50 to 150 mm deep. The param- a steel beam, headed studs, profiled decking, concrete slab
eters include the rib geometries, headed studs’ layout, and and wire mesh reinforcement. In the beginning, three differ-
reinforcing the concrete slab with a unique grid steel bars. ent element mesh sizes were used, including 10 × 10 mm,
The results obtained from the FE analysis are then com- 15 × 15 mm, and 20 × 20 mm. Based on the best compromise
pared with the nominal strengths of headed stud predicted between the accuracy in results and computational time, the
by the EC4 (2004), ANSI/AISC (2016), and some analytical medium mesh of 15 × 15 mm was selected. After meshing
methods proposed by Johnson and Yuan (1998), and Konrad all parts, they were assembled together to produce the push
(2011). test model. For steel beam, headed stud and concrete part,
three-dimensional continuum eight-node reduced integra-
tion elements (C3D8R) were used with an aspect ratio of
(1:1). Where necessary, six-node wedge elements (C3D6R)
2 Finite Element Modelling were also used in some places, especially the stud part with
an aspect ratio of (3:5). Four-node doubly curved shell ele-
2.1 Summary of Previous Experimental ments with reduced integration (S4R) were used for the
Investigations profiled steel decking. Finally, two-node three-dimensional
truss elements (T3D2) were used to model the wire-mesh
ABAQUS/Explicit is used to prepare the nonlinear three- reinforcement.
dimensional finite element model. This method is very
effective in terms of modelling complex contact interac- 2.3 Steel Material Properties
tions, and material and geometric nonlinearity. Two previ-
ous experimental studies from Rambo-Roddenberry (2002) The stud connectors and steel beam were modelled as a sin-
and Smith and Couchman (2010) are chosen for validation. gle geometry but with different material properties. All steel
The selected push tests from the first experiment consisted components were treated elastic perfectly plastic, except
of favourable and unfavourable headed studs welded through steel beam, which was treated fully elastic as the steel beam
trapezoidal decks of 50 and 76.2 mm deep each. Also, cen- has no effect on the behaviour of headed studs during the
tral headed studs were welded through a rectangular deck of test. Young’s modulus of elasticity (Es) for the steel beam
114.3 and 152.4 mm deep. Due to the narrow geometry of was taken as 210 GPa, while 200 GPa was assumed for the
decks deeper than 80 mm, push tests ended with rib shearing rest of steel components. For tests conducted by Smith and
failure and significant decrease in the shear connector resist- Couchman (2010), the yield stress (fy) of headed studs,
ance when compared to other results from tests with decks profiled deck and wire-mesh were taken as 475, 350 and
less than 80 mm deep. Using a wide a range of sheeting 500 MPa, respectively. For Rambo-Roddenberry (2002)
depths and geometries ensure that the results of FE model tests, the yield stress (fy) of headed studs, profiled, and
are acceptable for modelling composite beams with a deep wire reinforcement was taken as 400, 280, and 500 MPa,
profiled sheeting. The full details of both experiments are respectively.
summarised in Table 1. A bi-linear stress–strain relationship was used in
The second experiment involved 27 push tests to investi- ABAQUS to define the steel components with elastic per-
gate the mesh reinforcement position, slab depth, and num- fectly plastic behaviour. This relationship has two parts:
ber and transverse spacing between studs. Headed studs of first is the linear elastic region and second fully plastic part,
19 × 100 mm were welded through 60 mm deck with rib where stress remains constant beyond the yield point. The
geometry (bo/hp) of 2.35 and placed in the favourable side. linear elastic part is defined by the modulus of elasticity
Tests results showed that placing the mesh 25 mm from the of the material. While the yield stress of the material and
top surface of concrete slab caused 20% decrease in the zero plastic strain represents the second fully plastic region.
shear stud resistance compared to placing the mesh on the Modelling steel components in this way is reasonable, as the
deck surface. While the change in the stud spacing from failure is mostly concrete related in beams with composite
75 to 140 mm appeared to have a small effect on the shear profiled sheeting slab. Generally, fracture of steel, geometric
13
Table 1 Comparison of shear stud capacities obtained from experiments and FE analysis
Tested by Series Profiled sheeting (mm) Headed stud Slab depth (mm)
International Journal of Steel Structures
Rambo-Roddenberry (2002) S10 40.8 0.9 152.4 50.8 19 × 100 1 F 152.4 89.7 92.3 0.97
S12 40.8 0.9 152.4 50.8 19 × 100 1 U 152.4 63.4 66.7 0.95
S20 24.2 0.9 152.4 76.2 9.5 × 127 1 F 152.4 35.0 43.4 0.81
S21 24.2 0.9 152.4 76.2 22 × 127 1 F 152.4 60.5 63.5 0.95
S22 24.2 0.9 152.4 76.2 9.5 × 127 1 U 152.4 27.5 34.3 0.80
S23 24.2 0.9 152.4 76.2 22 × 127 1 U 152.4 70.3 74.6 0.94
S24 24.2 1.5 76.2 114.3 19 × 152 1 C 171.5 34.6 36.8 0.94
S25 24.2 1.5 88.9 152.4 19 × 197 1 C 222.3 27.6 27.8 0.99
Smith and Couchman (2010) A1U 18.9 0.9 155 60 19 × 100 1 F 140 91.2 90.4 1.01
A1D 19.0 0.9 155 60 19 × 100 1 F 140 116.4 111.6 1.04
B1U 19.0 0.9 155 60 19 × 100 1 F 225 112.1 112.0 1.00
A2DX 19.3 0.9 155 60 19 × 100 2 F 140 62.4 63.8 0.98
A2UY 19.3 0.9 155 60 19 × 100 2 F 140 56.0 50.8 1.10
Author's personal copy
13
personal copy
International Journal of Steel Structures
13
Author's personal copy
International Journal of Steel Structures
( ) [ ( )3 ] [( )]
dc = 1− 𝜎c ∕fcm (1) C1 w C2 w
f (w) = 1+ exp (5)
wc wc
d c 𝜎c
𝜀∼in
c
= 𝜀∼pl
c
+ (2)
1 − dc Eo
2.5 Contact Interactions and Constraints
2.4.2 Tensile Behaviour
A surface to surface contact pair algorithm was used to
The strain softening behaviour of concrete in tension was define stud-concrete and deck-concrete contact surfaces.
based on the fracture energy cracking method and using the The tangential and normal behaviour were used to define
exponential function given in Eq. (3) as proposed by Cornelis- the interaction properties between surfaces. The normal
sen et al. (1986). The fracture energy to any concrete grade behaviour was defined by the default option “Hard” contact
is represented by the stress-crack opening curve, particularly pressure-overclosure relationship. The tangential behaviour
the area under the unloading part. The tensile stress σt versus was defined via the penalty frictional formulation with a
crack opening displacement was calculated from Eq. (4). The coefficient of friction of 0.5 as recommended in Eurocode
tensile damage variable was found using the formula (dt = 1 4. The stud-concrete contact was defined by treating the stud
– σt/ft). The fracture energy was measured from the formula as a master and concrete as a slave surface. In deck-concrete
Gf = Gfo (fcm/fcmo)0.7 as suggested by MC 10 CEB-FIP model contact, concrete surface was treated as a master due to its
code (2010). higher stiffness and steel deck as a slave. Figure 6 shows a
surface to surface contact interaction of stud-concrete and
wc = 5.14Gf ∕ft (3) concrete-deck before the assembly. The beam-deck contact
was specified by general contact algorithm with frictionless
𝜎t = ft [f (w)−f (wc )] formulation. The nodes of steel decking around the headed
(4)
stud circumference were tied to the nodes of stud’s shank at
13
Author's personal copy
International Journal of Steel Structures
its base. The steel reinforcement bars were embedded inside (2010), which had profiled decking depths ranging from 50
the concrete slab using the embedded constraint method. to 150 mm. The validation was based on the shear connec-
tor resistance, load-slip behaviour and mode of failure. The
2.6 Boundary Conditions and Load Application FE results showed a close agreement with the experiments.
The average ratio of PTest/PFE is 0.97 with the corresponding
Figure 3 illustrates the boundary conditions and the load- coefficient of variation of 8.30%. The results obtained from
ing surface of the push test models. Same boundary condi- the numerical analysis are compared with the experimental
tions were applied to both validation and parametric studies. results in Table 1. Numerical load versus slip curves for
All nodes of the underside of the beam flange, denoted by selected push tests are compared with experiments from both
surface 1, were restrained from moving in the Y (upward) studies in Figs. 8 and 9, which shows a reasonable match.
direction and rotating in the X (transverse) and Z (longitudi- In case of the verified models for Rambo-Roddenberry
nal) direction due to symmetry. Surface 2, embedded to the (2002), the failure mode was stud shearing for the series
ground in experiments, representing profiled sheeting and S20–21, rib punching and stud shearing for the series
concrete slab nodes was prevented from translating in the Z S10, S12 and S22–23, and rib shearing or (rib cracking)
direction. Surface 3 denotes the shear loading surface on the for the series S24–25 which incorporate profiled sheeting
beam flange. All numerical models were loaded by applying greater than 100 mm deep. On the other hand, all specimens
a uniform velocity at a rate of 0.5 mm/s, after a convergence in Smith and Couchman (2010) failed by concrete cones
study of various loading rates. This loading rate was cho-
sen based on a compromise between accuracy and compu-
tational efficiency. To ensure the stability of the uniform 35
Slip (mm)
2.7 Validation of FE Model
Fig. 8 FE and experiment Load-slip curves for push test series (S25)
The numerical results were verified against the experiments from Rambo-Roddenberry (2002)
from Rambo-Roddenberry (2002) and Smith and Couchman
120
0.0035 100
0.003
Load per stud (kN)
80
ALLKE/ALLIE
0.0025
0.002 60
0.0015 A1U-FE
40
0.001 A1U-1-Exp.
20 A1U-2-Exp.
0.0005
A1U-3-Exp.
0 0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 18 20 22
Slip mm Slip (mm)
Fig. 7 The ratio of kinetic energy over internal energy (ALLKE/ Fig. 9 FE and experiment Load-slip curves for push test (A1U) from
ALLIE) versus slip capacity for the mass scaling factor of 10 Smith and Couchman (2010)
13
Author's personal copy
International Journal of Steel Structures
around headed studs and formation of a horizontal crack more precise behaviour of composite beams with decks
along the surface of the concrete slab. A very similar fail- deeper than 80 mm and to investigate different headed stud’s
ure pattern was observed in the numerical models as well distribution. The section of steel beam was chosen to be
(see Figs. 10, 11). The cracking is represented by the tensile 254 × 254 × 73 UC. The test program involved modelling
damage variable, DAMAGET (dt); the value equal to one composite beams with two different types of steel decks.
represents complete cracking and zero denotes no tensile One with a deep deck of 100 mm produced by Tata Steel
crack. After this extensive validation of the finite element (2016), and the other with a deep deck of 146 mm produced
model in terms of stud resistance, slip capacity and failure by Kingspan (2011). Although the first deck is marketed for
modes, it can be confidently used to model composite beams non-composite beams, it felt necessary to account it for the
with 100 and 146 mm deep decks. composite beam action herein to address more than one type
of steel deck deeper than 80 mm. Thus, the average rib width
(bo) was 100 mm and the rib deck ratio (bo/hp) was 1.0. In
3 Parametric Study case of the 146 mm deep deck, the average rib width was
97.5 mm, resulting in a rib deck ratio of 0.67. The sheeting
The influence of a deep profiled sheeting was investigated thickness of both steel decks was 1.2 mm. Full dimensions
on the behaviour of secondary composite beams. A full- of both steel decks are shown in Fig. 12.
scale push test model was created using five ribs of deep Because of the narrowness of the bottom surface of the
profiled sheeting. The use of five ribs was meant to achieve sheeting rib, it was not possible to place the headed studs
13
Author's personal copy
International Journal of Steel Structures
100 mm deep deck and four with the 146 mm deep deck. This
group was considered as the control case. The second group
(G2) resembles G1, except two headed studs were placed in
each rib at a transverse spacing of 5d. The third group (G3)
involved 8 tests (T17–24) similar to G1, but each test had
three single studs placed in every other rib. This means there
was no stud in the second and fourth ribs. The fourth group
(G4) contained 8 tests (T25–32) having the studs placed in
the three middle troughs, where there was no stud in the first
and last ribs. Figure 14 shows the distribution of studs in
G3–4. In the first four groups, all concrete slabs were rein-
forced by a single layer of wire-mesh bars placed on the deck
surface. The fifth group (G5) also had eight tests (T33–40)
similar to G1, except the concrete volume within ribs was
reinforced by unique grid bars. For inner three ribs, the grid
bars were positioned at the back and front to the stud connec-
tors, while the outer ribs had the layer placed either at front or
Fig. 12 Details of the 100 and 146 mm profiled steel decks back of the stud. The bars were made from A193 wire-mesh
(7 mm in diameter) and assembled with the conventional slab
reinforcement above the steel deck as one part. The model of
such grid bars is shown in Fig. 15.
4 Results and Discussion
13
Author's personal copy
International Journal of Steel Structures
Table 2 Details and shear stud capacities of push tests obtained from the parametric study
Group Test fc (MPa) Profiled sheeting Headed stud PFE (kN) Parameter
bo (mm) hp (mm) bo/hp d × hsc nr
pressure. This is indicated by a sudden drop in the load more moment is resisted by the studs causing a double
after further decline in the load bearing capacity as seen curvature (Lawson 1997).
in Figs. 16 and 17. The remaining studs experienced a This part of the research provides the first insight into
double curvature deformation, particularly the last three the correlation between narrow and wide rib decks regard-
studs away from the applied load as shown in Fig. 20. This ing the shear connector resistance. The shear stud capac-
deformation is common in narrow rib steel decks, where ities obtained from steel decks deeper than 80 mm are
13
Author's personal copy
International Journal of Steel Structures
Fig. 14 FE push tests: a studs placed alternatively in G3; b studs placed in the three middle ribs in G4
13
Author's personal copy
International Journal of Steel Structures
90
80
70
Load per stud (kN)
60
50
40
30 T1
T2
20
T3
10 T4
0
0 2 4 6 8 10 12 14
Slip (mm)
Fig. 16 Load-slip curves for tests with the 100 mm deep decking
90
80
T5
70 T6 Fig. 18 Side cut view of concrete failure in push models with deep
Load per stud (KN)
40
30
20
10
0
0 2 4 6 8 10 12
Slip (mm)
Fig. 17 Load-slip curves for tests with the 146 mm deep decking
13
Author's personal copy
International Journal of Steel Structures
100
Load per stud (kN)
90
40
30
20
C12 C20 C30 C40
Concrete grade (MPa)
80
40
30
C12 C20 C30 C40
Concrete grade (MPa)
no stud remained completely intact throughout the test. three ribs (refer to G4) did not add any benefit. The shear
With less concrete damage and up to 20% increase in the connector resistances obtained from this group were close
shear connector resistance, this unique layout of studs, if to the control case (G1). This observation proves that wide
implemented in practice, could be beneficial in composite longitudinal spacing between studs could play a significant
beams with deep decks. Contrarily to G3, omitting studs role in developing the shear connector resistance in deep
from the first and last ribs, and placing studs in the middle decks.
13
Author's personal copy
International Journal of Steel Structures
40
30
C12 C20 C30 C40
Concrete grade (MPa)
13
Author's personal copy
International Journal of Steel Structures
Fig. 24 Comparison between load-slip curves obtained from G1 and G5 for tests with the 100 mm deep deck
5.2 Eurocode 4 Provisions of PFE/PEC4 is 3.99, with the coefficient of variation of 27.5%.
Applying the dimensions of decks deeper than 80 mm with
The nominal strength of headed shear stud connector is their relevant stud height in Eq. (9) resulted in very low
taken as the lesser value obtained from Eqs. (7–8) and mul- reduction factor ranging from 0.11 to 0.28. For the conven-
tiplied by a reduction factor using Eq. (9). tional profiled decks, the ratio of the average rib width to
√ the rib height (bo/hp) is greater than 1.5 in most cases. How-
PEC4 = 0.29𝛼d2 fc Ec (7) ever, in the steel decks investigated herein, the rib deck ratio
(bo/hp) ≤ 1.0. This is the reason Eq. (9) resulted in low reduc-
𝜋d2 tion factor, which in turn gave very low predicted strengths. It
PEC4 = 0.8fu (8) is recommended that Eq. (9) should be calibrated to account
4
for the geometries of decks deeper than 80 mm.
[ ]
0.7 bo hsc
kt = √ −1 (9) 5.3 ANSI/AISC Provisions
Nr hp hp
The nominal strength of one headed shear stud connec-
where α = (h sc/d − 1) for (3 ≤ h sc/d ≤ 4) and α = 1.0 for
tor embedded in a solid concrete slab is determined using
(hsc/d > 4). The factor kt should not be taken greater than the
Eq. (10).
appropriate value ktmax given in Table 4.
√
The relationship between the load per stud obtained from PAISC = 0.5As fc Ec ≤ Rg Rp As fu (10)
the FE analysis and the EC4 predicted strengths is shown in
Fig. 26. Clearly, the EC4 equations highly underestimated √
the headed stud strengths in deep decks. Some predictions Ec = 0.043w1.5
o
fc (11)
were 3–5 times lower than the FE results. The average ratio
13
Author's personal copy
International Journal of Steel Structures
For decks oriented perpendicular, Rg is 1.0 and 0.85 for 5.4 Comparison with Analytical Approach
single and double studs, respectively, and Rp is 0.6. According to Johnson and Yuan (1998)
The FE results are compared to the ANSI/AISC predicted
strengths in Fig. 27. It was found that ANSI/AISC equations Based on a theoretical analysis carried by Johnson and
gave unsafe predictions in general. The average ratio of PFE/ Yuan (1998), the researchers proposed several equations
PAISC is 0.79, and the corresponding coefficient of variation is to predict the strength of headed stud in composite beams
16.96%. It was noticed that the controlling equation was the with transverse sheeting. The equations were based on
steel failure (Rg Rp As fu) when the strength of concrete was five modes of failure including shank shearing (SS), rib
higher than 20 MPa. The predicted strengths of the headed punching (RP), rib punching with shank shearing (RPSS),
stud in case of single and double studs remained constant when rib punching with concrete pull-out (RPCP), and concrete
the concrete strength was greater than 20 MPa. This suggests pull-out (CPT). In this research, two analytical approaches
that the use of equations in the American specifications would were selected, namely the developed equations from RPCP
result in unreliable values as they are not capable of dealing and CPT. The RP equations were not chosen as they do not
with different variables, such as, the concrete strength and consider the number of studs per rib.
number of studs per rib. Moreover, the American code comes
with the drawback that the geometries of steel decks are not
considered.
13
13
Table 3 Comparison of shear stud capacity obtained from FE analysis and current design equations
Group Test PFE (kN) PEC4 (kN) PAISC (kN) PRPCP (kN) PCPT (kN) PKonrad (kN) PFE/PEC PFE/PAISC PFE/PRPCT PFE/PCPT PFE/PKonrad
G1 T1 50.2 16.7 65.0 39.7 52.2 37.4 3.00 0.77 1.26 0.96 1.34
T2 59.3 22.7 89.3 46.9 62.6 45.2 2.61 0.66 1.26 0.95 1.31
T3 70.7 26.7 89.3 51.1 69.8 52.8 2.65 0.79 1.38 1.01 1.34
T4 82.5 26.7 89.3 51.2 70.2 59.2 3.09 0.92 1.61 1.18 1.39
T5 48.1 9.4 65.0 31.8 44.9 32.1 5.12 0.74 1.51 1.07 1.50
T6 57.2 12.8 71.4 38.7 53.0 38.9 4.48 0.80 1.48 1.08 1.47
T7 68.6 15.0 71.4 42.8 59.3 45.4 4.58 0.96 1.60 1.16 1.51
T8 79.1 15.0 71.4 42.9 59.7 50.9 5.28 1.11 1.84 1.32 1.55
G2 T9 39.1 11.8 65.0 48.1 45.4 29.9 3.31 0.60 0.81 0.86 1.31
T10 47.8 16.0 75.9 59.0 52.7 36.2 2.98 0.63 0.81 0.91 1.32
T11 54.8 18.9 75.9 67.1 57.4 42.3 2.91 0.72 0.82 0.95 1.30
T12 62.1 18.9 75.9 67.2 57.7 47.3 3.29 0.82 0.92 1.08 1.31
T13 38.3 6.6 60.7 40.4 37.3 25.7 5.77 0.63 0.95 1.03 1.49
T14 45.7 9.0 60.7 51.3 42.3 31.1 5.07 0.75 0.89 1.08 1.47
Author's personal copy
T15 50.3 10.6 60.7 59.5 46.0 36.3 4.75 0.83 0.85 1.09 1.38
T16 53.2 10.6 60.7 59.5 46.3 40.7 5.02 0.88 0.89 1.15 1.31
Mean 3.99 0.79 1.18 1.06 1.39
CoV (%) 27.48 16.96 29.84 11.00 6.41
International Journal of Steel Structures
Author's personal copy
International Journal of Steel Structures
5.4.1 Combined Concrete Pull-Out and Rib Punching where PEC4 is the shear connector resistance in a solid slab
Failure (RPCP) calculated from Eqs. (7–8). e is equal to 50 and 48.75 mm
for 100 and 146 mm deep decks, respectively.
For combined rib punching and concrete pull-out failure of For the concrete pull-out failure mode, the equations are
studs in slabs with two studs placed in series or diagonally as follows:
in a rib, the stud placed on the favourable side is assumed to PCP = kcp PEC4 (18)
fail by concrete pull-out, whilst the stud placed on the unfa-
vourable side is assumed to fail by rib punching. Eventually, [
( )0.5 ]/( )
the shear resistance of each stud is assumed to be the mean 2 2
kcp = 𝜂cp + 𝜆cp 1 + 𝜆cp −𝜂cp 1 + 𝜆2cp ≤ 1.0 (19)
value of PRP and PCP as given in Eq. (12).
( )
PRPCP = PRP + PCP ∕2 (12) [ ( )]/
h ( )
𝜂cp = 0.56vtu h2sc e + st − sc hp PEC4 if 0.75hsc ≤ e + st
For the rib punching failure mode, the equations are as 4
follow: (20)
( ( ) )]/
[
( )2 e + st ( )
𝜂cp = vtu e + st 0.75hsc − hp PEC4 if 0.75hsc > e + st (21)
3
13
Author's personal copy
International Journal of Steel Structures
90 90
80 80
70 Underestimation 70 Underestimation
60 60
50 50
40 Overestimation
40 Overestimation
30
30
20 Single studs (G1)
20 Single studs (G1)
Double studs (G2)
Double studs (G2) 10
10 FE results = predicted strengths
FE results = predicted strengths
0
0 0 10 20 30 40 50 60 70 80 90
0 10 20 30 40 50 60 70 80 90
Predicted strengths by Johnson and Yuan, PCPT (kN)
Predicteded strengths by Johnson and Yuan, PRPCP (kN)
𝜆cp = eTy ∕hp PEC4 (22) The predicted strengths considerably overestimated the
shear resistance of tests with stud pairs. It was noticed
that the predicted values were even higher than the ones
Ty ≅ 0.8As fu (23) achieved for the single studs which ideally should be
lower. The proposed equations from the RPCP failure
0.5
vtu = 0.8fcu ≤5 (24) would give unsteady results for the composite beams
with deep decks. The discrepancy was due to the way the
Theoretically, the equations developed from the RPCP equations were developed, which was based on the studs’
failure mode are not suitable for composite beams with arrangement.
deep decks, especially the equations for concrete pull-out
failure. These formulae were developed based on plac- 5.4.2 Concrete Pull-Out Failure (CPT)
ing two studs in series or staggered in a trough. In the
current research, it is not possible to place two studs in For concrete pull-out failure of studs in slabs with one stud
any arrangement but the same line due to the narrowness per trough, in a central or favourable position, the strength is
of rib. Nevertheless, if these equations were meant to be determined by multiplying the shear stud strength in a solid
used, then st would be zero in case of single studs, and st slab calculated from Eqs. (7–8) by the reduction factor kcpt
would be assumed as the transverse spacing between studs as shown below:
in case of stud pairs. Thus, Eq. (21) would be used for
single and double studs of 19 × 195 mm, as it justifies the PCPT = kcpt PEC4 (25)
condition of 0.75hsc > (e + st). On the other hand, Eq. (20)
[ ( )0.5 ]/( )
would be used for single and double studs of 19 × 140 mm,
as it justifies the condition of 0.75hsc ≤ (e + st). kcpt = 𝜂cpt + 𝜆cpt 1 + 𝜆2cpt −𝜂cpt
2
1 + 𝜆2cpt ≤ 1.0
The predicted strengths of the developed equation (26)
based on RPCP failure are compared to the FE results [ ( )]/
hsc
in Fig. 28. The average ratio of P FE/P RPCT is 1.18, and 𝜂cpt = 0.56 vtu h2sc bo − hp Nr PEC4 ≤ 1.0 (27)
the coefficient of variation is 29.84%. In case of the 4
push models with single studs, the developed equations
underestimated the shear resistance of headed stud when 𝜆cpt = e Ty ∕hp PEC4 (28)
the concrete grade was C12 and C20 with a deviation of
over 20%. However, the deviation in results became much Ty ≅ 0.8 As fu (29)
higher when the concrete grade was C30 and C40. The
predicted strengths remained within the range of 47–51 The predicted strengths based on CPT failure are com-
kN and 39–42 for tests with 100 and 146 mm deep decks, pared with the numerical results as shown in Fig. 29. In gen-
respectively and did not reveal a significant rise as the eral, the average ratio of PFE/PCPT is 1.06 with the coefficient
concrete strength increased. of variation of 11.00%. In terms of the push models with
13
Author's personal copy
International Journal of Steel Structures
results from the push models with stud pairs. The maximum 50
scatter in results was 15%. Overestimation
40
Although the developed equations from CPT failure did
not closely predict the stud capacity in some cases, they 30
Single studs (G1)
seemed more relevant than the ones from RPCP failure to 20
Double studs (G2)
deal with composite beams with deep decks. In fact, the 10
FE results = predicted strengths
equations consider the number of studs per trough, and they
0
were developed based on placing studs in a central or favour- 0 10 20 30 40 50 60 70 80 90
able position which meets the way the studs are arranged in
Predicted strengths by Konrad, PKonrad (kN)
this study (central position). Hence, the developed equations
from CPT failure could precisely predict the strength of stud
Fig. 30 FE results versus PKonrad predicted strengths
if they were calibrated again to cover the geometries of steel
decks deeper than 80 mm.
where AWulst,eff is the effective area of weld collar = 0.5
hWulst dWulst (see Table 5), kn = 1 for single studs per rib, and
5.5 Comparison with Analytical Approach kn = 0.8 for stud pairs.
According to Konrad (2011) Figure 30 presents the comparison between the results
from the FE analysis and the proposed equations by Kon-
The formulae developed by Konrad (2011) considered the rad (2011). It can be noticed that the proposed equations
influence of the welding position of the stud on the shear significantly underestimated the shear stud capacities of all
resistance. The mean shear strength of a stud in a profiled models, the average ratio of PFE/PKonrad is 1.39, and the cor-
decking is the smaller value among Eqs. (30–31). The reduc- responding coefficient of variation is 6.41%. This is because
tion factor (kt) in Eq. (32) is for the mid-position of stud of the value before the term (bo/hp)2 in Eq. (32) being very
and the case of hsc/hp ≤ 1.56. The author assumed that the small that resulted in a reduction factor (kt) ≤ 0.521. It is
geometry of decking does not affect the resistance of stud then recommended that if the equations proposed by Konrad
rupture, and thus the reduction factor is only included in the (2011) were meant to predict the shear connector resistance
concrete failure equation. in deep decks, Eq. (32) should be calibrated again to give
more realistic reduction factor which in turn would result in
Pm,s = 39.85AWulst,eff fc2∕3 + 0.59fu d2 (30)
more accurate strength capacity.
( )
Pm,c = kt 39.85AWulst,eff fc2∕3 + 3.75d2 fc1∕3 fu1∕2 (31)
6 Conclusions
[ ( )2 ( ) ( )]
kt = kn 6.79 × 10−4 bo ∕hp + 0.17 bo ∕hp + 0.25 hsc ∕hp ≤ 1.0
This research aims to bridge a crucial gap in knowledge and
(32) understating of steel–concrete composite beams with sheet-
ing deeper than 80 mm. Lack of design rules for composite
beams with deep decks and very limited previous research
Table 5 Effective area of weld Diameter of stud Height of weld collar Diameter of weld collar Effective area of weld
collar according to Konrad (mm) (hWulst) (mm) (dWulst) (mm) collar (AWulst,eff) (mm2)
(2011)
10 2.5 13.0 16.3
13 3.0 17.0 25.5
16 4.5 21.0 47.3
19 6.0 23.0 63.0
22 6.0 29.0 87.0
25 7.0 40.0 140.0
13
Author's personal copy
International Journal of Steel Structures
13
Author's personal copy
International Journal of Steel Structures
Qureshi, J., Lam, D., & Ye, J. (2011a). Effect of shear connector spac- Rambo-Roddenberry, M. D. (2002). Behaviour and strength of welded
ing and layout on the shear connector capacity in composite stud shear connectors. Ph.D. Thesis, Virginia Polytechnic Institute
beams. Journal of Constructional Steel Research, 67(4), 706–719. and State University.
Qureshi, J., Lam, D., & Ye, J. (2011b). The influence of profiled sheet- Robinson, R. (1988). Multiple stud shear connections in deep ribbed
ing thickness and shear connector’s position on strength and duc- metal deck. Canadian Journal of Civil Engineering, 15(4),
tility of headed shear connector. Engineering Structures, 33(5), 553–569.
1643–1656. Shin, D. K., Dat, B. V., & Kim, K. (2014). Compressive strength of
Rahnavard, R., & Hassanipour, A. (2015). Steel Structures analysis HPS box girder flanges stiffened with open ribs. Journal of Con-
using ABAQUS. Kerman: Academic Center for Education, Culture structional Steel Research, 95, 230–241.
and Research, Publishing Organization of Kerman Branch. Shin, D. K., Le, V. A., & Kim, K. (2013). In-plane ultimate compres-
Rahnavard, R., Hassanipour, A., & Mounesi, A. (2016). Numerical sive strengths of HPS deck panel with U-shaped ribs. Thin-Walled
study on important parameters of composite steel–concrete shear Structures, 63, 70–81.
walls. Journal of Constructional Steel Research, 121, 441–456. Smith, A. L., & Couchman, G. H. (2010). Strength and ductility of
Rahnavard, R., Hassanipour, A., Suleiman, M., & Mokhtari, A. headed stud shear connectors in profiled sheeting. Journal of Con-
(2017a). Evaluation on eccentrically braced frame with single and structional Steel Research, 66(6), 748–754.
double shear panel. Journal of Building Engineering, 10, 13–25. Suresh, K., & Regalla, S. P. (2014). Effect of time scaling and mass
Rahnavard, R., Khaje, M. T., Hassanipour, A., & Siahpolo, N. (2017b). scaling in numerical simulation of incremental forming. Applied
Parametric study of seismic performance of steel bridges pier Mechanics and Materials, 612, 105–110.
rehabilitated with composite connection. Journal of Struc- Tata Steel. (2016). Composite floor decking design and technical infor-
tural and Construction Engineering. https://doi.org/10.22065/ mation. United Kingdom.
JSCE.2017.92128.1259.
Rahnavard, R., Naghavi, M., Aboudi, M., & Suleiman, M. (2018). Publisher’s Note Springer Nature remains neutral with regard to
Investigating modeling approaches of buckling-restrained braces jurisdictional claims in published maps and institutional affiliations.
under cyclic loads. Case Studies in Construction Materials, 8,
476–488.
13