Construction and Building Materials 125 (2016) 1–11

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

New shear connector for Open Web Steel Joist with metal deck and
concrete slab floor system
Gregory Merryfield a, Amr El-Ragaby b,⇑, Faouzi Ghrib c
a
Windsor, Ontario, Canada
b
Department of Civil and Environmental Engineering, University of Windsor, Windsor, Ontario N9B 3P4, Canada
c
Department of Civil and Environmental Engineering, University of Windsor, Windsor, Ontario, Canada

h i g h l i g h t s

 Composite floor system of Open-Web Steel Joists and concrete slab is discussed.
 The composite action using shear stud connectors is reviewed and summarized.
 Puddle welds and Hilti-screw pins shear connectors are proposed and investigated.
 Two full-scale OWSJ composite deck slabs were constructed and tested.
 Puddle-welds shear connectors provided significant levels of composite action.

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

Article history: The composite floor system, consists of a concrete slab poured on steel sheets, supported over Open Web
Received 14 October 2015 Steel Joists (OWSJ); is widely used in commercial and industrial buildings. To achieve the desirable
Received in revised form 25 July 2016 strength, shear connector (studs) have to be welded on the OWSJ to ensure composite action of the three
Accepted 3 August 2016
components. Extending the application of this composite floor system to residential buildings, alternative
Available online 11 August 2016
shear connectors such as puddle-welds and Hilti-screws would greatly reduce the expense and accelerate
the construction. Yet the current design codes consider these alternates structurally inadequate due to
Keywords:
lack of research. The objective of this research is to investigate the ability of puddle-welds and Hilti-
Open Web Steel Joist
Shear connectors
screws to develop composite action for building applications. The experimental program consists of test-
Shear studs ing two full scale composite decks, each consisted of two OWSJs supporting a 2.4 m wide concrete slab
Longitudinal shear over 6.7 m span. Puddle welds and Hilti screws were used as main and only shear connector. Both test
Composite construction prototypes were tested until failure under different combinations of static and fatigue load cycles. Test
Composite action results showed that significant composite action is developed using puddle-weld shear connectors and
their behaviour meets the serviceability and ultimate strength requirements for residential applications.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction minimal increases in material as oppose to W-shape beams of

similar depths. Therefore, OWSJs are attractive structural elements
Over the last few decades, steel and concrete structures have for increased strength and stiffness while offering sufficient open-
been a very popular selection in construction; particularly compos- ing for air ducts and other services. The corrugated steel sheets act
ite floor systems consisting of a concrete deck poured on top of cor- as permanent in-place formwork and a thin layer of tensile rein-
rugated steel sheets, supported by Open-Web Steel Joists (OWSJ). forcement for the concrete slab. Most corrugated steel decking
Open Web Steel Joists (OWSJ) are steel trusses of relatively low used in composite flooring systems contain embossments within
mass with parallel or slightly pitched chords and triangulated the profile of the deck to maintain a superior bond to the concrete
web systems proportioned to span between walls, structural sup- slab cast on top. The concrete slab is designed to resist compressive
porting members, or both, and to provide direct support for floor stresses with only light, top reinforcement (welded wire mesh) to
or roof deck [1]. OWSJs offer enlarged effective depths with eliminate/control shrinkage cracks. The purpose of this construc-
tion practice is to build upon the strengths of each material and
to compensate for the weakness of each other simultaneously.
⇑ Corresponding author.
Composite construction is typically used in repeated floor
E-mail address: [email protected] (A. El-Ragaby).

http://dx.doi.org/10.1016/j.conbuildmat.2016.08.006
0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.
2 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11

buildings because the floor system can be replicated efficiently develop and maintain composite action, significant economic ben-
floor after floor. Composite elements also have a larger moment efits can be achieved. Easy and faster construction as well added
capacity, which allows for smaller section size and depth; which safety to the workers and construction site through eliminating
reduce the floor height over the course of a soaring structure. the trip and injury hazards from using projected shear studs are
Therefore, composite elements are lighter, slenderer, stiffer and just prime examples. Moreover, the anticipated construction cost
more economical while resisting identical loads compared to (include labour, tools and material) puddle-welds and Hilti-
non-composite ones. screws is less than 50 and 20% of shear studs, respectively. Further-
Composite action is the term used to describe the behaviour of a more. the extra strength from considering a composite action will
composite structure and has a great effect on the stress and strain be beneficial to reduce the depth of the section and/or allow cover-
of beam and floor composite systems. To ensure full strain compat- ing longer spans. These advantages will promote the composite
ibility at the interface between steel and concrete elements in floor system to residential building and make it more efficient in
composite section (composite action), stiff and rigid shear connec- industrial application.
tors, commonly called shear studs, are used [1]. (Fig. 1) illustrates The objective of this research is to determine if composite
the behaviour and strain profile of a composite to a non-composite action can be achieved without the use of shear stud and if partial
section [2]. The composite section undergoes less deflection and no or full-composite action can be achieved when puddle welds and
slip due to the addition of the shear studs. Shear studs are shear Hilti-screw are used as an alternate shear connector for shear
resisting dowels welded to the top flange of the steel sections studs.
and encased in concrete in order to transfer the longitudinal shear
forces between the deck and the supporting beam. The shear studs 2. Background
are the primary horizontal shear resistance component in the com-
posite element and are the only code-approved method for achiev- Robinson and Fahmy (1978) presented the experimental results
ing sufficient composite action [1]. However, shear studs impose and analysis of a number of composite open-web joists with metal
construction constraints such as extra cost, time, licensed profes- deck. Test results showed that composite open-web joists with
sionals and trip hazards. ribbed metal floors have greater stiffness, strength and ductility
Composite flooring with shear studs is typically used in indus- than non- composite open-web joists [5]. Kennedy and Brattland,
trial buildings, i.e. large spans and heavy live loads. In case of (1992) studied the effect of concrete shrinkage on the behaviour
lighter live loads and shorter spans, designers often opt for non- of composite steel joists and recommended the use of light top
composite options but have to use deeper cross sections for gird- reinforcement mesh [6]. Wang et al. (2010) conducted twelve
ers. Currently, during construction, arc-spot weld (puddle welds) push-out test specimens of stud shear connectors with large diam-
or screw pins (Hilti-screws) are used in order to temporarily fasten eter and high strength. It was found that overall capacity increased
the metal deck sheets in place on top of the supporting members. with the use large diameter and high strength studs [7]. Hedaoo
(Fig. 2) displays two alternate fasteners. To date, all design codes (2012) studied the structural behaviour of composite concrete
do not account for puddle welds [3] or screws [4] in the load resis- slabs with profiled steel decking with different shear span lengths,
tance, i.e. no composite action is exploited in a puddle weld or under static and cyclic loadings over simply supported conditions.
Hilti-screw connector design. This is mostly due to the lack of Steel sheets with embossments were used to increase the compos-
research allotted to this topic. If these two simple fasteners can ite interaction between the concrete and to improve their bond

(a) Composite beam with shear connectors (right) vs. non-composite beam (left)

(b) Dis-continuous strain profile (c) Continuous strain profile

(non/partial-composite section) (fully composite section)
Fig. 1. Composite vs. non-composite structures [2].
 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11 3

(a) Puddle weld [3] (b) Hilti-screws or steel deck

screws [4]
Fig. 2. Alternate shear connectors.

characteristics. It was concluded that as the shear span length connectors. It was concluded that both puddle-welds and large
increased, the longitudinal shear stress of the slab decreased [8]. diameter Hilti-screws (6.25 mm diameter) developed significant
Lakshmikandhan et al. (2013) investigated the longitudinal shear composite action compared to shear studs [10].
transfer mechanism at the interface between the steel section
and concrete deck using headed shear studs, shear rods or no con-
nection. The floor decks constructed with stud bolts and shear rods 3. Details of the experimental program
showed different behaviours when compared to that of the com-
posite slab without shear connectors. The insertion of shear con- 3.1. Test specimens
nector modifies the brittle behaviour of the composite slab into
ductile. The composite deck without shear connectors also slips This research includes construction, testing and analysis of two
and fails at the earlier load level [9]. Merryfield et al. (2015) tested full-scale composite floor prototypes to investigate the behaviour
T-shaped beam specimens each had an OWSJ with top concrete of alternative shear connections and to determine if composite
flange using shear studs, puddle welds and Hilti-screw shear action can be obtained without the use of the traditional shear

65 mm
116 mm

Shear
connector

600 mm 1200 mm 600 mm

(a) Cross section

254 mm

6.4 m

(b) Elevation view

(c) Corrugated Metal deck sheets (d) Casting the concrete slab

Fig. 3. Details of full-scale composite floor specimen.

4 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11

studs. Each of the composite decks was configured identical to one bars. A welded wire mesh was placed at mid-height of the concrete
another, with the exception of the shear connectors and secondary slab in test specimen CD-W to mitigate cracking on the surface of
reinforcement in the slab. The dimensions were based on current the deck while test specimen CD-S utilized a Glass Fiber Reinforce
practices where joist are spaced 1.2 m (4 feet) and spans typically Concrete (GFRC) slab in lieu of a Welded Wire Fabric mesh. Table 1
range from 5–10 m. presents summary of test specimens’ details.
The first composite deck utilized a 19 mm puddle weld to make
the deck-joist shear connection (CD-W); whereas, the second deck
employed a 6.35 mm Hilti-screw for the same purpose (CD-S). Each 3.2. Test set-up and loading procedures
deck measured 6.4 m long and 2.4 m wide, with a 65 mm clear
concrete slab cast on top of the 0.9 mm thick corrugated steel deck The composite floor system was simply supported in the longi-
sheets. Two OWSJ, each is 250 mm in depth, were spaced 1200 mm tudinal direction over 6.7 m span. It was loaded till failure under
apart in the transverse direction and were simply supported over two line loads at the third points as shown in (Fig. 4). A very stiff
6.4 m span in the longitudinal direction. Corrugated steel sheets, loading system was used to ensure uniform load is applied over
each of 900 mm wide and 2400 mm long, were welded (test spec- the entire width. The loading procedure according to the perfor-
imen CD-W) or screwed (test specimen CD-S) to the top chords at mance testing procedure according to the Canadian Institute for
single spot on each flute of the corrugated steel sheets (only on the Steel Construction, CISC, S16-1980, was followed [1]. The test pro-
exterior angle of the top cord of the OWSJ). The concrete slab has cedure specifies 4-point bending, in four successive cycles: load up
overall width of 2400 mm and was supported in the transverse to 25%, up to 60%, 100% of the predicted ultimate theoretical capac-
direction over 1200 mm span, on the 2 OWSJ, with two overhangs ity and lastly, load up to failure of the structure. Assuming failure
each of 600 mm wide. (Fig. 3) shows the details of test specimens. occurs at yielding of the bottom cord of the OWSJ, a failure load
The joist’s top and bottom chords are comprised of two of 133 kN using the section analysis was calculated. Accordingly,
L44  44  3.2 and two L51  51  4.8 back-to-back angles, test specimen CD-W was planned to undergo four monotonic load-
respectively. The web members are 17.5 mm in diameter round ing and unloading cycles at peak load levels of 33 kN, 80 kN,
133 kN and up to failure, as shown in (Fig. 5a). For CD-W test spec-
imen, the load was applied under load-controlled rate of 10 kN/
Table 1
min.
Test Matrix. On the other hand, CD-S was tested in both under monotonic
loading as well as cyclic fatigue loading. First, CD-S underwent
Test Shear Connector Concrete slab
Specimen Connector Type Spacing
monotonic loading and unloading cycles up to predetermined peak
load levels corresponding to the service and ultimate loads in res-
CD-W 19 mm Puddle 300 mm 65 mm reinforced with
weld Longitudinally wire mesh
idential and industrial building, 2.4 and 4.8 kPa, respectively,
CD-S 6.35 mm Hilti- 300 mm 65 FRC with glass fibres according to the National Building Code of Canada (2010), NBCC
screw Longitudinally [11]. Three peak loads, 40, 50 and 70 kN, were used and two load-
ing and unloading cycles were applied at each peak load, total of

(a) Auxiliary view of composite floor with load setup

(b) Test specimen under load (c) Stiff loading system

Fig. 4. Details of Test Set-Up.
 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11 5

120

PERCENTAGE Of Theoretical Capacity (%)

FAILURE
100

80

60

40

20

0
0 1 2 3 4 5
Load Cycle

(a) CD-W monotonic load cycles

140

120

100 SERVICE LOAD

ULTIMATE LOAD
Load (kN)

80

60

40

20

0
PRE-FATIGUE Cycles 50000 CYCLICAL Cycles POST-FATIGUE Cycles
(b) CD-S pre-fatigue, fatigue and post-fatigueloading
Fig. 5. Loading Procedures.

six pre-fatigue cycles, at loading rate of 20 kN/min. Thereafter, strain gauge was installed on the horizontal and vertical legs of
specimen CD-S was subjected to 50,000 fatigue loading cycles at one angel of both the top and bottom cords under the loading
peak load level corresponding to the service load, 40 kN and min- point, on the extreme top and bottom surfaces of the top and bot-
imum load of 8 kN at a frequency of 0.4 Hz. The CD-S specimen was tom cords at mid-span and on the first compression and tension
than subjected to five post-fatigue loading and unloading cycles at diagonals, as shown in (Fig. 6a). Also, three electric foil strain
increasing peak load levels of 40, 70, 90, 110 and 130 kN ‘‘failure gauges (70 mm long) were utilized. One strain gauge was installed
load” as shown in (Fig. 5b). Table 2 lists the details of the CD-S on the top of the concrete slab while the other two gauges were
loading pattern. installed on the front face across the depth of the concrete deck
at 15 and 35 mm from the top to monitor the strain distribution
3.3. Instrumentation across depth of the composite system. These strain measurements
will be used to develop cross sectional strain profiles to assess the
Eleven electric foil strain gauges (10 mm long) were installed on composite behaviour and the level of composite action. Three Lin-
the top and bottom cords as well as diagonals of the OWSJ. One ear Variable Displacement Transducers, LVDT’s, were also used to
capture the deflection at the third point, right under the line load,
Table 2 of the bottom cord, top cord and on top of the concrete slab over
Specimen CD-S full loading pattern. the OWSJ. (Fig. 6) display the distribution and location of different
Load Cycle # Type Max Load (kN)
sensors used in this study.
1 Pre-fatigue 40
2 40
3 50
3.4. Materials properties
4 50
5 70 All test specimens were constructed using normal weight,
6 70 ready-mixed concrete with a targeted 28-day concrete compres-
50,000 fatigue load cycles sive strength of 30 MPa. All test slab prototypes were cast and kept
7 Post-fatigue 40 in the laboratory, for 7 days, wrapped with plastic sheets in humid
8 70 environment for curing. The actual concrete compressive and ten-
9 90 sile strengths were determined based on the average value of com-
10 110
pressive and tensile splitting tests carried out on standard cylinder
11 Failure at 130
specimens of 100  200 and 150  300 mm, respectively, on the
6 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11

(a) Strain gauges and LVDT’s at 1/3 spans

(b) Concrete Strain gauges (c) Steel strain gauges (d) LVDT
Fig. 6. General test set-up and layout of instrumentation.

day of testing of the slabs. The standard cylinder specimens were showed a linear load-deflection behaviour up to 90 kN. A signifi-
cured under the conditions as their reference slabs. The obtained cant reduction in the flexural stiffness and non-linear load-
average concrete compressive and tensile strengths were about deflection relationship can be noticed when the loading continued
32 MPa and 3 MPa respectively. CSA grade 350 steel, of 375 MPa beyond 90 kN. Therefore, the full-composite action limit can be
yield strength and 200 GPa elastic modulus, was used in the fabri- assumed up to about 90 kN. This load is equivalent to a service live
cation of the OWSJ. load of 4.8 kPa, which is twice the NBCC load demand for residen-
tial building [11].
4. Test results and analysis (Fig. 7b) shows deflection behaviour for test specimen CD-S
during the pre-fatigue load cycles #2, 4 and 6 at peak load of
Composite action occurs when two or more elements/materials 40 kN, 50 kN and 70 kN (correspond to service (40 kN) and ulti-
are jointed together with stiff shear connectors to act as a single mate (60 kN) live load ratings for residential building as well as
structural element, with overall strength and stiffness more than the ultimate load for industrial buildings (72 kN)). It can be also
those of each individual element. The main challenge in composite noted the linear behaviour up to 70 kN load level as well as the 9
construction is to ensure that forces are transmitted effectively and to 13 mm deflection at the service live load level. (Fig. 7c) shows
safely between the two materials and there is full strain compati- the bottom chord deflection during monotonic loading before
bility at the interfaces. The composite action can be assessed from and after applying 50,000 fatigue cycles, cycles 6 and 8. It can be
the load-deflection and load-strain behaviours of composite struc- noticed that there is additional 5 mm accumulated residual deflec-
tures. Any change in the flexure stiffness of the composite element, tion due to the fatigue loading. Also, the post-fatigue cycle, cycle 8,
from the load-deflection relationship, before yielding of the steel or showed a non-linear load-deflection behaviour with reduced stiff-
crushing of the concrete, from the load-strain relationship, would ness compared to the pre-fatigue one, cycle 6. Both observations
indicate a decrease in the composite section inertia due to relative indicated that there is significant damage to CD-S, mainly in the
slip at the interface shear plan (loss of the composite action). Also, composite action, due to fatigue loading. (Fig. 7c) also compare
the cross sectional strain distribution (profile) is a unique method the pre-fatigue load-deflection behaviours of specimen CD-S (cycle
to investigate the overall performance of the composite system and 6) and specimen CD-W (cycle 2). It can be concluded that, even
to evaluate composite action behaviour. The following sections before the application of the fatigue loading, specimen CD-S has
present comparisons of the load-deflection, load-strain and strain more accumulated residual deflection compared to specimen CD-
profiles of the two test specimens at different load stages. W. (Fig. 7d) presents the post-fatigue load-deflection behaviour
of specimen CD-S up to failure, cycles 8 to 11. A significant reduc-
4.1. Load – deflection behaviour tion in the flexural stiffness and non-linear load-deflection rela-
tionship can be noticed with the increase of the applied load.
(Fig. 7a) compare the bottom chord load-deflection behaviour of Specimen CD-S failed at about 130 kN during cycle 11.
CD-W during the last cycle, cycle #3 up to failure, with cycles 1 and (Fig. 7e) shows the comparison of the deflection behaviour up to
2. It can be noticed that CD-W showed a perfect linear behaviour failure of specimens CD-W and CD-S. CD-W and CD-S failed at
up to 90 kN with minimum residual deflection. The 90 kN load sur- about 127 and 130 kN respectively. Specimen CD-S showed a sig-
passes the service (40 kN) and ultimate (60 kN) live load level for nificant reduction in the flexural stiffness and increased residual
residential buildings. At service live load level, 40 kN, the deflec- deflection due to the fatigue loading. The failure in both specimens
tion is about 8 mm compared to 13 mm allowable live load deflec- was by delamination/de-bonding between the corrugated metal
tion according to NBCC [11]. Specimen CD-W failed at about sheets and concrete and diagonal shear failure in the concrete slab
127 kN during the third load cycle and very close to the predicted under the load mainly due to local out-of-plan buckling of the top-
ultimate load of 133 kN. Therefore, the pre-planned fourth cycle cord and the buckling of first diagonal followed the failure. A sum-
was not performed. During the third load cycle, specimen CD-W mary of the failure is shown in (Fig. 8).
 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11 7

140
90
Cycle 3 80
120
Cycle 6
70
100
60
Load (kN)

Cycle 4

Load (kN)
80
Cycle 2 50

60 40 Cycle 2

30
40 Cycle 3
20 Cycle 2
Cycle 2 Cycle 4
20
10 Cycle 6
Cycle 1
0 0
0 10 20 30 40 50 60 70 0 5 10 15 20 25

Mid-span deflection (mm) Micro-strain

(a) CD-W load-deflection behaviour (b) CD-S pre-fatigue load –deflection behaviour

140
90
Cycle 8
CD-W cycle 2 Cycle 11
80 120 Cycle 9
CD-S cycle 6 Cycle 10
CD-S cycle 8
70 Cycle 11
100
60
Load (kN)
Cycle 9
Load (kN)

80
50

40 60
Cycle 10
30 CD-W Cycle 2 Cycle 8
40
20 CD-S Pre
fatigue cycle 6
20
10 CDS Post
fatigue cycle 8
0 0
0 5 10 15 20 25 0 10 20 30 40 50 60 70 80

Mid-span deflection (mm) Mid-spab deflection (mm)

(c) Comparisons of CD-S and CD-W (d) CD-S Post fatigue load –deflection behaviour
load –deflection behaviour

140
CD-S Cycle 11
120
CD-W cycle 3
100
Load (kN)

80

60

40

20 CD-W
CD-S
0
0 20 40 60 80
Mid-span deflection (mm)
(e) Comparison of CD-W and CD-S load-deflection
up to failure
Fig. 7. Load – deflection comparisons.

4.2. Load – strain behaviour before applying the 3rd cycle. The top chord exhibits elastic tensile
strain up to the 3rd load cycle until about 80 kN. This indicates the
Several strain gauges have been selected to generate load vs. neutral axis remains in the concrete deck above the OWSJ. As
max strain plots for each of the three cycles for test specimen expected, the bottom chord is showing completely positive and
CD-W. (Fig. 9a) shows the load vs. strain in the top and bottom elastic strain with little to no residual strain between cycles. The
chords of CD-W OWSJ during the three monotonic load cycles. strain behaviour becomes bilinear in the 3rd cycle when the load
The bottom chord strains are illustrated with solid lines whereas exceeds 90 kN and up to failure. The strain in the bottom and top
the top chord strains are represented by dotted lines. It can be seen cords at failure were 1600 and 500 micro-strain in tension and
that there are no residual strains in both the bottom and top cords compression, respectively. These strain values are below the
8 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11

(a) CD-S: Delamination of steel decking (b) CD-S: Shear connector at global failure

(c) CD-S: Buckling in top chord (d) CD-S: Floor deck at failure

(e) CD-W: Steel decking delamination (f) CD-W: top chord buckle
Fig. 8. Failure modes of test specimens.

yielding strains of 1750 micro-strain. This indicates that the failure be in and out of plane buckling of the OWSJ top cord and diagonals
was mainly due to the degradation of the composite action ‘‘de- as shown in (Fig. 8a to d). Furthermore, elastic linear behaviour can
lamination and slippage between the concrete deck, the corrugated be seen up to load levels of 90 and 100 kN for CD-W and CD-S,
sheets and the OWSJ”, as shown in (Fig. 8e and f), rather than yield- respectively. These load levels are well above the ultimate design
ing of the bottom or top cord. load for residential and industrial applications. It worth to mention
(Fig. 9b) shows the load vs. strain in the top and bottom chords that bilinear load-deflection behaviours were also observed with
of test specimen CD-S during the pre-fatigue load cycles. It can be reduced stiffness at these load levels as shown in (Fig. 7e).
noticed that the strain values in both the top and bottom cords are
far below the yielding even at load levels exceeding the ultimate
design loads. The load-strain relationships remain linear with very 4.3. Cross-sectional strain distribution
low residual tensile strains. (Fig. 9c) displays the top and bottom
chord strains during fatigue load cycles, measured during fatigue Identically, for a full and ideal composite floor section, the fail-
cycles number 100 and 49,900. The fatigue responses of the top ure mode must be initiated by yielding of the bottom cord of the
and bottom chords are still linear with maximum strain values of OWSJ while the top cord remains subjected to tension forces, as
600 micro-strains, which is about 30% of the yielding strains. shown in (Fig. 1c). This means that only one neutral axis is located
(Fig. 9d) compares the monotonic strain behaviour before and after within the concrete slab, indicating that no premature failure due
applying the fatigue load cycles. It can be concluded that specimen to buckling of the top cord would occur. Also, when the top cord
CD-S has about 100 microstrains strain accumulation (fatigue exhibits a small or insignificant compression force and the only
damage) due to fatigue loading. (Fig. 9e) shows comparisons of neutral axis is located outside the concrete deck, within the depth
the strain behaviour of specimen CD-W and CD-S up to failure. of the OWSJ, the failure should be initiated by yielding of the bot-
The strain in the bottom and top cords of CD-S at failure were tom cord in tension. This is the safest failure mode because the
1750 and 1500 tension and compression micro-strain, respectively. yielding of the bottom chord produces large, noticeable deflection.
These strain values are very close or below the yielding strains of This failure mode is comparable to the ductility requirements in
1750 micro-strain. The failure of specimen CD-S was also due to most design codes. However, in the case of non-composite action,
tension failure of the screws, delamination and slippage between each of the concrete deck and the OWSJ developed compression
the concrete deck, the corrugated sheets and the OWSJ followed and tensile stresses at the top and bottom surface of each material
 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11 9

80
140 Cycle 6 Top Cord
Cycle 3 Cycle 3 70 Cycle 6
120
60
100 Cycle 4
50
Load (kN)

Load (kN)
80 Cycle 2 Cycle 4
40
60 Cycle 2
30
40 20

20 Bottom Cord 10 Bottom Cord

Top Cord
0 0
-500 -250 0 250 500 750 1000 1250 1500 1750 -800 -600 -400 -200 0 200 400 600 800 1000
Strain (microstrain) Strain (microstrain)

(a) CD-WLoad -strain behaviour up to failure (b) CD-Spre-fatigue Load -strain behaviour

80
50 Top Cord
70 Cycle 6
Cycle 6
40 60
50,000 th fatigue cycle

Cycle 8
50
Load (kN)

Load (kN)
30 Cycle 8
40

20 30

20
10
10 Bottom Cord
Top Cord Bottom Cord
0 0
-800 -600 -400 -200 0 200 400 600 800 -1000 -800 -600 -400 -200 0 200 400 600 800 1000
Strain (microstrain) Strain (microstrain)
(c) CD-S:load vs. strain during fatigue cycles (d) CD-Scomparisons of pre and post fatigue
load -strain behaviour

140
CD-W
120
CD-S CD-W
100
CD-S
Load (kN)

80

60

40

20
Top Cord Bottom Cord
0
-2000 -1600 -1200 -800 -400 0 400 800 1200 1600 2000
Strain (microstrain)

(e) Comparisonof CD-W and CD-S load –strain

behaviour up to failure
Fig. 9. Load – strain comparisons.

and the strain profile will show two neutral axes, as shown in deck-joist connection at the top surface of the top cord and the bot-
(Fig. 1b). tom surface of the thick, 115 mm deep, discontinued part of the
In this section, Fig. 10 presents specimen’s CD-W and CD-S the concrete deck, as shown in (Fig. 10a). Test specimen CD-W cross
cross sectional strain profiles at service load levels. The graphs sectional strain profiles are presented in (Fig. 10b and c), at
show the measured strain values across the depth of the composite 40 kN and 70 kN. Both Fig. 10b and c show a linear, continuous
structure at the maximum load of the respective load cycle, i.e. strain distribution across the depth. The neutral axis is located
Each graph contains multiple gauges’ readings, through the depth within the concrete slab and the top chord developed tensile
from the top concrete surface, to the bottom of the OWSJs bottom strains in both cases, which confirms that puddle welds develop
chord. To simplify the interpretation of the data, two horizontal and maintain full composite action beyond the service load.
lines are plotted alongside the strain profile to help differentiate (Fig. 10d) shows the strain profiles of specimen CD-S before and
where the neutral axis(s) lie according to the profile of the compos- after fatigue loading at 40 kN. It can be noticed that there are
ite structure. The first line, the concrete line, represents the bottom two neutral axes and no compatibility in the strain values at the
level of the thin, 65 mm deep, continuous part of the concrete slab. shear plan. For both the pre and post-fatigue profiles, the concrete
The second line represents, shear plan, represents the level of the deck develops tensile strain just above the shear plan while below
10 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11

Strain (microstrain)
-800 -600 -400 -200 0 200 400 600 800
0
-500 -400 -300 -200 -100 0 100 200 300 400 500 0

Depth from top Concrete (mm)

-50
Concrete Line -50

Depth from Top of Concrete

-100

-150 Shear Line

-100

Deck (mm)
-200

-250 -150
-300

-350
-200
-400
-250
a) Strain profile setup (cross sectional profile) -300

-350

-400

b) CD-W strain profile at 40 kN

Strain (microstrain) Strain (microstrain)

-800 -600 -400 -200 0 200 400 600 800 -800 -600 -400 -200 0 200 400 600 800
0 0

Depth from Top Concrete (mm)

Depth from top Concrete (mm)

-50 -50

-100 -100
-150 -150
Post-fatigue
-200 -200
Pre-fatigue
-250
-250
-300
-300
-350
-350
-400
-400
c) CD-W strain profile at 70 kN d) CD -S pre and post fatigue strain profile at 40 kN

Fig. 10. Strain profile comparisons.

the shear plan; the top cord had compressive strains. This would 2. Puddle welds of 19 mm diameter showed a very promising
indicate partial or non-composite action. If the composite action performance in terms for ultimate capacity, stiffness/deflec-
is completely lost, the top and bottom cords should have the same tion and strain performances. The composite action was
strain values. However, the bottom cord tensile strain is about maintained until load levels that exceeds the service and
twice the top cord compressive strain in the pre-fatigue strain pro- ultimate load requirements. At service load level, the deflec-
file. This indicates that there is still some load sharing between the tion is less than 70% of the allowable limit as well as the
deck and the OWSJ due to partial composite action. However, after tension and compression strains are about 20% of the yield-
fatigue loading, both the top and bottom cords strains are equal ing strain.
which indicates that the composite action is completely compro- 3. Compared to the puddle-welds, the 6.25 mm diameter Hilti-
mised under fatigue. However, the strains remain well below the screws shear connectors provide lass composite action as indi-
1750 micro-strain yielding limit at service load level after fatigue. cated form the deflection and strain performance. However, it
still meets the ultimate and service requirements for residential
5. Summary and conclusions building applications.
4. Fatigue loads induced a significant damage in the stiffness and
The objective of this research is to determine if puddle welds composite action of composite OWSJ floor system with Hilti
and Hilti-screw pins can provide and maintain composite action screws of 6.25 mm diameter as shear connectors.
instead of shear stud connectors in composite concrete and OWSJ Future experimentations are still required to validate and inves-
floor system. This research includes construction, testing and anal- tigate the effect of different parameters such as Larger sizes of pud-
ysis of two full-scale composite floor decks; each is 2.4 m wide and dle welds and Hilti-screw, the number and layout of the shear
6.4 m long. The composite deck floor consists of 65/115 mm con- connectors as well as sustained and fatigue loads.
crete slab cast on top of 0.9 mm corrugated steel sheets. The deck
is supported in the transverse direction, over 1200 mm span, on
two OWSJ with an overhang of 600 mm on each side. The OWSJs Acknowledgments
are simply supported over 6.25 m span. The investigated parame-
ter is the shear connectors where puddle-welds and Hilti screw The authors would like to acknowledge the technical staff
pins were used. According to the analysis of test results, the follow- at the Structural Laboratory at The University of Windsor
ing conclusions can be obtained: for the technical expertise and support needed to fulfil the
experimental program. The Canadian Deck and Joist Inc. con-
1. The test results of composite OWSJ deck indicates that signifi- tribution to this research through providing the materials
cant level of composite action is achieved when puddle-welds and practical knowledge needed for this project is also
are used as shear connectors instead of shear studs. acknowledge.
 G. Merryfield et al. / Construction and Building Materials 125 (2016) 1–11 11

References [6] A. Brattland, L. Kennedy, Flexural tests of two full-scale composite trusses, Can.
J. Civil Eng. (1992) 279–295.
[7] A. Wang, A study on Composite end-plate connections with flexible tensile
[1] CAN/CSA-S16-09 (Design of Steel Structures), Canadian Institute of Steel
reinforcements and shear connectors, Can. J. Civil Eng. (2010) 1437–1450.
Construction, CISC, Markham, ON, Canada, 2009.
[8] N. Hedaoo, L. Gupta, N. Ronghe, Design of composite slabs with profiled steel
[2] Tata Steel Limited, 2015. The Principal of Composite Action. Retrieved
decking: a comparison between experimental and analytical studies, Int. J.
February 2015, from Tata Steel Europe: <www.tatasteelconstruction.com/en/
Adv. Eng. 3 (2012) 1.
reference/teaching-resources/architectural-teaching-resource/
[9] K.N. Lakshmikandhan, P. Sivakumar, R. Ravichandran, S.A. Jayachandran,
elements/composite-construction/the-principal-of-composite-action>.
Investigations on efficiently interfaced steel concrete composite deck slabs, J.
[3] SECB, T.S., 2013, March, Arc Spot Welding Steel Deck – A Primer. Retrieved
Struct. (2013) 1–10.
October 2014, from Structure Magazine: <www.structuremag.org/?p=918>.
[10] G. Merryfield, A. El Ragaby, F. Ghrib, New Shear Connector for Open Web Steel
[4] Hilti, 2015, June, Metal Construction Screws. Retrieved February 2015, from
Joist with Metal Deck and Concrete Slab Floor Systems (MASc thesis),
Hilti: <www.hilti.ca/screw-fastening/metal-construction-screws/sc-CLS_
University of Windsor, ON, Canada, 2015. 2015.
METAL_CONSTRUCTION_SCREWS>.
[11] NBCC, National Building Code of Canada, National Research Council of Canada,
[5] H. Robinson, E. Fahmy, The design of partially connected composite open-web
NRC, Ottawa, Canada, 2010, p. 2010.
joists, Can. J. Civil Eng. (1978) 611–614.