Engineering Structures 175 (2018) 467–482

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Experimental and numerical investigations of steel-polymer hybrid floor T

panels subjected to three-point bending
⁎
Jaeho Ryua, Yong Yeal Kimb, Man Woo Parkc, Sung-Won Yoond, Chang-Hwan Leee, Young K. Juf,
a
Department of Civil, Architectural and Environmental Engineering, The University of Texas at Austin, 10100 Burnet Road Building 24, Austin, TX 78758, USA
b
DongYang Structural Engineers Co., Ltd., 7 Beopwon-ro 11-gil Tower C Suite 1101, Songpa-gu, Seoul 05836, Republic of Korea
c
Steel Structure Business Unit, Dongbu Steel Co., Ltd., LG Seoulyeok Building, 98 Huam-ro, Jung-gu, Seoul 04637, Republic of Korea
d
School of Architecture, Seoul National University of Science & Technology, 232 Gongneung-ro, Nowon-gu, Seoul 01811, Republic of Korea
e
Department of Architectural Engineering, Pukyong National University, 45 Yongso-ro, Nam-gu, Busan 48513, Republic of Korea
f
School of Civil, Environmental, and Architectural Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Republic of Korea

A R T I C LE I N FO A B S T R A C T

Keywords: A new floor system for steel buildings was developed that can replace conventional concrete deck slab systems.
Composite floor The floor system is designed with a new type of composite panel with a polymeric material filling between the
Hybrid panel top and bottom steel plates. Its salient features are its light weight and simple installation that reduce structural
Polymeric material materials and shorten the construction period. Experiments with various independent variables were performed
Flexural behavior
to evaluate the flexural capacity of the proposed composite panel, and a finite element analysis was also con-
Composite interaction
ducted to examine the state of the stresses generated between the steel plate and the polymeric core. The test
Flexural test
Finite element analysis results showed that the proposed panel exhibited very ductile behavior and maintained its structural integrity
even after a maximum load. No other failure mode than the face yielding of the top and bottom steel plates was
observed. The bond strength between the polymeric material and the steel plates was confirmed to be sufficient,
even without any special surface treatment or any additional shear connectors, to maintain the stability of the
proposed panel and to resist the forces generated at the interface between the two materials. Design equations
for predicting the flexural strength and stiffness of the proposed panel were proposed, and its suitability was
verified. Additionally, the experimentally tested efficient methods for field applications of the proposed panel
regarding cutting and joining were presented in this paper.

1. Introduction
A concrete slab system has been the most common floor system for
buildings because of its convenience in controlling the volume forma-
tion in a large surface area. However, installation of the formwork and
curing of poured concrete are factors that prolong the construction
period and increase field work. The precast concrete (PC) floor system
was introduced to overcome such disadvantages and to save time as
prefabricated members are installed at the construction site. However,
to maintain the structural integrity of the entire system, skilled work-
manship is required for precise work on the joints, and a concrete
topping is necessary. Moreover, additional costs for PC production,
transport to the construction site, and lifting the PC members into place
are directly associated with higher construction costs. Owing to such
disadvantages, the PC floor system has not yet been widely used.
There is another floor system for buildings in which planar members
such as panels are placed on the structural beam and then connected to
each other to form a floor. In particular, sandwich panels consisting of
two thin skin layers and a low density core have excellent structural
efficiency and are suitable for flooring. A high-strength and high-stiff-
ness material is mainly used for the skin layers, and the skins provide
bending resistance and flexural stiffness to the panel. The core is
bonded to the skin layers and supports the skin layers to behave as a
continuum.
Various types of sandwich panels have been developed thus far and
are widely used in construction, shipbuilding, aircraft construction, and
other branches of industry. Sandwich panels are also characterized by
their diversity of materials and shapes according to their purposes and
functions [1,2]. In the aircraft construction and shipbuilding industries,
various types of sandwich panels have been mainly developed as the
exterior panels for aircrafts or ships. Lightweight metals such as alu-
minum are widely used as skin layers, and honeycomb and truss core

⁎
Corresponding author.
E-mail addresses: jryu@austin.utexas.edu (J. Ryu), yykim@dysec.co.kr (Y.Y. Kim), imwpark@dongbu.com (M.W. Park), swyoon@seoultech.ac.kr (S.-W. Yoon),
chlee@pknu.ac.kr (C.-H. Lee), tallsite@korea.ac.kr (Y.K. Ju).

https://doi.org/10.1016/j.engstruct.2018.08.030
Received 9 September 2017; Received in revised form 3 July 2018; Accepted 11 August 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
J. Ryu et al.
Fig. 1. Components of the iFLASH floor panel: (a) Basic core; (b) Core with a wire mesh; (c) Core with a wire mesh and hollow core balls.
structures are employed for efficient impact load absorption [3–7]. A
recent trend observed in all branches of industry is the replacement of
conventional core materials with metal foams, and metal-foam sand-
wich panels are being studied in various application fields [8–10].
In the construction sector, research on steel-concrete-steel sandwich
panels has been performed extensively since the late 1980s. The steel
plates fulfill both functions of serving as a formwork and resisting the
bending moment, thus demonstrating excellent production and struc-
tural efficiency. However, because of the weak bond strength between
the concrete and steel plates, shear connectors also need to be used to
ensure integrated behavior between the two materials as composite
structural elements. Oduyemi et al. [11] and Robert et al. [12] at-
tempted to increase the bond strength between the steel and concrete
by using welded stud connectors, and Richard Liew et al. [13] proposed
using a J-hook as shear connectors in the sandwich panel. Tata Steel
Europe Ltd. [14] developed the Bi-Steel panel in which steel round bars
used as shear connectors are directly connected to the two steel layers
to obtain the effective shear transfer. Leekitwattana [15] integrated bi-
directional corrugated strips with a core material to enhance the in-
teraction between the steel plates and the concrete.
An increasing number of studies in which the metal skins were re-
placed with glass fiber reinforced polymers (GFRPs) have been con-
ducted recently by several researchers [16–20,30]. Reis and Rizkalla
[16] introduced and studied a sandwich panel in which GFRP laminates
and a polyurethane foam were used as the skin layers and core re-
spectively, and top and bottom skin layers are connected together with
through-thickness fibers to overcome delamination problems. Fam and
Sharaf [17] investigated the effect of various configurations of internal
and exterior GFRP ribs on flexural performance of a GFRP sandwich
panel with low density polyurethane core. Awad et al. [18] developed a
new type of GFRP sandwich panel using a phenolic foam as a core
material. In this study, experimental and numerical analyses were
conducted to find an analytical model for the phenolic core material.
These sandwich panels with GFRP skins have been largely used for
flooring system of timber structures because the panels can be easily
connected to timber beams [18,20]. Therefore, to extend application of
the GFRP sandwich panels to floor system of a building constructed
with steel and concrete, it is necessary to investigate the connection
methods for structural integrity between steel or concrete beams and
GFRP floor panels.
Along with increasing interest in modular construction technolo-
gies, sandwich panels using plywood as skin layers and expanded
468
Engineering Structures 175 (2018) 467–482
polystyrene foam (EPS), extruded polystyrene foam (XPS), or poly-
urethane foam as the core have been developed and used for low-rise
housing buildings in some countries, especially in the USA [21]. They
are superior to metal-skin sandwich panels in terms of the unit product
cost and insulation performance and can also be used as wall members.
However, these panels are primarily targeted at modular housing with
two or three floors and are thus not suitable for high-rise buildings.
In this research, a new composite floor system was proposed that
can replace the concrete deck slab systems of steel structures and has
excellent workability. For field application of the developed floor
system, a series of studies have been conducted. As a first step in the
development of the floor system, this paper will focus on the static
response of the developed composite floor panel subjected to three-
point bending. The flexural capacity of the proposed panel was tested in
experiments with various independent variables that should be con-
sidered at construction sites. A finite element analysis (FEA) was also
carried out to verify the stress level at the interface between the skin
layers and the core material. Additionally, design equations for calcu-
lating the flexural strength and stiffness of the newly developed panel
were established.
2. iFLASH System
2.1. Concept
The composite floor system developed in this study is an assembly
system of composite panels consisting of top and bottom steel plates
with infilled polymeric material (Fig. 1(a)). A wire mesh for strength-
ening flexural strength under fire conditions and hollow core balls for
reducing the polymer weight can be added to the core (Figs. 1(b) and
(c)). The proposed system was named the iFLASH System, an acronym
for innovative, fireproof, lightweight, absorbed, shallow, and hybrid
system. Owing to the high length-to-thickness and strength-to-weight
ratios, the floor height and material requirements can be reduced, and
the construction time can be reduced owing to the simple in-situ in-
stallation and high workability. Moreover, the core material has an
elastomeric function; thus, it exhibits high vibration and noise ab-
sorption capacities, and its high thermal storage performance endows it
with excellent insulation and fire resistance capacities. Another salient
feature of the proposed composite panel is the recycling capacities of
both the steel plate and core so that no waste treatment is necessary.
The iFLASH System allows for modular manufacturing that is
J. Ryu et al.
Fig. 2. Construction process of the iFLASH System.
applicable to both new construction and renovation (Fig. 2). It can be
applied to a new building by lifting panels (∼20 m2 each) into place on
the frame using a crane and connecting them to the frame. On a re-
novation site, where a lifting device cannot be installed, downsized
modules (1 m2) are used for easy transport employing labor. The
iFLASH floor panels can be connected to the beam frame by welding or
bolting, and the connection between the floor panel and beam is as-
sumed to be a simple connection. Depending on the long-to-short span
ratio, the panel behaves as either a one-way or a two-way slab. For the
design of the proposed floor system, the following requirements should
be mainly considered: flexural capacity of the floor panel, flexural ca-
pacity of the composite beam consisting of the floor panel and steel
beam, floor diaphragm action, and floor vibrations.
2.2. Design of iFLASH floor panel for flexure
The nominal flexural strength of the iFLASH floor panel can be
determined based on the distribution of the flexural stresses acting on a
cross section at the ultimate limit state [22–26]. The stress distribution
plotted in Fig. 3 was established under the assumptions that debonding
failure does not occur and the steel/polymer integrated behavior lasts
until the ultimate limit state.
For the polymeric material, it maintains a linear relationship be-
tween stress and strain up to relatively large strains (4–5%) as com-
pared to steel. Accordingly, as shown in Fig. 3, the core material re-
mains linearly elastic at the ultimate limit state while the entire cross
section of the top and bottom steel plates is fully yielded. In addition,
the stress–strain relationships of the polymeric material in tension and
compression are significantly different. Especially, the modulus of
elasticity in tension is approximately three times larger than that in
Fig. 3. Strain and stress distributions at the ultimate limit state.
469
Engineering Structures 175 (2018) 467–482
compression. Because of these essential features of the polymeric ma-
terial, the bending stresses at the ultimate limit state for the core ma-
terial in tension or compression should be determined by multiplying
the strains at the ultimate limit state by the corresponding modulus of
elasticity.
Cs = Ts = As Fys (1)
1 1
Cc (or Tc ) = Ac, c σc, c _ max (or Ac, t σc, t _ max )
2 2 (2)
Mn = Cs d1 + Cc d2 + Tc d3 + Ts d4 (3)
Eqs. (1) and (2) are used to calculate the resultant tensile and
compressive forces for bending moment, and the nominal flexural
strength (Mn) of the proposed panel can be calculated with Eq. (3). The
symbols used in Fig. 3 and Eqs. (1)–(3) are defined as follows: Ac,c = the
cross-sectional area of the core in compression; Ac,t = the cross-sec-
tional area of the core in tension; As = the cross-sectional area of the
top or bottom steel plate; Cc = the resultant compressive force applied
to the core; Cs = the resultant compressive force applied to the top steel
plate; Ec,c = the modulus of elasticity of the core in compression;
Ec,t = the modulus of elasticity of the core in tension; Fys = the yield
strength of the steel; Tc = the resultant tensile force applied to the core;
Ts = the resultant tensile force applied to the bottom steel plate;
di = the distance between the neutral axis and each resultant force; tbsp
(or tsp) = the thickness of the bottom or top steel plate; tc = the thickness
of the core; ttotal = the total thickness of the iFLASH floor panel; εs = the
strain of the top or bottom steel plate; εys = the yield strain of the steel;
σc,c_max = the maximum compressive bending stress of the core; and
σc,t_max = the maximum tensile bending stress of the core.
With regard to the serviceability limit state, the floor deflection and
vibration capacity are commonly considered as the most important
evaluation items. Given that the floor vibration capacity is closely as-
sociated with the deflection [27], accurate calculation of deflection is
essential. The total deflection induced by the vertical loads on a typical
sandwich beam is usually determined as the sum of the deflections due
to the bending and shear deformations [28]. However, in the case of the
iFLASH floor panel, because the core material is surrounded by the
perimeter steel plates and integrated with them (Fig. 4), the shear de-
formations of the core are restrained. Accordingly, the shear deforma-
tions of the iFLASH floor panel can be neglected, and the total deflec-
tion (yiF) of the iFLASH floor panel can be calculated using Eq. (4) based
on the Euler–Bernoulli beam theory.
d 2yiF M
=
dx 2 Es Itr (4)
where M = the bending moment induced by total service load; Es = the
modulus of elasticity for the steel; and Itr = the moment of inertia for
the transformed section of the iFLASH floor panel. The transformed
moment of inertia for the iFLASH floor panel can be determined using
the modular ratio, n (= Es/Ec,c or Es/Ec,t), based on the transformed-
section method [29].
J. Ryu et al. Engineering Structures 175 (2018) 467–482

Table 1
Mechanical and thermal properties of solidified polymeric material.
Properties Values

3
Density (kg/m ) 1178
Tensile strength (MPa) 31.4
Modulus of elasticity in tension (MPa) 1277
Compressive strength (MPa) 23.1
Modulus of elasticity in compression (MPa) 461
Poisson's ratio 0.39

Thermal diffusivity (mm2/s) 25 °C 0.102

100 °C 0.083
200 °C 0.076

Thermal conductivity (W/m k) 25 °C 0.191

100 °C 0.195
Fig. 4. Prefabricated iFLASH floor panel. 200 °C 0.187

Specific heat (J/g K) 25 °C 1.613

The iFLASH floor panel can be cut at any location within the panel
to fit the floor space according to the floor framing plan of a building. In
this case, if the cut surface in the panel is properly reinforced with a
steel plate, the reinforced panel can be considered as a whole panel
without damage, and the flexural strength and stiffness of the panel can
be determined in the same manner as described above.
3. Investigation into material properties
3.1. Infilled polymeric material
100 °C 2.031
200 °C 2.127
failure as shown in Fig. 5(b). The average compressive strength and the
modulus of elasticity in compression were 23.1 MPa and 461 MPa, re-
spectively.
The laser flash analysis was used to measure the thermal properties
of the polymer such as thermal diffusivity, thermal conductivity, and
specific heat. The analysis results are summarized in Table 1 together
with the mechanical properties of the polymeric material.

A high density polymeric material is used as the core which was
developed by Chemtros Co., Ltd. for the iFLASH System. This polymer is
formed by reacting a diisocyanate with a polyol mixture and is a yel-
lowish-brown viscous liquid that is completely solidified after 24 h at
room temperature. In general, the strength and rigidity of a synthetic
polymer tend to increase as the polymer density increases [18,30]. The
developed polymeric material has a density of 1178 kg/m3 which is
high enough for use in structural applications.
Three tensile test coupons were prepared and tested in accordance
with ASTM D638-14 [31]. Fig. 5(a) shows a graph representing the
experiment results for the tensile strength of the developed polymer.
Similar to the steel plate, the polymeric material in tension exhibited a
region where the loading increases linearly and a post-yield region
where the strain increases without an increase in the loading. The
average tensile strength of the polymer was 31.4 MPa and the modulus
of elasticity in tension was 1277 MPa.
The uniaxial compression test for the developed polymer was con-
ducted according to the ASTM D695-15 [32]. Three test specimens were
prepared and tested. The specimen was in the form of a right rectan-
gular prism, the sizes of which were 12.7 by 12.7 by 25.4 mm. The
polymeric material in compression exhibited a nonlinear behavior up to
3.2. Steel plate
The grade of steel plates used as the skins in the iFLASH System was
SS400 (Fys = 235 MPa). Fabrication of three tensile test coupons and
the tests were performed in compliance with the standards specified in
KS B 0801-2007 and KS B 0802-2003, respectively [33,34]. As a result,
the average yield strength, the average ultimate strength, and the
modulus of elasticity of the steel used were verified to be 309.6 MPa,
462.4 MPa, and 205,766 MPa, respectively.
3.3. Tensile bond strength between steel plate and polymeric material
The proposed panel is expected to incur debonding failure at the
interface between the top steel plate and the infilled polymeric material
at the ultimate limit state. For this reason, it is important to secure the
proper bond strength between the two materials. The bond strength
between the steel plates and the infilled polymeric material can vary
depending on the degree of the surface treatment of the steel plate. To
verify the above statement, the tensile bond strength between the two
materials depending on the degree of surface treatment of the steel
plate was tested using the PosiTest Pull-off Adhesion Tester AT-A [35].

Fig. 5. Mechanical properties of polymeric material: (a) Tensile strength; (b) Compressive strength.

470
J. Ryu et al.
Fig. 6. Pull-off adhesion test.
Three degrees of surface treatment were designed for the steel plates: a
steel plate without surface treatment, a SP10 surface-treated steel plate,
and a SP5 surface-treated steel plate. Surface treatment was performed
in compliance with the standards specified by the Society for Protective
Coatings (SSPC) [36,37]. SP10 represents “near-white blast cleaning,”
defined to be the state of a steel surface with remaining post-blast re-
sidues of less than 5% of the blast-treated steel surface. SP5 represents
“white metal blast cleaning” and is defined to be a state free of all
stains.
The liquid polymeric material was poured on a steel plate to a
thickness of 20 mm and solidified, and a pull-off pressure was applied to
its surface using a steel pin (Ø 20 mm) fixed into the coupling of the
tester until debonding failure occurs (Fig. 6). This test was performed in
quintuplicate for each steel plate type, and the average of three median
values after excluding the maximum and minimum values was used as
the representative value, as summarized in Table 2. The average tensile
bond strengths of the specimens without surface treatment, with SP10
surface-treated, and SP5 surface-treated steel plates were measured to
be 6.34 MPa, 6.58 MPa, and 7.21 MPa, respectively. Compared with the
average bond strength of the specimen without surface treatment, the
SP10 surface-treated and SP5 surface-treated specimens exhibited 3.7%
and 13.7% higher average bond strengths, respectively.
4. Experimental program
4.1. Specimens
A test plan for a series of experiments was established to evaluate
the flexural capacity of the composite panel developed in this study. A
total of 12 specimens were fabricated and they were divided into four
groups depending on the variable characteristics. Table 3 and Fig. 7
present the details of the test specimens.
In Group 1 (n = 3), the degree of surface treatment of the steel plate
was considered as a primary variable that can influence the bond
strength between the steel plate and the polymeric material. Three test
specimens having different steel surface treatments were fabricated: a
specimen with steel plates without surface treatment (FP-NB), a spe-
cimen with SP10 surface-treated steel plates (FP-NWB), and a specimen
with SP5 surface-treated steel plates (FP-WB). All test specimens mea-
sure 500 mm × 2500 mm × 25 mm (width × length × thickness), and
the steel plate is 2.5 mm thick. The skin layers of the panel were fab-
ricated by cutting a sheet of steel into 525 mm × 2500 mm plates and
Table 2
Tensile bond strength of polymeric material depending on the degree of steel
surface treatment.
Blast cleaning Tensile bond strength (MPa)
1 2 3 Average
No Blast 6.69 6.17 6.15 6.34
SP 10 6.97 6.68 6.10 6.58
SP 5 7.81 7.52 6.29 7.21
471
Engineering Structures 175 (2018) 467–482
bending the plates at a point 12.5 mm from each widthwise end at right
angles (90°). Two channel-shaped steel plates thus prepared and two
end steel plates for headers were then welded to form the frame of the
composite panel, and a test specimen was fabricated by injecting the
polymeric material to fill the core. Steel solid bars with a rectangular
cross section (100 mm × 50 mm × 20 mm) were inserted at the sup-
ports and load points to prevent a local failure induced by stress con-
centration effect (Fig. 7(a)).
The second test specimen group consisted of specimens (n = 4) in
which shear connectors were installed in different modes to restrain the
relative motion between the top and bottom steel plates and to control
the shear deformations of the core material. In addition, the shear
connectors placed inside the panel contribute to transferring the long-
itudinal shear force occurring between two materials, especially in
cases when premature debonding failure occurs due to the diminished
bond strength caused by mill scales or stains on an untreated steel
surface. Inside each of the four test specimens, box-type or circular-
solid-type shear connectors were installed at 250 mm or 500 mm in-
tervals. As shown in Fig. 7(b), the box-type shear connector is a 200-
mm-long tube with a square cross section (20 mm × 20 mm). The cir-
cular-solid-type shear connector is a steel cylinder with a diameter of
20 mm (Fig. 7(c)).
In Group 3 and 4, an additional test plan was established to ex-
perimentally identify the optimal application method considering the
installation-related variables, such as cutting and joining of the panels.
The main concerns for the Group 3 test specimens were the reinforce-
ment or non-reinforcement of the cut surface and the method of at-
taching a reinforcing steel plate. If the last row of the composite floor
panels does not fit exactly into the remaining space, they need to be cut
to fit the space. The flexural capacity of a panel with a cut surface re-
inforced with a steel plate should be assessed for such cases. The most
reliable method of attaching the reinforcing steel plate is welding.
However, skilled workers are required for welding and it is costly. To
save the fabrication cost, a new reinforcing method was proposed in
this study. The reinforcing steel plate was bonded using a liquid-type
acrylic adhesive consisting of two blue and intense-red visco-liquids
with a 21 MPa bond strength. This adhesive was developed by
Chemtros Co., Ltd. As shown in Fig. 7(d), the third group of test spe-
cimens (n = 3) was cut at a point 70 mm from the widthwise end; the
cut surface of one of the specimens was left unreinforced (FP-NB-C),
and the surfaces of the remaining two were reinforced with a steel plate
(FP-NB-CB, FP-WB-CB). One of the specimens with a reinforcing steel
plate (FP-WB-CB) was fabricated by using SP5 surface-treated steel
plates to verify the extent of damages to the cut surface caused by the
cutting process depending on the degree of steel surface treatment.
The test specimens in the Group 4 (n = 2) were designed to test the
welding method for in-situ panel installation. Continuous welding is the
surest way to ensure the integrity and continuity between the panels,
but it is time-consuming and labor-intensive. Therefore, a test was
carried out to determine whether intermittent welds can have the same
effect as continuous welds. One test specimen was used for the con-
tinuous welding of three panels on the top and bottom steel plates (FP-
NB-CWC, Fig. 7(e)) and the other specimen was used for 60 mm in-
termittent welding on the top and bottom steel plates of the supports
and load point (FP-NB-IWC, Fig. 7(e)). Partial penetration groove welds
were used as a method to connect the panels. The connected parts were
groove welded with no edge preparation and no gap. The steel face
plates of the panels were only connected to each other through the
groove welds. For a conservative design, the effective throat of the
welds (a) and the shear force transferred between the panels were as-
sumed to be 1.0 mm and 20 kN, respectively. The required minimum
length of the intermittent welding (l) can be determined in accordance
with KBC 2016 [38], as follows:
ϕRn = 0.75·Aw ·Fw = 0.75·(le ·a)·(0.6·Fuw ) ⩽ 20 kN (5)
J. Ryu et al. Engineering Structures 175 (2018) 467–482

Table 3
Specimen details.
Group Specimensa Variables

Blast cleaning Cutting and bonding a reinforcing steel Shear connector Interval between shear Welding method between panels
plate connectors

1 FP-NB No blast-cleaned – – – –
FP-NWB SP10 – – – –
FP-WB SP5 – – – –

2 FP-NB-BS500 No blast-cleaned – Box tube type 500 mm –

FP-NB-BS250 No blast-cleaned – 250 mm –
FP-NB-CS500 No blast-cleaned – Circular solid type 500 mm –
FP-NB-CS250 No blast-cleaned – 250 mm –

3 FP-NB-C No blast-cleaned Only Cutting – – –

FP-NB-CB No blast-cleaned Cutting & Bonding – – –
FP-WB-CB SP5 – – –

4 FP-NB-CWC No blast-cleaned – – – Continuous

FP-NB-IWC No blast-cleaned – – – Intermittent

a
FP – iFLASH Panel; NB – No Blast; NWB – Near White Blast; WB – White Blast; C – Cutting; CB – Cutting and Bonding; BS – Box-type Shear connector; CS –
Circular-solid-type Shear connector; CWC – Continuously Welded Connection; IWC – Intermittently Welded Connection.

l = le + 2a
where Aw = the effective area of the weld; Fw = the nominal strength of
the weld metal; Fuw = the tensile strength of the weld metal (400 MPa);
and le = the effective length of the weld. From Eqs. (5) and (6), the
required length of the intermittent welding was 57.5 mm and ap-
proximated as 60 mm for workability.
4.2. Test setup
Experiments were performed using a 5000 kN universal testing
machine. Loads were applied at the displacement-controlled rate of
4–30 mm/min as one-point monotonic loading at the midspan using a
500 mm long loading beam [39,40]. Roller-based simple supports were
used, as shown in Fig. 8, with the distance between the supports being
2000 mm and the shear span-to-depth ratio was 40, suitable for flexural
capacity evaluation.
A 1000 mm capacity linear variable displacement transducer
(LVDT) was installed at the bottom steel plate of the test specimen to
measure the deflection at the center, and strain gauges were attached to
the specimen on the widthwise centerline at a point 125 mm from the
load point to measure the strains on the top and bottom steel plates.
5. Test results and discussions
5.1. Failure modes
Most of the Group 1 and 2 test specimens (FP-NB, FP-WB, FP-NB-
BS500, and FP-NB-CS250) displayed no other failure mode than the
face yielding of the top and bottom steel plates. These test specimens
maintained the cross-sectional integrity until the end of the experiment
and exhibited stable ductile behavior even after a maximum load.
Fig. 9(a) shows the failure mode of FP-NB. For FP-NB-NWB, the
wrinkling of top steel plate was observed around the load point as
shown in Fig. 9(b). However, this failure mode was not critical to the
flexural performance of the test specimen. Figs. 9(c) and (d) illustrate
the final deformed shapes of FP-NB-BS250 and FP-NB-CS500, respec-
tively. For these two specimens, out-of-plane buckling of the top steel
plate occurred with the expansion of debonding failure at the interface
between the top steel plate and the polymeric core. The local debonding
area was widely formed around the load point compared with other test
specimens. This was analyzed to be due to an inadequate curing of the
polymeric material exposed to low temperatures during the test spe-
cimen fabrication in winter, resulting in incomplete manifestation of
the bond performance.
(6) Figs. 9(e)–(g) show failure modes of the Group 3 test specimens that
were subjected to panel cutting experiments in order to check the
suitability of their field applications. The panel of FP-NB-C, which was
not reinforced with a side plate, deflected laterally with twist of the
cross section under vertical loading owing to the difference in the
stiffness between the left and right sides (Fig. 9(e)). This caused pre-
mature debonding of the top steel plate from the core. In contrast, FP-
NB-CB and FP-WB-CB that were reinforced with a steel plate at the cut
surface exhibited stable flexural behaviors without twist in cross-sec-
tional shape. As shown in Figs. 9(f) and (g), the test specimens finally
failed by out-of-plane buckling of the top steel plate resulting from
debonding around the load point. For these specimens, the reinforcing
side steel plate did not debond from the panel until the experiment
ended and contributed to ensure the stability of the panel. This in-
dicates that application of the proposed adhesive can be used as an
effective means of rigidly attaching the reinforcing steel plate to the cut
floor panel.
The failure mode of FP-NB-CWC, fabricated by continuous welding
between panels, was the wrinkling of top steel plates around the load
point (Fig. 9(h)). The debonding area did not widely appear because the
forces between the panels were uniformly transferred through the
continuous welding regions without local stress concentrations and the
three panels behaved as a single member. For FP-NB-IWC, the test
specimen with intermittent welds at the supports and load point, an
intense out-of-plane deformation of the top steel plates occurred around
the weld zones (Fig. 9(i)). This was analyzed to be due to a severe
torsional deformation of the side panels induced by the concentrated
load transfer when the load applied to the central panel is transferred to
the both side panels through the short weld zones.
5.2. Load–deflection relationships
Fig. 10(a) shows a graph representing the load–deflection re-
lationships of the Group 1 test specimens. FP-NB (steel plate without
surface treatment) showed an initial region where the load increased
linearly, and when the strains of the top and bottom steel plates reached
the yield strain, a rapid decrease in the stiffness was observed. After
reaching the yield load (17.5 kN), the specimen exhibited ductile be-
havior, and the maximum load (21.6 kN) was observed at the deflection
point of 320 mm. Both surface-treated test specimens, FP-NWB and FP-
WB, showed similar structural behavior to that of the untreated spe-
cimen FP-NB, with the maximum loads of 22.0 kN and 22.1 kN at the
deflection points of 263 mm and 295 mm, respectively. All three spe-
cimens in the flexural capacity test showed insignificant differences of

472
J. Ryu et al.
Fig. 7. Test specimens: (a) FP-NB & FP-NWB & FP-WB; (b) FP-NB-BS250; (c) FP-NB-CS250; (d) FP-NB-C & FP-NB-CB; (e) FP-NB-CWC & FP-NB-IWC (dimensions in
mm).
approximately 2% in the maximum load. This is because the bond
strength between the steel plate without surface treatment and the
polymeric material is not significantly lower than those between the
surface-treated steel plates and the polymeric material, as it can be seen
in the tensile bond strength test results. None of the three test speci-
mens incurred a rapid decrease in loading after reaching the maximum
load, and they exhibited additional ductile behavior by demonstrating a
load carrying capacity at the 90% level of the maximum load.
The Group 2 test specimens showed very similar structural beha-
viors to those of the Group 1 test specimens, as can be seen in the
473
Engineering Structures 175 (2018) 467–482
load–deflection curves in Fig. 10(b). This means that the perimeter steel
plates provide a sufficient restraint against the shear deformations of
the polymeric core even without the shear connectors. The test speci-
mens, except for FP-NB-BS250, showed an average maximum load of
21.9 kN, which is the same resistance level as the Group 1 test speci-
mens. However, unlike the other test specimens, FP-NB-BS250 and FP-
NB-CS500 showed a sharp drop in the load due to premature debonding
failure. From the test results for the Group 1 and 2 test specimens, it is
demonstrated that the flexural capacity of FP-NB was the same as those
of the other specimens even without any special surface treatment or
J. Ryu et al.
Fig. 7. (continued)
any additional shear connectors. This is a very meaningful result be-
cause the fabrication process of the proposed panel can be greatly
simplified in the shop, and it leads to lower production costs.
Fig. 10(c) shows a comparison of the test results of the Group 3 test
specimens. The test specimen FP-NB-C showed a rapid decrease in
loading after the maximum load because of debonding of the top steel
plate from the core resulting from the lateral-torsional behavior of the
specimen. The maximum load of FP-NB-C was 15.7 kN. In contrast, FP-
NB-CB and FP-WB-CB showed ductile behavior of over 250 mm until
they reached their maximum loads, similar to the Group 1 and 2 test
specimens. The maximum loads of FP-NB-CB and FP-WB-CB were
17.6 kN and 18.6 kN, respectively, which were 12% and 18% higher
474
Engineering Structures 175 (2018) 467–482
than that of FP-NB-C. However, FP-WB-CB, the blast-treated test spe-
cimen, outperformed FP-NB-CB by 5.7% with regard to the maximum
load and showed more stable post-maximum load behavior. This de-
monstrates the positive effect of the blast treatment in reducing the
influence of microscale damage to the cut surface caused by vibration
and friction while cutting.
Finally, the load–deflection curves of the Group 4 test specimens
were compared and are presented in Fig. 10(d). FP-NB-CWC showed
similar behavior to those of a single panel, demonstrating a three times
higher maximum load than a single panel (66.2 kN vs. 21.9 kN). In
contrast, because of premature debonding failure of the top steel plates,
FP-NB-IWC showed a rapid decrease in loading after the maximum load
J. Ryu et al. Engineering Structures 175 (2018) 467–482

Fig. 8. Test setup (dimensions in mm).

Fig. 9. Failure modes: (a) FP-NB; (b) FP-NWB; (c) FP-NB-BS250; (d) FP-NB-CS500; (e) FP-NB-C; (f) FP-NB-CB; (g) FP-WB-CB; (h) FP-NB-CWC; and (i) FP-NB-IWC.

475
J. Ryu et al.
Fig. 10. Load–deflection relationships: (a) Group 1 test specimens; (b) Group 2 test specimens; (c) Group 3 test specimens; (d) Group 4 test specimens.
(55.0 kN) without exhibiting ductile behavior. Based on the preceding
discussions of failure modes and load–deflection relationships, it can be
concluded that continuous welding is a more preferable connection
method in cases where the iFLASH panels are connected side by side.
5.3. Strain distributions
As illustrated in Fig. 8, the longitudinal strains were measured on
the outer surfaces of top and bottom steel plates, and the strains ob-
tained from the gauges symmetrically installed on the left and right
sides of test specimen were averaged. Fig. 11 shows the strain dis-
tributions at a point 125 mm from midspan under different load stages.
In the case of FP-NB, because the polymeric material has a much
larger modulus of elasticity in tension than in compression, the neutral
axis was located below the centerline of the cross section until the
applied load reached 20 kN (Fig. 11(a)). After 20 kN, the tensile strain
at the extreme fiber of bottom steel plate showed a rapid increase. This
is because the axial tensile force, induced by the slip resistance between
the test specimen excessively deflected and the support, occurred along
the longitudinal axis of the test specimen. Accordingly, the neutral axis
moved toward the centerline of the cross section until the maximum
load. The strain distributions of the other test specimens without a wide
debonding failure (FP-NWB, FP-WB, FP-NB-BS500, FP-NB-CS250) were
very similar to those of FP-NB.
FP-NB-BS250 and FP-NB-CS500 exhibited similar strain distribu-
tions. Fig. 11(b) shows the strain distributions of FP-NB-BS250. Because
of the large out-of-plane deformation of the top steel plate, the tensile
strain of the bottom steel plate increased more rapidly than the com-
pressive stain of the top steel plate. As a result, the neutral axis of the
476
Engineering Structures 175 (2018) 467–482
cross section gradually moved toward the compressive edge as de-
bonding area of the top steel plate increased.
The strain distributions of FP-NB-C and FP-WB-CB are plotted in
Figs. 11(c) and (d), respectively. The strains at the maximum load of FP-
NB-C were only about half as large as those of FP-WB-CB due to early
lateral-torsional failure. In the case of FP-WB-CB, the strain values at
the maximum load were −1.55% for top steel plate and 1.17% for
bottom steel plate, respectively. These values are not significantly dif-
ferent from the experimental results of the Group 1 and 2 test speci-
mens. This means that this test specimen underwent a sufficiently large
deformation in flexure, maintaining its stability owing to the reinfor-
cing side steel plate.
In the case of the Group 4 test specimens, as the applied load in-
creased, the debonding area of the top steel plate gradually enlarged.
Thereafter, out-of-plane deformation of the top steel plate took place,
and this failure mechanism caused a small increment in the compressive
strain of the top steel plate. In contrast, the tensile strain of the bottom
steel plate showed a large increase until the maximum load.
Consequently, the position of the neutral axis gradually moved upward
in the cross section (Figs. 11(e) and (f)).
5.4. Flexural strengths and stiffnesses
The maximum loads of all test specimens measured in the experi-
ments were compared with the calculated values obtained using Eqs.
(1)–(3) and are presented in Table 4. The average of the actual thick-
nesses measured on the test specimens was used for calculating the
theoretical strengths: 2.4 mm for the steel-plate thickness and 19.6 mm
for the polymeric core. The maximum compressive and tensile bending
J. Ryu et al.
Fig. 11. Strain distributions: (a) FP-NB; (b) FP-NB-BS250; (c) FP-NB-C; (d) FP-WB-CB; (e) FP-NB-CWC; (f) FP-NB-IWC.
stresses at the extreme fiber of the polymeric core (σc,c_max and σc,t_max)
were determined based on the strain distributions of the test specimens:
6.0 MPa for σc,c_max and 16.6 MPa for σc,t_max. When considering that the
strains were measured at a point 125 mm from midspan, these stresses
can be used as the reasonable values for a conservative estimate of the
flexural strength of the proposed panel.
The maximum loads of all test specimens of Groups 1 and 2, except
for FP-NB-BS250 which incurred premature failure, were over 20%
higher than the theoretically calculated value of 17.7 kN. Compared
with the theoretical ultimate strength of the test specimens of Group 3,
FP-NB-C which was not reinforced with a side steel plate exhibited only
a 4% higher bending resistance while the other two test specimens, FP-
NB-CB and FP-WB-CB, exhibited 15% and 22% higher flexural
477
Engineering Structures 175 (2018) 467–482
strengths, respectively. In the case of the test specimens of Group 4, the
calculated strength was 53.1 kN. Although the flexural strength of FP-
NB-CWC, the specimen with continuous welds, demonstrated a 25%
higher value, FP-NB-IWC, the specimen with intermittent welds,
showed only a 4% higher flexural strength. From these comparisons, it
can be confirmed that the flexural strengths predicted by Eqs. (1)–(3) is
about 80% of the actual performance, and these values are sufficiently
conservative to be used as a nominal strength of the proposed panel.
The initial elastic stiffness (Ke), yield load (Pye), and yield dis-
placement (δy) are defined as shown in Fig. 12 [41]. The theoretical
stiffness (Kt) of each test specimen can be determined by applying the
relevant load and boundary conditions to Eq. (4). If a concentrated load
is applied at midspan of a simply supported test specimen, the
J. Ryu et al. Engineering Structures 175 (2018) 467–482

Table 4
Test results.
Group Specimens Load Displacement Initial elastic stiffness

a
Pye (kN) b
Pue (kN) c
Put (kN) Pue/Put δy (mm)
d
δue (mm) μ f
Keg (N/mm) Kth (N/mm) Ke/Kt

1 FP-NB 17.5 21.6 17.7 1.22 82.3 320.4 3.89 329.2 365.3 0.90
FP-NWB 18.9 22.0 17.7 1.24 100.3 263.3 2.63 319.8 365.3 0.88
FP-WB 19.2 22.1 17.7 1.25 98.7 295.2 2.99 342.9 365.3 0.94

2 FP-NB-BS500 18.3 21.7 17.7 1.23 90.8 356.1 3.92 343.9 365.3 0.94
FP-NB-BS250 17.0 20.0 17.7 1.13 70.4 136.1 1.93 357.1 365.3 0.98
FP-NB-CS500 18.4 21.5 17.7 1.21 94.2 240.5 2.55 342.9 365.3 0.94
FP-NB-CS250 18.7 22.4 17.7 1.27 95.5 314.0 3.29 351.5 365.3 0.96

3 FP-NB-C 11.5 15.7 15.1 1.04 69.8 174.6 2.50 234.4 313.0 0.75
FP-NB-CB 14.9 17.6 15.3 1.15 91.6 268.4 2.93 267.6 314.7 0.85
FP-WB-CB 16.0 18.6 15.3 1.22 105.2 325.0 3.09 278.9 314.7 0.89

4 FP-NB-CWC 53.3 66.2 53.1 1.25 89.7 319.9 3.57 941.9 1095.8 0.86
FP-NB-IWC 48.0 55.0 53.1 1.04 78.9 138.0 1.75 823.2 1095.8 0.75

a
Experimental yield strength.
b
Experimental ultimate strength.
c
Theoretical ultimate strength.
d
Displacement at the yield strength.
e
Displacement at the ultimate strength.
f
Ductility (μ = δu/δy).
g
Experimental initial elastic stiffness.
h
Theoretical initial elastic stiffness.

to estimate the load-transferring capacity of the welded connections.

The ductility, μ, for each test specimen is also summarized in
Table 4. Most of the test specimens exhibited adequate ductilities with a
range of 2.63–3.92 except for the test specimens showing the abrupt
load reduction due to premature debonding failure or lateral-torsional
behavior. Considering that the proposed panel maintains a high load-
carrying capacity after the maximum load (at about 90% of the max-
imum load), the ductility of this composite panel can be actually
evaluated higher than the values presented in Table 4.

6. Finite element analysis (FEA)

An FEA was performed using ABAQUS 6.14 [42] to suggest a proper

Fig. 12. The definitions of initial elastic stiffness (Ke), yield load (Pye), and yield
displacement (δy).
load–deflection relationship can be defined with Eq. (7), and the the-
oretical stiffness can be calculated with Eq. (8).
PL3
yiF =
48Es Itr
48Es Itr
Kt = P / yiF =
L3
where P = the load applied and L = the span length.
The theoretical stiffness of a single composite panel (Groups 1 and
2) was calculated to be 365.3 N/mm from Eq. (8), and that for Groups 3
and 4 were 314.7 N/mm and 1095.8 N/mm, respectively. In the
load–deflection responses, the Group 1 and 2 test specimens exhibited
the flexural stiffnesses equal to 93.4% of the theoretical stiffness on
average. This means that when calculating the stiffness of the proposed
panel, use of the moment of inertia for the transformed section (Itr)
tends to overestimate the actual stiffness by 7% on average.
In the case of the Group 3 and 4 test specimens, they exhibited a
11%–25% lower stiffness than the theoretical value. This is because, for
the Group 3 test specimens, the bond between the cut surface and the
attached steel plate did not function like the welds, and because, for the
Group 4 test specimens, the load was only applied to the central panel
numerical model for the iFLASH panel, and to examine the state of the
stresses generated between the steel plate and the polymeric core. In
this study, the main target specimen is FP-NB with no steel surface
treatment and no shear connectors. From the experimental results for
the Group 1 and 2 test specimens, it was found that the flexural per-
formance of FP-NB was equivalent to those of the other specimens, even
without any special surface treatment or any additional shear con-
nectors. In addition, for the Group 3 and 4 test specimens, the optimal
application methods related to cutting and joining of the panels were
(7) fully verified by tests. For these reasons, the FEA was conducted for
only FP-NB, not all specimens. Instead, to confirm the suitability of the
proposed equations for predicting the flexural strength and stiffness,
(8)
additional iFLASH panels having different dimensions and span lengths
were modeled.
6.1. Finite element models
A single iFLASH panel was modeled with the dimensions measured
on the FP-NB test specimen. As shown in Fig. 13, only one quarter of the
entire test specimen was considered to reduce the computational cost of
the analysis, and symmetry boundary conditions were applied to the cut
surface of the model. Analogous to the experiments, the loading beam
and the support were also modeled by using an analytical rigid surface
which is never allowed to be deformed.
The C3D8, which is an 8-node linear brick element with full in-
tegration, was used to generate the meshes, and the element size was
varied from 3 mm to 10 mm. The finest mesh (3 mm) was used at the

478
J. Ryu et al.
Fig. 13. Configuration and boundary conditions of the finite element model.
locations where stress concentrations or significant deformations oc-
curred, and a relatively course mesh was used for the rest of the FE
model.
The nonlinear stress–strain curves obtained from the experiments
were applied as the material properties for the FEA. In the case of the
polymeric material, the hyperelastic material model was selected to
simulate its mechanical response, and the Marlow model was used as a
parameter of the hyperelastic strain energy potentials [43]. A geometric
nonlinearity was added to reflect the nonlinear behavior induced by
large deformation.
The contact pair approach was used to model the contact between
the exterior surface of the test specimen and the analytical rigid surface
such as the loading beam or the support. Two mechanical contact
properties were assigned to these contact pairs: hard contact for the
normal behavior and frictional contact for the tangential behavior. The
friction coefficient between the exterior surface of the test specimen
and the analytical rigid surface was assumed to be 0.4.
Additional contact pairs were defined between the steel plate and
the polymeric core. In order to simulate debonding failure between the
two materials, surface-based cohesive behavior and damage model
were selected among the mechanical contact properties. Fig. 14 shows
the traction–separation response for the cohesive surface used in this
analysis. As illustrated in Fig. 14, prior to damage, a linear elastic be-
havior was assumed. The cohesive stiffness (Kn) in both normal and
tangential directions was assumed to be the same as the tensile modulus
of elasticity of the polymeric core. The average tensile bond strength
(6.34 MPa for the steel plate without surface treatment) obtained from
the pull-off adhesion test was used as a peak value of the traction stress.
The damage to the cohesive surface was assumed to initiate when the
result calculated on the left side of Eq. (9) reaches a value of 6.34.
tn2 + ts2 + tt2 = 6.34
where tn = the normal traction stress; ts = the shear traction stress in
the first direction; and tt = the shear traction stress in the second
Fig. 14. Traction–separation response for the cohesive surfaces.
Engineering Structures 175 (2018) 467–482
direction. The ts and tt are two orthogonal shear traction stresses tan-
gent to the contact surface. Once the damage initiation criterion is met,
damage occurs and the cohesive stiffness is degraded. In this analysis, a
linear softening law was chosen as a damage evolution. To solve the
severe convergence difficulties caused by the stiffness degradation, a
viscosity coefficient was specified with a value of 1.0 × 10−5 which is
small enough not to affect the analysis results. Furthermore, to clarify
the post-failure behavior of the cohesive surface, the hard and frictional
contact (with a friction coefficient of 0.3) models were also added to the
mechanical contact properties.
In the same manner as described above, three iFLASH panels with
different thicknesses for the steel plate and polymeric core and span
lengths were modeled. The details of the FE models were presented in
Table 5.
6.2. FE analysis results
The FE analysis results for the FP-NB model were compared with the
experimental results of all Group 1 and 2 test specimens because they
exhibited very similar behaviors. As shown in Fig. 15(a), the analysis
and experimental results were confirmed to be in good agreement up to
the maximum load. The only difference is that the FE model behaved in
a linearly elastic manner with no reduction in stiffness until the yield
point while the actual test specimens showed a gradual decrease in
stiffness near the yield point. This is due to the presence of residual
stresses induced by cold forming and welding and the initial geometric
imperfections of the thin steel plates. However, this difference is not a
key issue that has a significant effect on the reliability of the FE model.
The maximum load and initial elastic stiffness obtained from the FE
analysis were 22.3 kN and 345.3 N/mm, respectively. These values are
(9) not significantly different from the experimental results of the Group 1
and 2 test specimens. Additionally, as shown in Fig. 15(b), the strains
obtained from the FE analysis increased in a similar manner as those
measured from the gauges attached to the top and bottom steel plates.
These results demonstrate that the proposed FE modeling method is
appropriate for predicting the flexural behavior of the iFLASH panel.
Fig. 16 illustrates the degree of damage to the cohesive surfaces
between the steel plates and the polymeric core under different load
stages. In this figure, the black areas indicate a region where the da-
mage initiation criterion is met, and the letter A and B denote the center
and the end in the direction of the width of the test specimen, respec-
tively.
No debonding failure between the two materials was observed until
the yield point (point a). When the applied load reached 19.7 kN (point
b), the initial debonding between the steel solid bar and the polymeric
core occurred locally owing to the tensile bending stresses caused by
bending. At a load of 20.0 kN (point c), the lower part of the interface
between the steel solid bar and the polymeric core was fully debonded,
479
J. Ryu et al. Engineering Structures 175 (2018) 467–482

Table 5
FE analysis results for the models with different dimensions and span lengths.
FE models Thickness Span length Ultimate load Initial elastic stiffness

a
tsp (mm) tc (mm) ttotal (mm) L (m) Pua (kN) Put (kN) Pua/Put Ka (N/mm) Kt (N/mm) Ka/Kt

FP-S2.4-P19.6 (FP-NB) 2.4 19.6 24.4 2.0 22.3 17.7 1.26 345.3 365.3 0.95
FP-S3-P30 3.0 30 36 2.0 46.7 33.8 1.38 952.4 1033.6 0.92
FP-S4-P40 4.0 40 48 2.5 65.6 60.6 1.08 1163.3 1261.1 0.92
FP-S5-P50 5.0 50 60 2.5 107.9 95.6 1.13 2253.1 2476.2 0.91

a
The subscript ‘a’ denotes ‘analytical values’ obtained from the FE analyses.

and thereafter the gap between the two parts was gradually widened as
shown in Fig. 17. This local debonding resulted in an increase in the
longitudinal shear stress at the contact surface between the bottom steel
plate and the polymeric core, which led to debonding of the bottom
steel plate near the steel solid bar. As the load increased, the local de-
bonding between the bottom steel plate and the polymeric core pro-
gressed into the adjacent areas (point d). At the maximum load, 22.3 kN
(point e), the debonding failure zone (the black area) extended to a
point 150 mm from the load point. However, this local debonding of the
bottom steel plate never causes structural instability of the proposed
panel because the thin steel plate is always in tension. In addition, the
steel solid bars were installed to prevent a local failure of the test
specimen which occurs due to the stress concentrations induced by
loading. Considering that no steel solid bar is installed within the actual
iFLASH panel, the extensive debonding of the bottom steel plate from
the polymeric core is not expected until the ultimate limit state, which
is similar to the interaction between the top steel plate and the core in
the FE model. The contact between the top steel plate and the polymeric
core was maintained without debonding until the ultimate load. At the
peak load, a local debonding of the top steel plate took place over a very
small area adjacent to the steel solid bar. However, this debonding
failure zone was not large enough to cause out-of-plane buckling of the
top steel plate. From the experimental and FE analysis results, it can be
concluded that the bond strength between the steel plates and the
polymeric core is sufficient to maintain the stability of the proposed
panel and to resist the forces generated at the interface between the two
materials when the proposed panels are subjected to bending without
in-plane load.
Table 5 and Fig. 18 shows the FE analysis results of the three iFLASH
panels with different dimensions and span lengths. The FE analysis
results were compared to the theoretical values computed from Eqs.
(1)–(3) and (8). As in the calculation of the flexural strengths of the test
specimens, the maximum compressive and tensile bending stresses at
the extreme fiber of the polymeric core were assumed to be 6.0 MPa and
16.6 MPa, respectively.
The FE models showed a flexural strength of 46.7 kN for FP-S3-P30
with a span length of 2.0 m, 65.6 kN for FP-S4-P40 with a span length of
2.5 m, and 107.9 kN for FP-S5-P50 with a span length of 2.5 m. The
ratios of the analytical strength to the theoretically calculated strength
(Pua/Put) varied from 1.08 to 1.38, depending on the polymeric core
thickness-to-steel plate thickness ratios (tc/tsp) and the span-thickness
ratios (L/ttotal). However, Eqs. (1)–(3) always gave conservative design
strengths for the iFLASH panels subjected to bending, regardless of the
variables considered in the FE analyses. In the load–deflection re-
sponses, the FE models exhibited the initial elastic stiffnesses equal to
91.9% of the theoretical stiffnesses on average, which was very similar
to the experimental results. Therefore, it is recommended that for a
more accurate deflection calculation of the proposed panel, the calcu-
lated transformed moment of inertia should be reduced by 10% (that is,
0.9Itr should be used). The bond interactions between the steel plate
and polymeric core in the FE models were similar to that of the FP-NB
model.
7. Conclusions
In this study, the iFLASH System was developed as a new floor
system for steel structure buildings to replace the concrete deck slab
systems. The flexural capacity of the iFLASH floor panel was evaluated
by performing experiments applying various independent variables. An
FEA was also performed to examine the state of stresses generated be-
tween the steel plate and the polymeric core. The following summarizes
the results presented herein.
(1) The flexural behaviors and bending resistances of the test speci-
mens of Groups 1 and 2 were very similar, independent of the
variables. Except for FP-NB-BS250 and FP-NB-CS500 which in-
curred extensive debonding failure of the top steel plate due to an
inadequate curing of the polymeric core during fabrication, all test
specimens of Groups 1 and 2 exhibited very ductile behavior and
maintained its structural integrity even after a maximum load. No

Fig. 15. A comparison of the FE analysis and experimental results: (a) Load–deflection relationships; (b) Load–strain curves.

480
J. Ryu et al.
Fig. 16. The degree of damage to the cohesive surfaces between the steel plates and the polymeric core under different load stages: (a) P = 18.8 kN (point a); (b)
P = 19.7 kN (point b); (c) P = 20.0 kN (point c); (d) P = 21.4 kN (point d); (e) P = 22.3 kN (point e).
Fig. 17. The gap between the steel solid bar and the polymeric core
(P = 21.4 kN).
other failure mode than the face yielding of the top and bottom
steel plates was observed. From these test results, it was verified
that the proposed floor panel can be fabricated and applied in the
field without any special surface treatment or any additional shear
connectors.
(2) FP-NB-C, which was not reinforced with a steel plate at the cut
surface, exhibited early lateral-torsional behavior due to the
asymmetric cross-sectional shape while FP-NB-CB and FP-WB-CB
that were reinforced with a side plate exhibited stable flexural be-
haviors. In addition, the reinforcing side steel plate bonded using
the proposed adhesive did not debond from the panel until the
experiment ended and contributed to ensure the stability of the
panel. Based on these results, for the convenient reinforcement, the
use of the proposed adhesive is recommended as an effective al-
ternative to welding when the cut panel is reinforced with a side
steel plate.
481
Engineering Structures 175 (2018) 467–482
Fig. 18. Load–deflection curves for the FE models with different dimensions
and span lengths.
(3) In the case of the Group 4 test specimens with different welding
methods, FP-NB-IWC, the test specimen with intermittent welds,
showed an intense out-of-plane deformation of the top steel plates
and a rapid decrease in loading without ductile behavior because of
a concentrated load transferred through the intermittently welded
connections. Therefore, in cases where the proposed panels are
connected side by side, continuous welding is recommended as a
more preferable connection method.
(4) The FE analysis results were in good agreement with the experi-
mental results. The maximum load and initial elastic stiffness ob-
tained from the FE analysis were 22.3 kN and 345.3 N/mm,
J. Ryu et al.
respectively. These values were not significantly different from the
experimental results. Therefore, it was demonstrated that the FE
modeling method suggested in this study is appropriate for pre-
dicting the flexural behavior of the proposed panel.
(5) At the maximum load, the bending stresses in the core material
remained within the linearly elastic range while the entire cross
section of the top and bottom steel plates was fully yielded. The
flexural strength of the proposed panel can be determined based on
the distribution of the bending stresses acting on a cross section at
the ultimate limit state, and the maximum compressive and tensile
bending stresses at the extreme fiber of the polymeric core can be
conservatively assumed to be 6.0 MPa and 16.6 MPa, respectively,
based on the test results. The theoretically calculated strengths
were about 80% of the actual performance, which was sufficiently
conservative to be used as a nominal strength of the developed
panel.
(6) The deflection of the proposed panel under service loads can be
determined based on the Euler–Bernoulli beam theory and the
transformed-section method. The test and FE analysis results
showed that the actual stiffnesses were on average 7.8% lower than
the theoretically calculated values. This was because use of the
moment of inertia for the transformed section (Itr) tends to over-
estimate the actual stiffness by 8% when calculating the stiffness of
the proposed panel. Therefore, it is recommended that for a more
accurate deflection calculation of the proposed panel, the calcu-
lated moment of inertia for the transformed section should be re-
duced by 10%.
(7) The tensile bond strength between the steel plate without surface
treatment and the infilled polymeric material, measured from the
pull-off adhesion test, was 6.34 MPa. Based on the experimental and
FE analysis results, it was demonstrated that this bond strength is
sufficient to resist the forces generated at the interface between the
two materials and to maintain the stability of the proposed panel
when the proposed panels are subjected to bending without in-
plane load.
Acknowledgements
This research was financially supported by a grant from the Korea
University and the National Research Foundation of Korea (NRF-
2013R1A2A2A01067872 and NRF-2017R1A2B3006531). The writers
are grateful to the authorities for their support.
References
[1] Davies JM. Sandwich panels. Thin Wall Struct 1993;16:179–98.
[2] Karlsson KF, TomasÅström B. Manufacturing and applications of structural sand-
wich components. Compos Part A-Appl Sci 1997;28(2):97–111.
[3] Meraghni F, Desrumaux F, Benzeggagh ML. Mechanical behaviour of cellular core
for structural sandwich panels. Compos Part A-Appl Sci 1999;30(6):767–79.
[4] He M, Hu W. A study on composite honeycomb sandwich panel structure. Mater
Design 2008;29(3):709–13.
[5] Abbadi A, Tixier C, Gilgert J, Azari Z. Experimental study on the fatigue behaviour
of honeycomb sandwich panels with artificial defects. Compos Struct
2015;120:394–405.
[6] Queheillalt DT, Murty Y, Wadley HNG. Mechanical properties of an extruded pyr-
amidal lattice truss sandwich structure. Scr Mater 2008;58(1):76–9.
[7] Deshpande VS, Fleck NA. Collapse of truss core sandwich beams in 3-point bending.
Int J Solids Struct 2001;38(36–37):6275–305.
[8] Hao Q, Qiu S, Hu Y. Development on preparation technology of aluminum foam
sandwich panels. Rare Metal Mat Eng 2015;44(3):548–52.
[9] Jiang B, Li Z, Lu F. Failure mechanism of sandwich beams subjected to three-point
bending. Compos Struct 2015;133:739–45.
482
Engineering Structures 175 (2018) 467–482
[10] Yan LL, Han B, Yu B, Chen CQ, Zhang QC, Lu TJ. Three-point bending of sandwich
beams with aluminum foam-filled corrugated cores. Mater Design 2014;60:510–9.
[11] Oduyemi TOS, Wright HD. An experimental investigation into the behaviour of
double-skin sandwich beams. J Constr Steel Res 1989;14(3):197–220.
[12] Roberts TM, Edwards DN, Narayanan R. Testing and analysis of steel-concrete-steel
sandwich beams. J Constr Steel Res 1996;38(3):257–79.
[13] Richard Liew JY, Sohel KMA. Lightweight steel-concrete-steel sandwich system
with J-hook connectors. Eng Struct 2009;31(5):1166–78.
[14] Bowerman HG, Gough MS, King CM. Bi-steel design and construction guide.
Scunthorpe: British Steel Ltd; 1999.
[15] Leekitwattana M. Analysis of an alternative topology for steel-concrete-steel sand-
wich beams incorporating inclined shear connectors [Ph.D. thesis]. UK: University
of Southampton; 2011.
[16] Reis EM, Rizkalla SH. Material characteristics of 3-D FRP sandwich panels. Constr
Build Mater 2008;22:1009–18.
[17] Fam A, Sharaf T. Flexural performance of sandwich panels comprising polyurethane
core and GFRP skins and ribs of various configurations. Compos Struct
2010;92:2927–35.
[18] Awad ZK, Aravinthan T, Zhuge Y. Experimental and numerical analysis of an in-
novative GFRP sandwich floor panel under point load. Eng Struct 2012;41:126–35.
[19] Tuwair H, Hopkins M, Volz J, ElGawady MA, Mohamed M, Chandrashekhara K,
et al. Evaluation of sandwich panels with various polyurethane foam-cores and ribs.
Compos Part B-Eng 2015;79:262–76.
[20] Van Erp G, Rogers D. A highly sustainable fibre composite building panel. In:
Proceedings of the international workshop on fibre composites in civil infra-
structure – past, present and future. University of Southern Queensland,
Toowoomba, Queensland, Australia, 1–2 December 2008.
[21] Mullens MA, Arif M. Structural insulated panels: impact on the residential con-
struction process. J Constr Eng M 2006;132(7):786–94.
[22] American Institute of Steel Construction (AISC). Steel construction manual.
Chicago, IL: AISC; 2011.
[23] Ju YK, Chun SC, Kim SD. Flexural test of a composite beam using asymmetric steel
section with web openings. ASCE J Struct Eng 2009;135(4):448–58.
[24] Ryu J, Ju YK, Yoon SW, Kim SD. Bending capacities of glass fibre reinforced plastic
composite slab. Mater Res Innov 2013;17(Suppl 2):s12–8.
[25] Plantema FJ. Sandwich construction: the bending and buckling of sandwich beams,
plates, and shells. New York, NY: John Wiley & Sons Ltd; 1966.
[26] Ryu J, Ju YK, Yoon SW, Kim SD. Bending capacity of glass fibre steel composite
plate (GSP) reinforced composite slab. Mater Res Innov 2013;17(Suppl 2):s27–34.
[27] Smith AL, Hicks SJ, Devine PJ. Design of floors for vibration: a new approach, SCI
P354. UK: The Steel Construction Institute; 2007.
[28] Allen HG. Anlaysis and design of structural sandwich panels. London: Pergamon
Press; 1969.
[29] Gere JM. Mechanics of materials. sixth edition Belmont, CA: Thomson Learning,
Inc.; 2004.
[30] Manalo AC, Aravinthan T, Karunasena W, Islam MM. Flexural behaviour of struc-
tural fibre composite sandwich beams in flatwise and edgewise positions. Compos
Struct 2010;92:984–95.
[31] American Society for Testing and Materials (ASTM). ASTM D638-14: Standard test
method for tensile properties of plastics. West Conshohocken, PA: ASTM
International; 2014.
[32] American Society for Testing and Materials (ASTM). ASTM D695-15: Standard test
method for compressive properties of rigid plastics. West Conshohocken, PA: ASTM
International; 2015.
[33] Korean Standards Association (KSA). KS B 0801-2007: Test pieces for tensile test for
metallic materials. Seoul, South Korea: KSA; 2007.
[34] Korean Standards Association (KSA). KS B 0802-2003: Method of tensile test for
metallic materials. Seoul, South Korea: KSA; 2003.
[35] DeFelsko Corp. Manual for the DeFelsko AT-A PosiTest Adhesion Tester.
Ogdensburg, NY: DeFelsko Corp; 2013.
[36] The Society for Protective Coatings (SSPC). SSPC-SP5/NACE No.1 White metal blast
cleaning. Pittsburgh, PA: SSPC; 2007.
[37] The Society for Protective Coatings (SSPC). SSPC-SP10/NACE No.2 Near white blast
cleaning. Pittsburgh, PA: SSPC; 2007.
[38] Architectural Institute of Korea (AIK). Korean Building Code and Commentary
2016. Seoul, South Korea: Kimoondang; 2016.
[39] Ju YK, Kim DH, Kim SD. Experimental assessment of the shear strength of an
asymmetric steel composite beam with web openings. Can J Civil Eng
2005;32(2):314–28.
[40] Ryu J, Lee CH, Oh J, Yoon SW, Ju YK. Shear resistance of a biaxial hollow com-
posite floor system with GFRP plates. ASCE J Struct Eng 2017;143(2):04016180.
[41] Whittaker AS, Bertero VV, Thompson CL, Alonso LJ. Seismic testing of steel plate
energy dissipation devices. Earthq Spectra 1991;7(4):563–604.
[42] ABAQUS [Computer software]. Providence, RI: Dassault Systèmes Simulia.
[43] Dassault Systèmes Simulia. Abaqus 6.14, Analysis User’s Guide Volume III:
Materials. Providence, RI: Dassault Systèmes Simulia; 2014.