You are on page 1of 14

Parametric Study on Mechanical Responses of

Corrugated-Core Sandwich Panels for Bridge Decks


Mehdi Tehrani1; Farshad Hedayati Dezfuli, S.M.ASCE2; M. Shahria Alam, M.ASCE3; and Abbas S. Milani4

Abstract: A critical challenge in bridge design and the construction process is to reduce the weight of the bridge deck. Specifically, in small
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

aged bridges, light modules provide an easy and fast bridge deck renewal. Sandwich panels were introduced as such lightweight bridge decks
a few decades ago. Low density and high specific strength of the panels provide remarkable advantages for a wide variety of industrial applica-
tions. The objective of this study was to investigate the effect of geometric parameters on the mechanical behavior (deflection and shear force)
of a corrugated-core steel sandwich panel and predict its response by developing mathematical regression models. The results reveal that the
core and the face sheet thicknesses highly affect the panel deflection response, whereas the weld spacing has the highest contribution to the
maximum shear force response. DOI: 10.1061/(ASCE)BE.1943-5592.0001026. © 2017 American Society of Civil Engineers.
Author keywords: Corrugated sandwich panel; Bridge deck; Spot weld shear capacity; Panel deflection.

Introduction (Burlayenko and Sadowski 2010; Yazici et al. 2014; Yang et al.
2015), polymer (Liu et al. 2016), a truss (Lok and Cheng 2000;
Demand for lighter and modular structures is increasing in recent Yuan et al. 2016), or precast concrete (Mugahed Amran et al.
years due to driving factors in construction projects, such as tight 2016). The outer layers usually correspond to flat metal sheets
scheduling, labor, management, and overall cost. For instance, attached to the core by means of an adhesive agent or welding tech-
reducing the required man hours is highly favorable for construction nology. The core guarantees that both face sheets are tightly
companies. The use of prefabricated modular structures leads to a involved together to resist against the applied load on top of the
lower number of workers on site and instead, a longer fabrication sandwich panel. It also contributes to stiffening the face sheets to
time in the shop, which is translated to lower costs. Moreover, in the postpone the local buckling. On the other hand, the face sheets can
renewal of aged and deteriorated bridges, the installation of modu- significantly protect the core material from any physical damage or
lar superstructure components definitely helps minimize the disrup- degradation.
tion to public transportation. The first investigation into sandwich panels was conducted by
Along with the modular concept, which contributes to an easy Libove and Batdorf (1948). They suggested a simplified deflec-
and fast construction, the weight of the bridge superstructure (e.g., tion theory by presenting an equivalent elastic modulus for the
deck) also plays an important role in the design and construction sandwich panel. Three years later, Libove and Hubka 1951 pre-
levels. To achieve the minimum possible weight for deck structures, sented a method for obtaining an equivalent elastic modulus for a
corrugated-core steel sandwich panels with a significant high stiff- corrugated-core sandwich panel. Plantema (1966) and Allen (1969)
ness to weight ratio are introduced. Low density and high specific described the mechanical behavior of the sandwich panels. They
strength of such panels provide remarkable advantages for a wide explained that the sandwich concept plays an important role in the
variety of industrial applications. In structural applications, sand- efficient engineering system development due to their significant
wich panels have two main components: an inner core and top and advantages. The most remarkable features include high stiffness to
bottom outer layers. The core usually corresponds to a section weight ratio, high thermal insulation property, and high impact and
between outer layers filled with a lightweight material or hollow vibration absorption rates (Zenkert 1995). Kennedy et al. (2002)
sections formed with corrugated sheets. Depending on the applica- discussed the need for a lightweight and cost-efficient deck for
tion of the panel, the core can be filled with compressed foam portable bridges. They pointed out that bridge decks made of tradi-
tional steel plates are expensive due to the amount of required
1 welding. They compared a traditional steel box girder with a stiff-
Master’s Graduate, School of Engineering, Univ. of British Columbia,
Kelowna, BC, Canada V1V 1V7. E-mail: mehdi.tehrani@austinengineering ened sandwich plate box girder and a composite core sandwich
.ca plate box girder, which showed that sandwich plate girders are
2
Postdoctoral Fellow, School of Engineering, Univ. of British more cost-efficient and have superior performances at ultimate, fa-
Columbia, Kelowna, BC, Canada V1V 1V7. E-mail: farshad.hedayati@ tigue, and serviceability limit states. They concluded that such sys-
alumni.ubc.ca tems are attractive alternatives for traditional bridge decks due to
3
Associate Professor, School of Engineering, Univ. of British reduced welding parts and ease of construction. Depending on the
Columbia, Kelowna, BC, Canada V1V 1V7 (corresponding author). application of sandwich panels, the structural behavior of panels
E-mail: shahria.alam@ubc.ca needs to be analytically characterized. Such investigations have
4
Associate Professor, School of Engineering, Univ. of British Columbia,
been performed for panels with truss and honeycomb core configu-
Kelowna, BC, Canada V1V 1V7. E-mail: abbas.milani@ubc.ca
Note. This manuscript was submitted on February 22, 2016; approved
rations (Rathbun et al. 2004; Zok et al. 2005; Chi et al. 2010; Sun
on November 15, 2016; published online on January 16, 2017. Discussion and Gao 2013; Abbadi et al. 2015; Zhang et al. 2016). Steel sand-
period open until June 16, 2017; separate discussions must be submitted wich panels have been also investigated due to their multifunc-
for individual papers. This paper is part of the Journal of Bridge tional advantages, such as lightweight structure and high resist-
Engineering, © ASCE, ISSN 1084-0702. ance, in applied blast loading (Qiu et al. 2003; Xue and Hutchinson

© ASCE 04017002-1 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


Fig. 1. Corrugated steel sandwich panel configurations (adapted from Kujala and Klanac 2002)

continuous and discontinuous. A continuous core is fabricated by


folding one steel sheet repeatedly, and a discontinuous one is pro-
duced from several steel-sheet cuts. Based on the application of the
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

sandwich panels, core geometries for the panels can be designed in


different forms and shapes. The most common core configurations
applied to steel sandwich panels are shown in Fig. 1.
Fig. 2. Geometry of the corrugated-core sandwich panel The selected geometry for the sandwich panel presented in this
study is obtained from the experimental work accomplished by Tan
et al. (1989). Fig. 2 illustrates the sandwich panel, which is made of
2003; Poirier et al. 2013). In these investigations panels with differ-
a continuous corrugated core with a modified V configuration. The
ent core geometries, such as truss, square honeycomb, and corru-
assumed panel width, length, and height are 2,120, 5,996, and
gated cores, have been studied.
107.5 mm, respectively. To increase the rigidity and strength of the
One of the most significant criteria that should be taken into
entire panel, two thick surrounding boundary plates are also added
account in steel bridge design process is the dead load due to the
to two longitudinal sides of the panel.
weight of the superstructure (bridge deck). By considering the den-
The core’s corrugation is a repeated form of a unit cell along the
sity of the reinforced concrete in the range of 2,500–2,600 kg/m3,
panel cross section (Fig. 2). In the presented simulation, the core
the applied dead load due to the weight of a concrete slab with
includes four similar unit cells. At each unit cell, the core sheet is
30-cm thickness will be approximately 750 kg/m2. On the other
attached to face sheets by four paths of spot welding through the
hand, the dead load of corrugated-core steel sandwich panels with
depth; two paths of spot welds attach the core to the top face sheet
an adequate live-load capacity will be approximately 250 kg/m2.
and two other paths attach it to the bottom face sheet. Each path con-
This shows that, by using a sandwich panel, the dead load can be
sists of an equal number of spot welds through the depth of the
significantly reduced (up to 65%). Also, a higher amount of dead
panel. The distance between two consecutive spot welds in a path is
load remarkably decreases the live-load capacity of the deck. In
called weld spacing, which is considered as a design parameter in
addition, with the same live-load capacity, a lighter deck leads to a
this study. Fig. 3 shows the assumed dimensions of the unit cell.
smaller size of girders and piers by which the cost of construction
The thickness of the sandwich panel (the height of the unit cell),
may be significantly reduced. To attain this objective, a targeted
which is also considered as a design parameter, is assumed to be
bridge superstructure should be designed based on a minimum pos-
107.5 mm. Because the spot weld elements are applied to the inter-
sible weight, while fulfilling other critical design requirements, face of the core and face sheets in the meshing phase, a 1.25-mm
such as acceptable deflection and force levels in the structure. gap is considered between the two sheets. Weld paths are located at
In this study, a concrete deck is suggested to be replaced with distances of 115 and 415 mm, respectively (Fig. 3).
corrugated-core steel sandwich panels for small bridge deck appli-
cations. Because the fabrication cost depends on the number of spot
welds, decreasing the number of required welding points will be FE Modeling
cost-effective. To present an efficient design for the panel, the me-
chanical behavior of sandwich panels with different geometric pa- To model and analyze the corrugated sandwich panel, ANSYS
rameters, including thickness of core and face sheets, panel height, Mechanical APDL 14.0 software is used in this study.
spot weld distance, and spot weld radius, will be studied. In the con-
sidered bridge deck, the welded spots play an important role in the Geometry and Boundary Conditions
overall integrity and response of the panel. On the other hand,
deflection of the panel can significantly affect the weld’s perform- To reduce the complexity of geometry in the finite-element model-
ance. Therefore, deflection and weld strength are considered as two ing (FEM) simulation, the size of the model is reduced by applying
important design criteria. Here, two responses are monitored: the symmetry planes. As a result, a geometrical model with a lower
global maximum deflection and the shear force distribution at the number of elements, which leads to a lower numerical computation
interface of corrugated core and face sheets. Regarding this, finite- and a shorter processing time, is generated. Here, two planes of
element (FE) models are generated and numerical simulations will symmetry are applied to the sandwich panel on the longitudinal and
be performed, validated, and then extended to different cases transverse centerlines and divide the model into four equal sections
through a parametric study. To predict responses of the panel, math- (Fig. 4). The shaded section shows the quarter model, which will be
ematical regression models are also developed. analyzed in this study.
Both face and core sheets are made of stainless steel, which is
characterized as a bilinear material with mechanical properties
Corrugated Sandwich Panels listed in Table 1. The behavior of steel is simulated using the bilin-
ear kinematic hardening material model with a yield strength of
The flexural properties of the corrugated sandwich panel highly 310 MPa and a postyield hardening ratio of 0.6% defined as a ratio
depend on the geometry of the panel cross section and especially of the postelastic stiffness to the initial stiffness. This material
the core configuration. Corrugations are classified into two groups, model is capable of capturing the Bauschinger effect, which takes

© ASCE 04017002-2 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Corrugated-core unit cell with the weld model

Table 1. Mechanical Properties of Steel Material

Mechanical property Value


Modulus of elasticity (GPa) 209
Poisson’s ratio 0.3
Density (kg/m3) 7,800
Yield stress (MPa) 310
Postelastic stiffness (GPa) 1.23

panel maximum deflection graph, the deflection increases from


8.01 to 8.11 mm for the FEM models with 48,000 and 65,000 ele-
ments, respectively, which denotes a 1.2% increase in the deflec-
tion. For the maximum shear force, the same increase in the number
of elements changes the shear force by 0.18%. Therefore, the FEM
model with approximately 65,000 elements is chosen.
Fig. 4. Applied symmetry lines in longitudinal and transverse Based on the test setup used by Tan et al. (1989), the sandwich
directions panel is assumed simply supported on all sides. As shown in Fig. 4,
the nodes located on the left and front sides of the quarter model
have zero displacement in the y-direction. The displacement of all
into account the effect of tensile strength on the compressive nodes located on the symmetry planes is restricted to in-plane
strength. The simulated welds and the panel have the same material movements, and there is no translation perpendicular to the plane of
properties, because there is no filler material in the spot welding symmetry. It also should be noted that due to thick boundary plates
process. that surround the panel, the side plates are assumed to be rigid in
The core and face sheets are meshed with the SHELL181 ele- comparison with the core and face sheets. As a result, all nodes
ment with the plane stress condition. Bending and membrane placed on the side plates are assumed to have negligible displace-
options are considered for the element stiffness, and the thickness ment in the y-direction.
of element is assumed to be uniform along its length. This rectangu- Face and core sheets are prevented from penetrating each other
lar element with four nodes and 6 degrees of freedom at each node by defining contact areas. Each contact pair, which is considered to
can be used to analyze thin to moderately thick shell structures be surface-to-surface, consists of target and contact (i.e., the one that
(ANSYS 2012). Although SHELL181 is a two-dimensional ele- moves toward the target surface) elements. Element CONTA174 is
ment, the sheet thickness can be added to the element properties located on the surface of shell elements to represent contact and slid-
through the real constant feature in ANSYS. The BEAM188 ele- ing between the target surface and a deformable surface. Contact
ment is selected to mesh spot welds at the interface of core and occurs when the contact element surface penetrates into one of the
face sheets. This element, which is a three-dimensional (3D) two- target segment elements on a specified target surface (ANSYS
node beam element with 7 degrees of freedom at each node (trans- 2012). The target surface is discretized by a set of segment elements,
lations, rotations, and warping), has the capability of capturing TARGE170, and is paired with the associated contact surface.
shear-deformation effects (Timoshenko beam theory).
A convergence study is performed to obtain the required number
Spot Weld
of shell elements. Regarding this, mechanical responses of the sand-
wich panel (e.g., the maximum global deflection of the panel and In this study, the resistance welding, also known as spot welding,
the maximum shear force at spot welds) are considered to assess the was considered for attaching the core sheet to the face sheets. The
numerical convergence of the model and justify the required num- welding process is a thermoelectric process that generates heat at
ber of elements. Fig. 5 presents the convergence of outputs in terms the interface of two metal sheets by conducting electrical current
of the number of elements. The rate of change of the obtained through the electrode heads to the metal parts. Sheets with a mini-
results decreases as the number of elements increases. Based on the mum thickness of 6.35 mm (0.25 in.) can be spot welded.

© ASCE 04017002-3 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


8.5

Panel maximum deflection


3950

Maximum shear force at spot


8
3900
7.5

(mm)

welds (N)
7 3850
6.5 3800
6
3750
5.5
5 3700
0 20000 40000 60000 80000 0 20000 40000 60000 80000
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

Number of elements Number of elements

(a) (b)

Fig. 5. Element convergence for (a) maximum deflection of the panel and (b) maximum shear force at spot welds

In ANSYS, the spot weld feature is applied to the model by intro-


ducing a beam element in the considered gap between the two
sheets, as shown in Fig. 3. This feature adds a contact interface
between the two sheets at weld locations to prevent the sheets from
penetrating each other in the case of large deformation.
Furthermore, this feature is capable of considering the electrode’s
diameter by applying constraint equations to the nodes on both
sheets at the vicinity of the beam elements. Based on the selected Fig. 6. Spot welding paths
electrode diameter, the ANSYS spot welding feature constrains (i.e.,
coupled) all of the nodes, which are located inside the weld circle
traffic pattern is assumed as a single lane at each direction with a
on both sheets, to have the same displacement. Therefore, the elec-
total width of 6 m. Considering this traffic pattern and the panels’
trode head diameter feature guarantees the perfect simulation of
assembly arrangement, the deck may experience different loading
fused zone in the spot welding process. According to the spot weld
scenarios. In the first case, it is assumed that two 5,000-kg two-axle
wizard in ANSYS, the beam element is assigned as a rigid connec-
cars pass at the same time and the wheel base (i.e., the distance
tion, which makes the entire simulated panel stiffer than the real
between the centers of the front and rear wheels) is less than
case. Moreover, the weld may experience a deflection or even fail-
2,120 mm. As a result, the total weight of the vehicle transfers to a
ure as a result of detachment. Therefore, the rigid connection is not
an appropriate assumption. To more accurately model the spot single panel. It should be mentioned that, in this case, the position
welds, instead of using the spot weld wizard, welds are manually of the tire on the panel will be close to the panel longitudinal sup-
modeled by creating beam elements between face and core sheets. ports. In terms of experiencing maximum deflection, the described
Then, by assigning mechanical properties of the steel material loading scenario is not of interest because the load of the tire should
(Table 1) to the BEAM188 elements, the weld behavior is simulated be applied to the longitudinal centerline of the panel. In the second
with the same stiffness as the face and core sheets. In the simulation, loading scenario, two vehicles are considered to follow each other
the fused zone diameter is considered to be 10 mm (Spot Welding in each lane. In this case, the rear wheels of the front car and the
Consultants, Inc. 2014). It should be added that all of the nodes, front wheels of the rear car (in each lane) can be located on a single
which are located within a distance equal to the radius of a spot panel. Fig. 7 illustrates the described traffic pattern on three adjacent
weld, on either top or bottom sheet, are coupled to be displaced to- sandwich panels. As shown, four axles from four vehicles are
gether (Fig. 3). located on the middle panel.
BEAM188 elements are placed at the interface of two SHELL Although the load percentage of front and rear axles of the vehi-
elements along straight paths through the depth of the panel with an cle, respectively, are around 40 and 60% of the total weight
assigned weld spacing. This distance (weld spacing) is one of the (5,000 kg) (CSA 2006), to symmetrically apply the vehicle’s load,
parameters that plays a significant role in the panel stiffness. If the and consider the worst-case scenario in terms of the maximum
distance decreases, the number of welds will increase and, as a applied load, the axle load for all vehicles is assumed to be 50% of
result, the panel will be stiffer. However, because the design of the the total weight (3,000 kg). It should be noted that there are three
panel should be optimized in terms of the number of required welds, moments (Mx, Mz, and Mxz according to the chosen coordinate sys-
the weld spacing should be limited to a maximum value. Fig. 6 tem shown in Fig. 4) acting on each edge of the panel; thus, the
shows the spot welding location in the quarter panel cross section. effect of moments should be considered in the modeling. However,
Four paths are located on the top face sheet, and four paths are by applying symmetric boundary conditions to the panel on the
located on the bottom face sheet. symmetry planes, the effect of these moments are taken into account
by the software while solving the problem. This consideration can
be verified when the FE and experimental results are compared in
Loading
the validation section. The load transferred from each axle to the
According to the application of panels in small bridge decks, the panel acts as a uniform distributed load over two equal areas (i.e.,
load limit is considered based on the lightweight vehicles passing the contact surface between wheels and the road; 10  40 cm). The
over the bridge with a restriction of a total weight of 5,000 kg. The dead load (10 kN/m2) due to the weight of road pavement should

© ASCE 04017002-4 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


Fig. 7. Traffic pattern considered for vehicles on the bridge deck
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

Because the center point of the panel is stiffened by attaching the


core to the face plates, the maximum deflection occurs at the center
of the unit cell rather than the center point of the entire panel.
Fig. 10 illustrates a comparison between the ANSYS FEM results
and experimental tests performed by Tan et al. (1989). The graph
depicts a good agreement between the two results. At a loading
value of 14 kN/m2, the maximum difference between the two meth-
ods is around 5%.

Results and Discussions

Five geometric parameters are chosen as input factors including


core and face sheets thickness (Tc and Tf, respectively), distance
Fig. 8. Total load applied to the quarter model of the sandwich panel between two adjacent welding paths in the unit cell in transverse
direction (P), spot weld spacing in the longitudinal direction (S),
and spot weld radius (R). To obtain the effect of these five input fac-
also be considered. Therefore, in addition to a uniformly distributed tors on the mechanical responses, each factor varies at two levels
load of 10 kN/m2 due to the dead load (e.g., road pavement), which (low and high) and, as a result, 25 (32) runs are established through
is applied to the entire panel, the weight of vehicles is uniformly dis- a full factorial analysis. Table 2 illustrates the considered input fac-
tributed over two areas (contact surfaces) of 10  40 cm (Fig. 8). It tors along with their values.
should be noted that in this study, the applied load (i.e., dead load
and vehicles’ load) is considered as a static load, and it effects of Maximum Deflection
snow and dynamic loading are ignored.
To solve the model, a transient method is used. However, it Table 3 depicts the results obtained from ANSYS for the panel maxi-
should be noted that for the simulated model with assumed loading mum deflection response, Dmax. The effect of each parameter is cal-
and boundary conditions, and considered welding and contact fea- culated by subtracting the low average value from the high average
tures, a static analysis could also be used because it is assumed that value (Table 4). A positive effect value means that increasing the
the applied loads do not change by time (i.e., moving loads due to input factor increases the output response, and a negative effect
the weight of the vehicles are assumed to be transferred as static value indicates that increasing the input decreases the output. To
loads). In the solution control options, the Newton-Raphson method determine the contribution of each parameter to the response, the
with a line search is used, and the large-deflection (large strain) normalized effect of each factor is calculated based on the values of
effect is considered. Also, the time step size in each load step is effects. For normalizing the effects, each value is divided by the
specified to be 0.1 s. summation of effect values. As seen, [Tc] and [Tf] with the same
effect value of 41% are the factors with highest contributions.
To qualitatively compare the contribution of factors with the
FE Model Validation panel maximum deflection response, the effect plot is presented in
Fig. 11. The horizontal axis shows low and high levels, and the ver-
To determine the correctness of FE simulations, the numerical tical axis depicts the corresponding deflection response. The slope
model should be validated with experimental tests conducted on of lines indicates the variation of the response when input factors
the same model. Tan et al. (1989) performed experiments on a increase. The deflection response is more sensitive to the input fac-
corrugated-core sandwich panel by considering a uniformly dis- tor (line) with a higher slope.
tributed load of (14 kN/m2) on the top face sheet. The test is per- The main advantage of designing the simulation experiments
formed for two boundary conditions: simply supported in all with 32 runs is to provide a regression model that can mathemati-
edges of the panel, and simply supported only at two ends. Fig. 9 cally predict the mechanical behavior of the sandwich panel for fur-
shows the distribution of the vertical deflection in the y-direction ther investigation. The regression model links the panel desired
for the simulated panel at the last load step of the applied loading. response to the factors and their interactions by introducing a math-
As expected, based on the applied simply supported boundary ematic formula. Based on the order of considered interactions, the
conditions, the maximum deflection occurs at the top right corner of accuracy of regression may change. Including higher order interac-
the model, which indicates the center point of the full-sized panel. tions may lead to more accurate results; however, it can increase the

© ASCE 04017002-5 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Displacement contours in the y-direction for the quarter model (Note: Units are in millimeters)

6 Table 2. Input Factors Used in the Parametric Study with Their Low and
High Levels
Applied mechanical load

5
Level (mm)
4 Parameter Symbol Low [−1] High [þ1]
(kN/m2)

3 Weld spacing S 80 160


Weld radius R 5 20
2 Core sheet thickness Tc 4 6
Exp. Face sheet thickness Tf 4 6
1
FEM Weld paths distance P 65 115
0
0 2 4 6 8 10
The obtained regression model of the panel maximum deflec-
Panel maximum deflection (mm) tion, Dmax, is shown in Eq. (2). The unit of maximum deflection in
this equation is in millimeters
Fig. 10. Comparison of FEM results with experimental results
obtained by Tan et al. (1989) Dmax ¼ 15:67 þ 0:47 S  0:35 R  3:88 T c  3:87 T f  0:83 P
 0:14 SR  0:12 ST c  0:23 ST f þ 0:20 SP
complexity of the model. The regression equation with a polyno-
mial formula is presented in Eq. (1) þ 0:13 RT c þ 0:17 RT f  0:11 RP þ 1:18 tc T f

y ¼ y þ a1 x1 þ a2 x2 þ    þ an xn þ a12 x1 x2 þ a13 x1 x3 þ     0:20 T c P þ 0:33 T f P (2)

þ anm xn xm (1) Neglecting the second-order and higher order interactions may
affect the accuracy of model prediction. However, the plot provided
This equation includes an intercept term that is the average of in Fig. 12 indicates that assuming just the first-order interaction effects
responses over the entire 32 runs, y; coefficients of main factor, an; in Eq. (2) has been accurate enough to calculate the sandwich panel
and coefficients of first-order interaction effects, anm . It should be maximum deflection response. Fig. 12 compares the panel maximum
added that these coefficients can be obtained by dividing the calcu- deflection predicted from the regression equation, obtained from
lated effect values, listed in Table 4, by the number of factor levels, Design-Expert 9.0, with the results obtained from FE simulations
or by means of regression curve-fitting tools available in Microsoft (ANSYS Mechanical APDL 14.0). The distance of each point from the
Excel or other mathematical and statistical software. Because the plotted line demonstrates the difference between the predicted and
effect values are calculated based on the input factors varying in the actual values. The distribution of plotted points reveals that the regres-
range of [−1] to [þ1], the coefficients in the regression equation are sion equation can properly present the panel maximum deflection.
the effects divided by the number of levels, which is 2. Table 5 Table 6 shows the ANOVA results for the panel maximum
depicts the coefficients associated with factors and their main and deflection response. This table presents the P value for all factors
first-order interaction effects. and interactions. In the ANOVA, the confidence level is assumed to

© ASCE 04017002-6 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


Table 3. Maximum Deflection of the Panel for 32 Runs in the Full 0.022
S

Panel maximum deflection (m)


Factorial Design
0.02 R
Run Dmax (m) tc
1 25.6 0.018 tf
2 26.4 P
3 24.5 0.016
4 25.5
0.014
5 15.8
6 16.2 0.012
7 15.6
8 15.9 0.01
9 14.6
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

-2 -1 0 1 2
10 15.7 Input level
11 14.1
12 14.9 Fig. 11. Geometric factors main effect plot for sandwich panel maxi-
13 9.7 mum deflection response
14 9.8
15 9.7
16 9.7
17 22.8 Table 5. Coefficients of Main Factors and First-Order Interactions in the
Regression Equation
18 26.2
19 22.4 Factor Coefficient
20 23.3
21 12.3 Intercept 15.67
22 15.1 [S] 0.47
23 11.7 [R] −0.35
24 13.2 [Tc] −3.88
25 13.9 [Tf] −3.87
26 14.9 [P] −0.83
27 13.6 [SR] −0.14
28 14.1 [STc] −0.12
29 8.4 [STf] −0.23
30 8.7 [SP] 0.20
31 8.3 [RTc] 0.13
32 8.5 [RTf] 0.17
[RP] −0.11
[TcTf] 1.18
Table 4. Effect Values of Five Factors for the Panel Maximum Deflection [TcP] −0.20
Response [TfP] 0.33
Factor Level Average (mm) Effect (mm) Normalized effect (%)
[S] High 16.1 0.9 5
Low 15.2
[R] High 15.3 −0.7 −4
30 Maximum Deflection
Low 16.0
[Tc] High 11.8 −7.8 −41
Low 19.5 25
[Tf] High 11.8 −7.7 −41
Low 19.5
Predicted

[P] High 14.8 −1.7 −9 20


Low 16.5
15 R2 = 0.958
be 95%, so terms with a P value of less than 0.05 are considered to
have significant contributions to the response. These terms are [S], 10
[R], [Tc], [Tf], [P], [STf], [SP], [TcTf], [TcP], and [TfP]. Insignificant
terms may be eliminated from the regression equation [Eq. (2)] with
a minimal loss of model predictability. 5
For any random geometry, the regression equation should cor- 5 10 15 20 25 30
rectly estimate the maximum deflection response of the sandwich
Actual
panel. To validate the regression model, two new random models
are generated in ANSYS and applied to the regression equation. The Fig. 12. Predicted versus actual results for the maximum deflection
selected geometries for the two models are listed in Table 7. It is response (Note: Units are in millimeters)
noted that the actual values chosen for the five factors should be in

© ASCE 04017002-7 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


Table 6. ANOVA Test Results for the Maximum Deflection Response Table 8. Maximum Deflection Response of Two Random Models

Source Sum of squares Degrees of freedom Mean square P value Method Model I (mm) Model II (mm)
−3 −5
Model 1.049  10 15 6.994  10 <0.0001 ANSYS simulation 21.7 11.1
[S] 7.058  10−6 1 7.058  10−6 <0.0001 Regression 19.1 13.4
[R] 3.993  10−6 1 3.993  10−6 0.0004
[Tc] 4.817  10−4 1 4.817  10−4 <0.0001
[Tf] 4.789  10−4 1 4.789  10−4 <0.0001 Table 9. Maximum Shear Force at Spot Welds for 32 Runs in the Full
[P] 2.192  10−5 1 2.192  10−5 <0.0001 Factorial Design
[SR] 6.393  10−7 1 6.393  10−7 0.0921
[STc] 4.854  10−7 1 4.854  10−7 0.1380 Run Fmax (kN)
[STf] 1.631  10−6 1 1.631  10−6 0.0113 1 12.86
1.263  10−6 1.263  10−6
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

[SP] 1 0.0228 2 21.92


[RTc] 5.688  10−7 1 5.688  10−7 0.1104 3 13.90
[RTf] 8.866  10−7 1 8.866  10−7 0.0509 4 26.56
[RP] 3.633  10−7 1 3.633  10−7 0.1955 5 8.50
[TcTf] 4.479  10−5 1 4.479  10−5 <0.0001 6 16.53
[TcP] 1.290  10−6 1 1.290  10−6 0.0216 7 9.26
[TfP] 3.570  10−6 1 3.570  10−6 0.0006 8 19.47
9 10.30
10 19.76
Table 7. Random-Model Geometrical Parameters 11 11.86
12 20.67
Model I Model II 13 7.74
Test value Scaled Test value Scaled 14 15.22
Input factor (mm) value (mm) value 15 7.74
Weld spacing (S) 100 −0.50 130 0.25 16 16.83
Weld radius (R) 15 0.33 10 −0.33 17 13.50
Core sheet thickness (Tc) 4.5 −0.50 5.2 0.20 18 22.77
Face sheet thickness (Tc) 4.5 −0.50 5.6 0.60 19 15.18
Welding path distance (P) 80 −0.40 95 0.20 20 27.70
21 9.37
22 17.93
the range of considered levels ([−1] and [þ1]) to substitute them 23 9.84
into Eq. (2). 24 20.70
Table 8 demonstrates the panel maximum deflection values 25 10.97
obtained from ANSYS and the regression equation for the two 26 18.89
random models. The comparison between the two deflection val- 27 11.82
ues for each model indicates that the regression equation can 28 22.91
predict the panel maximum deflection response with an accepta- 29 8.75
ble amount of error. The differences between the regression val- 30 16.66
ues and ANSYS results are 12 and 20% for Models I and II, 31 9.61
respectively. 32 18.84

Shear Force Table 10. Effect Values of Five Factors for the Maximum Shear
Because the spot welds are the main component for transferring Response
the shear force from the face sheets to the corrugated core, they Factor Level Average (kN) Effect (kN) Normalized effect (%)
are considered as hot-spot locations for the potential failure.
[S] High 20.21 9.51 49.8
Because the shear force capacity of the spot weld is limited by the
Low 10.70
welding schedule, the shear force flow at the interface of face
[R] High 16.43 1.95 10.2
sheets and the core should be less than the spot weld shear force
Low 14.48
capacity. Therefore, once the ANSYS solution is obtained, shear
[Tc] High 13.31 −4.29 −22.4
forces for all spot welds are listed, and the maximum value is cho-
Low 17.60
sen. The maximum shear force values, Fmax, obtained from 32
[Tf] High 14.28 −2.34 −12.2
ANSYS runs are listed in Table 9.
Low 16.62
Table 10 summarizes the effect values of input factors for the
[P] High 15.96 1.02 5.3
maximum shear force response. As the weld spacing increases, the Low 14.94
shear force magnitude at each spot weld increases. This behavior
occurs because when increasing the weld spacing, the number of
spot welds at each welding line and, consequently, the total number Comparing the effect values of [Tc] and [Tf] indicates that [Tc]
of welds decreases. Therefore, the share of each spot weld from the has more contribution to the total shear force reduction. As men-
total shear forces increases. tioned earlier, in the corrugated sandwich panel the total shear force

© ASCE 04017002-8 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


is transferred to the core through face sheets. Therefore, the core Table 11. ANOVA Test Results for the Maximum Shear Force at Spot
thickness plays an important role in the shear capacity of the panel. Welds
Increasing the thickness of the core sheet contributes more to the Source Sum of squares Degrees of freedom Mean square P value
flexural property improvement than an increase in the face sheet’s
thickness. In addition, increasing the thickness of the core sheet Model 9.820  108
15 6.547  107
<0.0001
leads to a lighter panel compared with the thickness of the face sheet [S] 7.235  108 1 7.235  108 <0.0001
with the same value. For example, as Tc increases from 3.5 to [R] 3.045  107 1 3.045  107 <0.0001
4.5 mm, the panel weight changes from 1,020 to 1,140 kg (12% [Tc] 1.470  108 1 1.470  108 <0.0001
increase). However, with the same change for Tf, the panel weight [Tf] 4.382  107 1 4.382  107 <0.0001
varies from 1,104 to 1,304 kg (18% increase). Therefore, to provide [P] 8.318  106 1 8.318  106 <0.0001
a higher flexural property, it is highly efficient to increase the thick- [SR] 8.810  106 1 8.810  106 <0.0001
ness of the core sheet. Contribution of the weld spacing parameter [STc] 2.764  106 1 2.764  106 0.0081
[STf] 3.229  106 1 3.229  106 0.0048
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

to the shear force response, with around 5% normalized effect


value, is not noticeable among four other factors. [SP] 2.065  105 1 2.065  105 0.4205
According to Table 10, [S] has the highest contribution and [Tc] [RTc] 2.022  106 1 2.022  106 0.0199
is the second-most contributing factor. Because the spot welds [RTf] 1.642  106 1 1.642  106 0.0331
transfer the total shear force from the face sheets to the core, the [RP] 5.777  105 1 5.777  105 0.1858
total number of applied spot welds plays an important role in the [TcTf] 9.010  106 1 9.010  106 <0.0001
shear force distribution at the interface of face and core sheets. As [TcP] 6.436  105 1 6.436  105 0.1638
the number of spot welds increases, the share of each weld in load [TfP] 3554.55 1 3554.55 0.9150
transferring decreases and, as a result, the maximum shear force
reduces.
According to the P values presented in Table 11, all input factors Table 12. Maximum Shear Force at Spot Welds Calculated from the
contribute to the shear force response. However, because [SP], Regression Equation and ANSYS
[RP], [TcP], and [TfP] have P values greater than 0.05, their contri-
Run ANSYS (kN) Regression (kN)
butions are not significant.
To develop the regression equation for the maximum shear force 1 12.86 12.86
response, Fmax, the coefficients are calculated and substituted into 2 21.92 22.38
the general regression equation [Eq. (1)]. The maximum shear force 3 13.90 14.45
presented in Eq. (3) is computed in kilonewtons 4 26.56 26.07
5 8.502 8.32
Fmax ¼ 15:45 þ 4:76 S  0:98 R  2:14 T c  1:17 Tf  0:51 P 6 16.53 16.67
7 9.26 8.90
 0:52 SR  0:29 ST c  0:32 ST f þ 0:08 SP
8 19.47 19.35
 0:25 RT c  0:23 RTf þ 0:13 RP þ 0:53Tc Tf 9 10.30 10.53
10 19.76 18.78
þ 0:14 T c P þ 0:01 Tf P (3) 11 11.86 11.21
12 20.67 21.56
Table 12 summarizes the calculated regression values for the 32 13 7.74 8.11
models and compares them with ANSYS results. A good match 14 15.22 15.19
between two methods, with a maximum relative error of 5.5%, 15 7.74 7.79
shows that the regression model is capable of correctly predicting 16 16.83 16.96
the shear force at spot welds. 17 13.50 13.15
18 22.77 22.99
19 15.18 15.27
Effect of Panel Height on the Response of Panel
20 27.70 27.22
In the fabrication of the corrugated-core steel sandwich panels, 21 9.37 9.17
the panel height (i.e., the distance between the face sheets) is con- 22 17.93 17.84
sidered as an important geometric parameter because it is a driv- 23 9.84 10.29
ing factor in adjusting the stiffness to weight ratio. In this section, 24 20.70 21.06
changes of the maximum deflection and shear force distribution 25 10.97 10.85
at spot welds are studied by changing this parameter. When the 26 18.89 19.43
height of the panel increases, the moment of inertia of the panel 27 11.82 12.07
cross section and the flexural strength both increase. Generally, 28 22.91 22.75
plastic yielding, elastic (Euler) buckling, plastic buckling, and 29 8.75 9.00
debonding are typical failure modes of the core in corrugated- 30 16.66 16.40
core sandwich panels, which are subjected to out-of-plane com- 31 9.61 9.22
pressive loads (Kooistra et al. 2007; Zhang et al. 2016). However, 32 18.84 18.71
in the considered steel sandwich panel with the specified height
(H = 107.5 mm), only face sheets experience local buckling, as
shown in Fig. 13. surface in the core while the base length of the unit cell is fixed, or
According to the panel cross section, there are two solutions for by decreasing the base length and keeping the slope constant.
changing the height with minor changes in other parameters. The Because there are two welding paths on the top and bottom of each
height can be increased either by increasing the slope of the inclined unit cell, decreasing the length of the horizontal section of the unit

© ASCE 04017002-9 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Buckling mode shapes of the steel sandwich panel (H = 107.5 mm)

© ASCE 04017002-10 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


cell is limited by the welding distance due to maintaining the num- indicates that the slope of the load capacity curve increases as h
ber of welding paths at each section. The main reason for not chang- increases. The generated FE code is capable of capturing the mate-
ing the number of welding points is to be consistent in the shear rial nonlinearity and simulating the nonlinear response of the sand-
force distribution pattern in spot welds to be able to distinguish the wich panel. To observe such nonlinearity, the load was increased to
height effect on the shear force response. Thus, variation of height around 100 kN, and the response of the panel was studied. Fig. 16
is done by changing the slope of the inclined surface in the core. shows that the panel with H = 107.5 mm, up to a 60-kN load, experi-
Fig. 14 shows three cross sections with three chosen heights: 107.5, ences a maximum deflection of 30 mm. The panel behaves elasti-
138.5, and 170.5 mm. By increasing the height, the slope of the cally, and above this value and the response becomes inelastic.
inclined sheet in the corrugated core increases. Similarly, for heights of 138.5 and 170.5 mm, the sandwich panel
To investigate the effect of height on the buckling of the sand- undergoes inelastic behavior when the load exceeds 65 and 70 kN,
wich panel, Eigen buckling analysis was performed on panels with respectively, with corresponding maximum deflections of 25 and
three considered heights. Results showed that no local buckling 20 mm.
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

occurred in the core sheets by increasing the height from 107.5 to In the previous section, the shear force response was based on a
170.5 mm. From Fig. 15, which depicts the first five buckling modes single value of maximum shear force at spot welds; however, in this
for different heights, it is observed that only face sheets undergo part, the shear force profile along a specific welding path is plotted.
local buckling. Fig. 6 indicates the welding path numbers on the panel cross
Fig. 16 shows the sandwich panel maximum deflection response section.
versus the applied load for the three heights. The panel maximum Fig. 17 illustrates the shear force value at each spot weld along
deflection decreases by increasing the panel height. The graph eight welding paths in the sandwich panel. Each graph shows the
shear force distribution for three panel heights. It is noted that z = 0

100
90
80

Applied load (kN)


70
60
50
40
30 H = 107.5 mm
20 H = 138.5 mm
10 H = 170.5 mm
0
0 10 20 30 40 50 60
Maximum deflection (mm)
Fig. 14. Sandwich panel cross section with three different heights:
(a) 107.5 mm; (b) 138.5 mm; (c) 170.5 mm Fig. 16. Effect of panel height on the maximum deflection response

Fig. 15. First five buckling modes of the sandwich panel with three considered heights

© ASCE 04017002-11 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


on the horizontal axis refers to the quarter panel end side with the complete panel to both ends, the shear force increases. In other
plane of symmetry, which denotes the middle of the complete panel, words, the spot welds placed around the center of the panel experi-
and z = 2.98 m refers to the other end side with simply supported ence the lowest amount of shear force, whereas the ones located
boundary conditions. near the panel support transfer at a higher amount of shear force to
The graph reveals three significant points. First, except for weld- the core. Second, the shear force decreases by increasing the panel
ing Path 1, the shear force value increases from z = 0 to 2.98 m, height. In all graphs, the shear force profiles for h = 170.5 mm and
which means that as the weld location moves from the center of the h = 139 mm are located below the shear profiles for h = 107.5 mm.

2.0 0.6 h=107.5 mm

Spot weld shear force (kN)


Spot weld shear force (kN) 0.5 h=139 mm
1.5 h=170.5 mm
0.4
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

1.0 0.3
0.2
0.5 h=107.5 mm
h=139 mm 0.1
h=170.5 mm
0.0 0.0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
(a) Spot weld location in z axis (m) (b) Spot weld location in z axis (m)

0.7 2.5
h=107.5 mm

Spot weld shear force (kN)


h=107.5 mm
Spot weld sher force (kN)

0.6 h=139 mm h=139 mm


h=170.5 mm 2.0 h=170.5 mm
0.5
0.4 1.5
0.3 1.0
0.2
0.5
0.1
0.0 0.0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
(c) Spot weld location in z axis (m) (d) Spot weld location in z axis (m)

1.6 2.5
Spot weld shear force (kN)

h=107.5 mm h=107.5 mm
Spot weld sher force (kN)

1.4 h=139 mm h=139 mm


1.2 h=170.5 mm 2.0 h=170.5 mm
1.0 1.5
0.8
0.6 1.0
0.4
0.5
0.2
0.0 0.0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
(e) Spot weld location in z axis (m) (f) Spot weld location in z axis (m)

1.8 2.0
Spot weld shear force (kN)

Spot weld shear force [kN]

h=107.5 mm h=107.5 mm
1.6 h=139 mm h=139 mm
1.4 h=170.5 mm 1.5 h=170.5 mm
1.2
1 1.0
0.8
0.6
0.4 0.5
0.2
0 0.0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
(g) Spot weld location in z axis (m) (h) Spot weld location in z axis (m)

Fig. 17. Effect of panel height on the shear force response at different welding paths: (a) Path 1; (b) Path 2; (c) Path 3; (d) Path 4; (e) Path 5; (f) Path
6; (g) Path 7; (h) Path 8

© ASCE 04017002-12 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


Third, to obtain the shear profile for the whole panel in the range which are available for North America’s bridges [e.g., AASHTO
of z = −2.98 to þ2.98 m, the plotted curves should be mirrored (2005)], for the applied loads.
with respect to the vertical axis.

Acknowledgments
Conclusions and Recommendations
The financial contribution of the Natural Sciences and Engineering
The obtained results from full factorial analysis with five geomet-
Research Council (NSERC) of Canada through a Discovery
ric parameters revealed that core and face sheet thicknesses are
Grant was critical in conducting this study and is gratefully
the most important factors because they have significant contribu-
acknowledged.
tions (41% for each) to the panel maximum deflection response.
Among first-order interaction effects, it was shown that [TcTf]
and [TfP] have significant effects on the panel maximum deflec- References
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

tion. For the spot weld maximum shear force response, it was
observed that the spot weld spacing and the core sheet thickness AASHTO. (2005). LRFD bridge design specifications: Interim revisions,
have the highest contributions (49 and 22%, respectively). [SR] Washington, DC.
and [TcTf] are the most effective interaction effects in the maxi- Abbadi, A., Tixier, C., Gilgert, J., and Azari, Z. (2015). “Experimental study
mum shear force response. on the fatigue behaviour of honeycomb sandwich panels with artificial
As the panel height increases, the panel flexural strength defects.” Compos. Struct., 120, 394–405.
Allen, H. G. (1969). Analysis and design of structural sandwich panels,
increases and, as a result, the panel maximum deflection decreases.
Pergamon Press, London.
Eigen buckling analyses showed that no local buckling occurs in ANSYS. (2012). ANSYS software documentation, ANSYS, Canonsburg,
the core sheet for the considered range of the panel height (107.5 to PA.
170.5 mm). It was also found that for the minimum and maximum ANSYS Mechanical APDL 14.0 [Computer software]. ANSYS, Canonsburg,
heights, the steel sandwich panel behaves elastically for vertical PA.
loads up to 60 and 70 kN, respectively, which correspond to maxi- Burlayenko, V. N., and Sadowski, T. (2010). “Effective elastic properties of
mum deflections of around 30 and 20 mm. Furthermore, the shear foam-filled honeycomb cores of sandwich panels.” Compos. Struct.,
force profile graphs in all welding paths except for Path 1 depicted 92(12), 2890–2900.
that shear force values at spot welds decrease when the panel height Chi, Y., Langdon, G. S., and Nurick, G. N. (2010). “The influence of core
height and face plate thickness on the response of honeycomb sandwich
increases.
panels subjected to blast loading.” Mater. Des., 31(4), 1887–1899.
The results of this study can be applied to further investiga- CSA (Canadian Standard Association). (2006). “Canadian highway bridge
tions to produce more efficient and economical steel sandwich design code (CHBDC).” CAN/CSAS6-06, Rexdale, ON, Canada.
panels. The presented outcome of the effect analysis, as a qualita- Design-Expert 9.0 [Computer software]. Stat-Ease, Inc., Minneapolis.
tive approach, can help structural engineers efficiently design cor- Kennedy, D. J. L., Dorton, R. A., and Alexander, S. D. B. (2002). “The sand-
rugated-core sandwich panels for small bridge decks by evaluat- wich plate system for bridge decks.” Proc., 19th Annual Int. Bridge Conf.,
ing the effect of various design parameters on the mechanical Intelligent Engineering Limited, Ottawa.
response of the panel and determining the influential factors that Kooistra, G. W., Deshpande, V., and Wadley, H. N. G. (2007). “Hierarchical
significantly affect the response. For instance, if the designed corrugated core sandwich panel concepts.” J. Appl. Mech., 74(2),
panel does not exceed the applicable deflection criterion, this 259–268.
Kujala, P., and Klanac, A. (2002). “Plate analysis of all steel sandwich pan-
research will determine the parameters that can be adjusted to
els under uniform pressure load.” Proc., 15th Symp. on Theory and
improve the flexural behavior of the panel. Furthermore, mathe- Practice in Shipbuilding, SORTA, Trogir, Croatia.
matical regression models provide an applicable tool for bridge Libove, C., and Batdorf, S. B. (1948). “A general small-deflection theory
engineers to systematically predict responses of a bridge deck for flat sandwich plates.” NACA TN 2289, Langley Aeronautical
made of corrugated-core steel sandwich panels. Laboratory, Langley Field, VA.
To establish the general applicability of the presented results and Libove, C., and Hubka, R. E. (1951). “Elastic constants for corrugated-core
develop comprehensive mathematical models for predicting sandwich plates.” NACA TN 2289, Langley Aeronautical Laboratory,
responses of the panel, the following future recommendations Langley Field, VA.
should be considered: (1) investigating the model that is fabricated Liu, J., Kan, T., Lou, J., Xiang, L., Zhu, X., and Tang, Y. (2016). “Localized
damage response of carbon fiber reinforced polymer composite sand-
using the laser welding technology [in the FE model, weld spacing
wich panel after thermal exposure.” Polym. Test., 50, 33–40.
(i.e. the distance between each two spot welds) should be signifi- Lok, T., and Cheng, Q. (2000). “Elastic stiffness properties and behavior of
cantly reduced to simulate a continuous line of weld made by laser truss core sandwich panel.” J. Struct. Eng., 10.1061/(ASCE)0733
technology]; (2) considering the number of unit cells in the sensitiv- -9445(2000)126:5(552), 552–559.
ity analysis (because a higher number of corrugations increases the Mugahed Amran, Y. H., Rashid, R. S. M., Hejazi, F., Safiee, N. A., and
panel stiffness); (3) studying different core geometries (O-core, Z- Abang Ali, A. A. (2016). “Response of precast foamed concrete sand-
core, etc.); (4) considering multiple girders as support for the deck; wich panels to flexural loading.” J. Build. Eng., 7, 143–158.
(5) analyzing the behavior of the sandwich panel at different load- Plantema, F. (1966). Sandwich construction, Wiley, New York.
ing scenarios and also under the ultimate limit state; (6) taking into Poirier, J. D., Vel, S. S., and Caccese, V. (2013). “Multi-objective optimiza-
account the effect of uncertainties in the mechanical properties and tion of laser-welded steel sandwich panels for static loads using a
genetic algorithm.” Eng. Struct., 49, 508–524.
geometries; (7) considering the effect of bridge span length on the
Qiu, X., Deshpande, V. S., and Fleck, N. A. (2003). “Finite element analysis
panel responses; (8) increasing the range of input factors with the of the dynamic response of clamped sandwich beams subject to shock
purpose of extending the responses to the nonlinear region; (9) loading.” Eur. J. Mech. A. Solids, 22(6), 801–814.
determining the shear capacity of the sandwich panel through a Rathbun, H. J., et al. (2004). “Measurement and simulation of the perform-
shear failure analysis; (10) performing buckling analysis while ance of a lightweight metallic sandwich structure with a tetrahedral truss
increasing the panel height; and (11) using standard guidelines, core.” J. Appl. Mech., 71(3), 368–374.

© ASCE 04017002-13 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002


Spot Welding Consultants, Inc. (2014). “CMW spot welding catalog.” Yazici, M., Wright, J., Bertin, D., and Shukla, A. (2014). “Experimental
hhttp://www.spotweldingconsultants.com/cmw/cmw.htmli (May 25, and numerical study of foam filled corrugated core steel sandwich struc-
2015). tures subjected to blast loading.” Compos. Struct., 110, 98–109.
Sun, Y., and Gao, L. (2013). “Mechanical behavior of all-composite pyram- Yuan, W., Song, H., and Huang, C. (2016). “Failure maps and optimal
idal truss cores sandwich panels.” Mech. Mater., 65, 56–65. design of metallic sandwich panels with truss cores subjected to thermal
Tan, K. H., Montague, P., and Norris, C. (1989). “Steel sandwich panels: loading.” Int. J. Mech. Sci., 115–116, 56–67.
Finite element, closed solution, and experimental comparisons, on a 6 m Zenkert, D. (1995). An introduction to sandwich construction, Engineering
x 2.1 m panel.” Struct. Eng., 67(9), 159–166. Materials Advisory Services, London.
Xue, Z. Y., and Hutchinson, J. W. (2003). “Preliminary assessment of Zhang, J., Qin, Q., Xiang, C., Wang, Z., and Wang, T. J. (2016). “A theoreti-
sandwich plates subject to blast loads.” Int. J. Mech. Sci., 45(4), cal study of low-velocity impact of geometrically asymmetric sandwich
687–705. beams.” Int. J. Impact. Eng., 96, 35–49.
Yang, B., Wang, Z., Zhou, L., Zhang, J., Tong, L., and Liang, W. (2015). Zok, F. W., Rathbun, H., He, M., Ferri, E., Mercer, C., and
“Study on the low-velocity impact response and CAI behavior of McMeeking, R. M. (2005). “Structural performance of metallic
foam-filled sandwich panels with hybrid facesheet.” Compos. Struct., sandwich panels with square honeycomb cores.” Philos. Mag.,
Downloaded from ascelibrary.org by University Of British Columbia on 04/12/19. Copyright ASCE. For personal use only; all rights reserved.

132, 1129–1140. 85(26–27), 3207–3234.

© ASCE 04017002-14 J. Bridge Eng.

J. Bridge Eng., 2017, 22(5): 04017002

You might also like