Engineering Structures 91 (2015) 197–209

Contents lists available at ScienceDirect

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

Analysis of partially composite foam insulated concrete sandwich

structures
Fengtao Bai ⇑, James S. Davidson
Department of Civil Engineering, Auburn University, Auburn, AL 36849, United States

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

Article history: This paper presents the theoretical development that defines the small displacement behavior of foam
Received 8 October 2014 insulated concrete sandwich panels. Composite theories presented by other researchers are first
Revised 9 December 2014 thoroughly reviewed and scrutinized in the context of their use for precast concrete beams. A more
Accepted 25 February 2015
rigorous Discrete Model that incorporates the complex shear deformation behaviors and independent
Available online 25 March 2015
flexural resistance of the concrete wythes is then derived. Experimental data from full-scale precast sand-
wich panel tests are used to validate the developed methodology. Finally, it is demonstrated that this
Keywords:
study provides a rigorous analysis methodology for foam insulated concrete sandwich structures.
Partially composite beam
Sandwich panels
Ó 2015 Elsevier Ltd. All rights reserved.
Interlayer slip
Layered wood system

1. Introduction and background
1.1. Historical development of composite structures
Recently, due to energy efficiency interests, foam Insulated
Concrete Sandwich Panels (ICSP), as illustrated in Fig. 1, have
gained popularity in civil engineering construction. ICSPs provide
thermal efficiency advantages over other traditional building
façade construction due to the layer of insulation sandwiched
between the structural concrete wythes. Unfortunately, this layer
of insulation, along with the shear connectors, results in significant
behavior complexity, and an absence of versatile analysis metho-
dology. This research focuses on developing a suitable ICSP theory
after reviewing typical composite structures and associated theo-
retical developments. The research developments presented herein
are limited to elastic small displacement behavior of sandwich
panels that have symmetric wythes.
Partially composite structures of three layers and more, some-
times called sandwich or laminated structures, have been used
for structural purposes for more than a half century in the aero-
space engineering and nailed timber construction industries.
During this time period, a number of analysis methods have been
derived [2–11]. In the late 1940s, Granholm [1] published his the-
ory and work in the field of nailed timber structures. Granholm’s
theory focused on the equilibrium of axial force within individual
⇑ Corresponding author.
E-mail address: fzb0011@auburn.edu (F. Bai).
layer and overall bending moment of the whole cross-section.
More than a decade later, Holmberg [2] adopted and improved
Granhom’s theory by considering additional transverse action
and applied it to various concrete structures. Meanwhile, Allen
[3] and Hartsock [4] respectively published essentially the same
governing equations with each other by considering the kinematics
relation between interior and exterior wythes, and overall shear or
bending moment equilibrium. Later, based on very similar govern-
ing equations of Hartsock [4], Ha [5,6] and Davies [7] focused on
aspects of matrix formation and finite element algorithm. Also,
there are a number of studies that focus on a particular aspect of
sandwich structures construction, such as design optimization,
structures with thin or thick wythes [8], and development of
various composite elements [9]. Therefore the list of multi-layer
composite structures theories is extensive, and a comprehensive
review and study of over 1300 publications is available [10].
In bridge engineering, two-layer composite structures such as
composite steel concrete T beams have been widely used and stud-
ied theoretically [12–18]. In 1951, Newmark [12] published his
work on composite T beams; his theory was derived from the
strain compatibility of the steel concrete interface. In 1967,
Goodman [11] proved that Newmark’s theory is the same with
Granholm’s [1]. Since then, a number of studies targeting different
aspects of composite T beam mechanics have been carried out.
Girhammar [13] developed a second order analysis approach.
Ranzi [14–16], Salari [17], and Sousa [18] published studies involv-
ing finite element formation. Fabbrocino [19] employed predefined
moment curvature relation and force equilibrium to study the
mechanics of composite T beams. Xu [20,21] considered the

http://dx.doi.org/10.1016/j.engstruct.2015.02.033
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
198 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209
Fig. 1. Typical sandwich panel geometry.
composite cross-section in a plane stress state and derived theo-
retical solutions.
The mechanical behaviors of metallic sandwich panels, nailed
timber systems, composite bridge beams and ICSPs are similar in
some respects. For example, they all consist of a relatively deform-
able middle layer that results in large overall shear deformation.
However, there are also significant differences. For example, metal
faced sandwich structures, due to the slenderness of the face
plates, are more susceptible to face buckling and other local failure
modes. For nailed timber and composite bridge beams, compo-
nents are connected tightly through shear connectors, leaving little
room for shear deformation. Actually, those two-layer composite
structures could have fairly high composite action that in some
circumstances may be considered as fully composite. Whereas
metallic sandwich panels and ICSPs have a thick middle insulation
bearing considerable amount of shear deformation and therefore
are usually partially composite.
1.2. ICSP literature
Increasing building constructions involving ICSPs call for accu-
rate analysis theory. The majority of available literatures focus on
experimental studies [22–42], although there have also been
attempts to define sandwich panel behavior by force equilibrium
[22,23], classical beam theory [24,25], and adapting various exist-
ing composite theories [26,27]. Most of those theories however are
not appropriate for ICSP behavior and not derived rigorously nor
validated by experiments. Consequently, few of them properly pre-
dict the response behavior of concrete sandwich structures to
transverse loading.
ICSP analysis methodologies by others [22,24,25,28] typically
consider sandwich structures to be classical Euler–Bernoulli beams
involving flexural deflection only, and then attribute the additional
unexpected deflection of the sandwich beam to a reduced moment
of inertia through the concept of composite ratio. The present
study uses a closed form solution to demonstrate that the addi-
tional unexpected deflection, along with other unexplained behav-
iors, is the consequence of shear deformation within the insulation
layer. The moment of inertia should still be calculated without
reduction until cracking. In this way, the correct stress and strain
distributions can be determined.
2. Sandwich structure mechanics
Before proceeding with the theoretical development, sandwich
structures’ general behavior will be briefly introduced. One of the
essential differences distinguishing ICSPs from conventional solid
beams is the significant shear deformation. Therefore the deflec-
tion can be considered to consist of two components, shear deflec-
tion and flexural deflection, as illustrated in Fig. 2(b) and (c),
respectively. These two components interact with each other
through shear connectors and the insulation layer, resulting in four
internal forces: two identical local bending moments around two
wythes and one pair of opposite axial forces that form an overall
bending moment around the entire cross section. External moment
is resisted by means of summing these internal bending moments
at any given cross-section.
The flexural deflection can be obtained readily by Euler–
Bernoulli beam theory, but the shear deflection is complicated
and can vary due to different shear connector stiffness, wythe
thickness, theories employed and assumptions. The primary differ-
ence of various theories lays in this shear deflection and will also
be discussed thoroughly in later sections.
3. Typical theoretical developments
Theories that could potentially be applied are mainly from three
areas: thin face metal laminated plate [3–9], composite steel con-
crete T beam [12–21], and nailed timber structures [1,11]. Among
these three areas, some theories are similar in terms of deriving
governing equations and therefore are applicable to each other in
the elastic range with modification. Here two major strong form
theories are chosen prior to proposing a new model.
Assumptions shared by both are listed as:
(1) Structural behavior is limited to elastic and small displace-
ment, and the constitutive relation is assumed to be linear.
(2) The wythes have the same deflections.
(3) Perfect bond is achieved at the interfaces between the
middle foam layer and the wythes, meaning that no relative
sliding occurs at the two interfaces. However, the two
wythes move relative to each other due to the shear defor-
mation of the middle layer, and in the literature, this relative
movement is referred to as ‘‘slip’’. Therefore, even though no
sliding occurs and the use of the word ‘‘slip’’ may not be
mechanically accurate, ‘‘slip’’ will be used herein to be con-
sistent with other prominent literatures.
(4) Continuous and constant middle layer shear stiffness exists
along the span.
3.1. Single cross section sandwich theories
The first category of theories [3–5] considers the two wythes
connected by the middle layer as one whole unit, and only one
governing equation is derived in this manner. Among these theo-
ries, Allen’s [3] and Hartsock’s [4] theories are most frequently
referenced since their assumptions are most realistic. Also, these
two are equivalent as demonstrated by Ha [5]. Therefore, Allen’s
theory is used to illustrate the approach herein.
From the geometry and shear deformation kinematics shown in
Fig. 3, the shear deformation can be represented as follows:
c
y0s ¼ c ð1Þ
2r
where ys = shear deflection; c = 2r  d = middle layer thickness;
d = wythe thickness; 2r = distance between centroids of two
wythes; and c = shear strain of middle layer. The shear stress caused
by flexural deformation in the middle layer can be considered to be
constant since the stiffness of the middle layer foam is much lower
than that of concrete. Also, because of the existence of the shear
stress and strain in the middle layer, the wythes must retain com-
patibility with the middle layer shear strain, resulting in additional
 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209 199

qb
M M
b
d N N
r
r x
d N N
b M M
y
(a)

qb

ys

(b)

qb

yf
(c)
Fig. 2. Illustration of shear and flexural deflections.

where B ¼ b  ð2rÞ2 =c. Recall Q 1 ¼ Ef Iy000 000

f and Q 2 ¼ Ef If ys , and
plug them into Eq. (4),
d e
 
f Q1 If
c 1 ¼ y0s ð5Þ
ys' BG I
c 2r
Defining shear force equilibrium,
b
Q ¼ Q1 þ Q2 ð6Þ
a
where Q = external applied shear force.

a2 Q ¼ a2 Q 1  a2 Q 2 ð7Þ

Fig. 3. Kinematics of shear deformation.
where a2 ¼ BG=ðEf If ð1  If =IÞÞ. Recall Q 2 ¼ Ef If y000
s , and plug it along
with Eq. (5) into Eq. (7),
shear deflection in the wythes. Therefore, shear force equilibrium a2 Q ¼ a2 Q 1 þ Q 001 ð8Þ
can be established as:
Eq. (8) is a second order linear differential equation and can
ðQ 1 þ Q 2 Þ ¼ bð2rÞs þ Ef If ðy000 000
f þ ys Þ ð2Þ now be solved to first obtain Q 1 and then the other unknowns.

where Q 1 ¼ Ef Iy000

f internal shear force resultant of flexural deflec-
tion in wythes and middle layer; Q 2 ¼ Ef If y000
s = internal shear force
3 2
resultant of shear deflection in wythes; I ¼ bd =6 þ 2bdr = moment
of inertia of the whole cross section; b = width of cross section;
s = shear stress in the middle layer; Ef = modulus of elasticity of
3
wythes; If ¼ bd =6 = moment of inertia of two wythes; and
yf = flexural deflection. From Eq. (1), the shear stress can be defined
as:
s ¼ y0s 2rG=c
where G = shear modulus of the middle layer. Note that in Eq. (3),
the shear modulus G is only suitable for relatively small shear stiff-
ness. For beams with a comparably stiff middle layer, an effective
Gm should be used to account for nonlinearity of line cb in Fig. 3
[3]. Substitute Eq. (3) into Eq. (2),
ðQ 1 þ Q 2 Þ ¼ BGy0s þ Ef If ðy000 000
f þ ys Þ
3.2. Interactive wythes theories
This category of theories may also be referred as layered wood
theory, interlayer slip theory or partially composite beam theory.
This type of theory, as mainly applied to nailed timber structures
and composite steel–concrete T beam, considers the interior and
exterior wythes as separate parts connected by shear connecters
for shear transfer. Although fundamental theories were indepen-
ð3Þ dently developed by Granholm [1] and Newmark [12], Goodman
[11] subsequently proved that they are actually the same. In
Granholm’s theory, two coupled governing equations establish
interactive equilibrium between the axial forces of wythes and
the middle layer shear forces, and the external bending moment
and the sum of the internal bending moment. Newmark’s work
addresses the shear strain compatibility of the steel concrete inter-
face. As a result, only one governing equation of axial force in the
ð4Þ wythes is obtained. In the present study, Granholm’s approach is
200 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209

taken as an example. The middle layer shear transfer mechanism in

this study consists of shear connectors and foam insulation.
y'
3.2.1. Axial forces equilibrium
The strain due to axial force in each wythe is equal to the first y'
derivative of its axial displacement, which is half of the total axial
displacement. Then the axial stress is obtained as:
u02
r ¼ Ew ð9Þ
2
2r
where u2 = slip between wythes due to axial displacement; Ew =
moduli of elasticity of wythes. Notice that the middle layer shear
force causes the change in axial force in the two concrete wythes. Fig. 5. Slip due to bending.
Therefore if a differential length of dx is considered, the axial force
equilibrium is achieved as shown in Fig. 4.
sbdx ¼ N0 dx ð10Þ 3.2.2. Bending moment equilibrium
Considering the distance of the pair of opposite resultant axial
sb ¼ ðArÞ0 ð11Þ forces in each concrete wythe as 2r, the moment equilibrium can
be established as:
where s = shear stress in the middle layer; N = resultant axial force
in the wythe; b = wythe width; A = bd = cross section area of wythe; Mex ¼ Min þ M ex þ 2rN ð18Þ
and d = thickness of wythe. After substituting Eq. (9) into Eq. (11),
the shear stress is obtained as: where Mex = external bending moment; Min = internal bending
moment of interior wythe; Mex = internal bending moment of exter-
Ew A 00 ior wythe. Using Eq. (9) and the assumption that the two wythes
s¼ u ð12Þ
2b 2 have the same deflection, Eq. (18) can be rewritten as:
The shear stress in the middle layer could also be determined by  
  u0
shear stiffness of connectors and corresponding relative slip as: Mex ¼ Ew 2Isgl y00 þ 2r AEw 2 ð19Þ
2
Ku ¼ s ð13Þ 3
where Isgl ¼ bd =12 = moment of inertia of each wythe. Substituting
where K = shear stiffness of the middle layer; u ¼ u1 þ u2 = total Eq. (15) into Eq. (19) yields:
slip between wythes; u1 = relative slip due to bending.
Substituting Eq. (12) into Eq. (13):
  ðu0  2ry00 Þ
Mex ¼ Ew 2Isgl y00 þ 2rAEw ð20Þ
2
Ew A 00
Ku ¼ u ð14Þ Eq. (20) can be rewritten as:
2b 2
  
With u1 as defined in Fig. 5, u2 can be expressed as: Mex ¼  2Isgl þ ð2rÞrA Ew y00 þ rAEw u0 ð21Þ

u2 ¼ u  2ry0 ð15Þ By using Itotal ¼ 2Isgl þ 2r 2 A = moment of inertia for the whole
cross-section; a2 ¼ 2r2 A=Itotal . The second governing equation can
where y = deflection; r = distance from neutral axis of wythe to the
now be determined from Eq. (21) as:
overall neutral axis. Substituting Eq. (15) into Eq. (14) yields:
2bK a2 Mex
u  u00 ¼ 2ry000 ð16Þ y00  u0 ¼  ð22Þ
Ew A 2r Ew Itotal
Eqs. (17) and (22) are combined to form the coupled set of govern-
After rearranging terms and using v2 ¼ 2Kb=ðEw AÞ, the first gov-
ing equations:
erning equations is determined as:
(
u00  v2 u ¼ 2ry000 ð17Þ
u00  v2 u ¼ 2ry000
2
y00  a2r u0 ¼  EwMI ex
total

dx 3.3. Theory for discrete shear connectors

As shown at the beginning of Section 3, one of the common

A N bdx N+N'dx
d assumptions is that the shear stiffness is constant and continuous
E I along the span. However, in reality, this is an unrealistic assump-
r tion since discrete, perhaps with non-uniform spacing, shear con-
nector distributions are often used. Therefore, an arbitrary shear
stiffness function is implemented in the present research. By doing
so, any form of shear connector pattern can be analyzed, but as a
r
result, the governing equations are nonlinear and can only be
A N N+N'dx
d solved numerically.
E I bdx In this study, a rectangle wave is chosen to represent the stiff-
ness of discrete shear connecters. Therefore the constant shear
stiffness K is replaced by a shear stiffness function K f , which is
Fig. 4. Force equilibrium. obtained through Fourier Transform as:
 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209 201

(    
t X 1
1 2pn 2pn 2bK f 2r
K f ðxÞ ¼ K in þ sin cos x u00  u¼ M 0ex ð25Þ
T n¼1 np T T b2 Ew A b2 Ew Itotal
X 1
1     )
2pn 2pn rA 0 M
þ
np
1  cos
T
sin
T
x ð23Þ y00  u ¼  ex ð26Þ
n¼1 Itotal Ew Itotal
where K f = the shear stiffness function of the whole structure; K in = where b2 ¼ ð1  a2 Þ. Then the first order standard form of the set of
the shear stiffness of individual shear connecter; t = length of posi- governing equations is determined based on Eqs. (25) and (26) for
tive phase; T = period length. It is suggested that K in ¼ KðT=tÞ when numerical solution as:
the dimension of shear connecter cannot be accurately obtained. In
Z 01 ¼ k1  Z 2  k2 ð27Þ
this way, the total shear stiffness of all the shear connecters remains
constant, even though the shear connecter width varies. The
Z 02 ¼ Z 1 ð28Þ
rectangular wave function with T ¼ l=6 (l = span length), and
t = 0.05T is used in this study, which is normalized and plotted in
Z 03 ¼ k3  Z 1  k4 ð29Þ
Fig. 6.
K is then replaced with K f in Eq. (16), and the first governing
Z 04 ¼ Z 3 ð30Þ
equation is obtained as:
2bK f where: u
Z1 ¼ 0 ; Z2 ¼ u; Z 3 ¼ y0 ; Z 4 ¼ y; k1 ¼ 2bK f =ðb2 Ew AÞ;
u  u00 ¼ 2ry000 ð24Þ k2 ¼ 2rM ex =ðb2 Ew Itotal Þ; k3
0
¼ rA=Itotal ; k4 ¼ M ex =Ew Itotal .
Ew A
Finally, Eqs. (27)–(30) can be solved with specific boundary
The new set of governing equations is now rewritten by replac-
conditions as the Boundary Value Problem (BVP), and numerical
ing Eq. (17) with Eq. (24) as:
schemes such as finite difference can be employed.
8
< u00  2bK f u ¼ 2ry000 In summary, Allen’s method is concise in the governing equa-
Ew A
tion and its solution, but as a tradeoff, it involves assumptions
: y00  a2 u0 ¼  Mex
2r Ew I total
regarding the shear strain distribution in the middle layer that
may result in inaccuracy in some extreme cases. Granholm’s theory
In order to solve the coupled governing equations, Eq. (24) is does not involve those additional assumptions and is mechanically
mathematically manipulated further to the second order by closer to ICSPs but yields complexity in both of the governing
eliminating the third order term of deflection. The processed set equations and solution. However, they both have an unrealistic
of governing equations is determined as: assumption of constant and continuous shear stiffness in the mid-
dle layer along the span length. Finally, the arbitrary shear resis-
tance theory developed herein is better conditioned in terms of
1 shear stiffness but can only be solved numerically.

0.8 4. Solutions and comparisons

All solutions are based on a simply supported beam subjected to

0.6
K/Kmax

uniform pressure on the top and x = 0 located at the mid-span.

0.4
4.1. Allen’s solution

0.2 A solution to Allen’s governing equation can be obtained by

solving Eq. (8) with specific boundary conditions. The original
boundary conditions are shown in (a) of Fig. 7. However, in order
0
to simplify the solution by symmetry, the problem is solved with
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
equivalent boundary conditions of a half span cantilever beam sub-
jected to the same loads shown in (b) of Fig. 7.
x/ (0.5l)
Deflection and shear stress in the middle layer are solved as:
Fig. 6. Shear stiffness function.

l
l 2

qb qb

x x

y
-q b l/2 -q b l/2 -q b l/2
(a) (b)
Fig. 7. Simply supported beam.
202 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209

2
" !# p p
1 qbl 2x2 cosh e; u3 ¼ sin e  sinh e; k¼ 2=4ðl=l0 Þ; e ¼ 2=4ðx=l0 Þ; l0 ¼
y¼ x2 3  2 p ffiffi
48 Ef I l
4
ðEw  i=2k  bÞ; i ¼ 2Isgl .
 2 " 2 2
# The final solution:
qb If x b l
 1 þ 2 ð1  cosh axÞ ð31Þ yex ¼ jy1 j þ jy2 j þ jy3 j
BG I 2 4h ð36Þ
yin ¼ jy1 j  jy2 j  jy3 j
 
qbl If 2x Bending Moment in the Wythes:
s¼ 1  b2 sinh ax ð32Þ "   !
2bd I l  2 !#
1 2 2b
2
cosh vb x 1 2 2x
M1 ¼ qbl a2 1 þ b 1  ð37Þ
1
where h ¼ a  l=2; b2 ¼ ðh  coshhÞ ; l = span length. 8 vl cosh vl 2
2b
l

4.2. Holmberg’s solution to Granholm’s theory M2 ¼ 0 ð38Þ

pffiffiffi
Based on Granholm’s governing equations, Holmberg provides 2 2 l0 1  0 
comprehensive solution cases. The solution system provided by M3 ¼  qbl u u  u01 u3 ð39Þ
4 l w 3 1
Holmberg [2] is considered more complex and more accurate.
The specific condition in this present study, shown in Fig. 8(a), is The final solution:
(
determined by superposing the three cases illustrated in Mex ¼ jM 1 j  jM3 j
Fig. 8(b)–(d). The first case shown in Fig. 8(b) determines the ð40Þ
deflection shared by both wythes, then the second and third cases Min ¼ jM 1 j þ jM 3 j
shown in Figs. 8(c) and (d), respectively, determine the relative Axial Force in the Wythes (Prestress not included):
vertical compression between the wythes. All of the three cases "   !   2 !#
are presented here, but the second and third cases are regarded
2
1 qbl 2 2b
2
cosh vb x 1 2x
N1 ¼ a  1 þ 1 ð41Þ
as beam bending on soft medium, details of which can also be 8 r vl cosh vl 2 l
2b
found from related references. In this study, those later two cases
are considered to be trivial. N2 ¼ 0 ð42Þ
Deflection:
4  N3 ¼ 0 ð43Þ
5 qbl 24
x2 16
x4 1
y1 ¼ 1 þ þ
384 Ew Itotal 5 l 5 l 16 The final solution reads
4  2  2 " 2 !  2 !#
qbl a 2b 2b cosh vb x 1 2x Nex ¼ N1
  1 þ 1  ð44Þ
Ew Itotal b vl vl cosh vl 2b
2 l Nin ¼ N 1
ð33Þ Slip between the Wythes:
3  2 " v ! #
q 1 qrbl 2b 1 2b sinh b x 2x
y2 ¼  ð34Þ u¼  ð45Þ
4k 4 Ew Itotal vl b2 vl cosh vl l
2b
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where b = ð1  a2 Þ; k = vertical compressive stiffness of middle Middle Layer Shear Stress:
layer; l = span length. " v ! #
pffiffiffi 1 ql 2 2b sinh b x 2x
2 qbl l0 1  0  s¼ a  ð46Þ
y3 ¼  u1 u1 þ u03 u3 ð35Þ 4 r vl cosh 2b
vl l
2
2 Ew Itotal b w
Or alternatively
where W ¼ u01 ðu02 þ u04 Þ  u03 ðu02 þ u04 Þ; u01 ¼ cos k  coshk; u02 ¼
cos k  sinh k; u03 ¼ sin k  sinh k; u04 ¼ sin k  cosh k; u1 ¼ cos e s ¼ Ku ð47Þ

l -q b l/4 -q b l/4
+q b +q b/ 2

x x

y
+q b/ 2
-q b l/2 -q b l/2 -q b l/4 y -q b l/4
(a) (b)
+q b l/4 +q b l/4
+q b/ 2

x x

y
-q b/ 2
y -q b l/4 -q b l/4

(c) (d)

Fig. 8. Load cases.

 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209 203

4.3. Comparison Z 1 ðl=2Þ ¼ Z 2 ð0Þ ¼ Z 3 ð0Þ ¼ Z 4 ðl=2Þ ¼ 0 ð48Þ

The normalized results of deflection, slip, bending moment and
4.3.1. Allen’s and Holmberg’s solutions
axial force (normalized to 0.39) of both models for half of the span
As shown in Fig. 9, Allen’s solution shown in Eq. (31) and either
are compared and plotted as:
Granholm’s original solution or the first case of Holmberg’s solu-
As shown in Fig. 10, deflection and relative slip are not signifi-
tion shown in Eq. (33) would yield identical results for specimens
cantly affected by the Discrete Models. However, the internal
of the same dimensions, even with various middle layer shear stiff-
forces, such as axial force and bending moment, are significantly
ness, which includes both shear connectors and foam. However, it
affected at the vicinity of the shear connectors. Furthermore, two
should be noted that, for the simply supported condition as shown
parameters, t and T as defined in Eq. (23), will affect the results,
in Fig. 8(a), Holmberg supplies the two additional load cases shown
even though the total shear stiffness remains constant. The ratio
in Fig. 8(c) and (d) to conclude the computation although they are
of t/T is usually referred to as ‘‘duty cycle,’’ which ranges from zero
trivial in this study, whereas Allen does not consider those
to one. The bigger the duty cycle is the smoother the results will
scenarios.
become. When the duty cycle is equal to 1, the Discrete Model
becomes the continuous model. For example, in Fig. 10, the
4.3.2. Continuous and Discrete Models
Discrete Model has a duty cycle of 0.05 and the continuous model
The continuous model refers to Granholm’s Model or the first
can be regarded as the Discrete Model with duty cycle of 1, and
case of Holmberg’s solution. Eqs. (27)–(30) can be solved with sim-
they all have the same middle layer total shear stiffness.
ply supported beam boundary conditions, which are given in Eqs.
Therefore it is generally sound to conclude that the continuous
(48). Note that among these boundary conditions, Z 2 ð0Þ ¼ 0,
shear connector model is adequate for general analysis; however,
Z 3 ð0Þ ¼ 0, and Z 4 ðl=2Þ ¼ 0 are directly obtained from the facts that
the maximum stress may no longer occur at mid-span, but rather
slip and slope are zero at the mid-span, and deflection is zero at the
in the vicinity of shear connectors near mid-span and actual stress
end support. However, Z 1 ðl=2Þ ¼ 0 is derived from Eq. (26), with near the shear connecters may be larger than predicted by the con-
the conditions of both external applied bending moment M ex and
tinuous model. Special attention should be paid to the problems
internal bending moment EW Isgl y00 equal to zero at the ends. with large shear stiffness, but small duty cycle.

4.4. Property of solution

4.5 ↓ Non-composite

4 4.4.1. Classification of deflection components

Both Eqs. (31) and (33) consist of two components: flexural
Allen
3.5 deflection and shear deflection. The flexural deflection terms are
Holmberg
identical to the solution of the equivalent solid Euler–Bernoulli
Deflection (mm)

3 ← PCS1
beam:
2.5 2  
1 qbl 2 2x2
yf ¼  x 3 2 ð49Þ
2 48 Ef I l
← PCS2
1.5 4 
5 qbl 24
x2 16
x4
yf ¼ 1 þ ð50Þ
1 384 Ew Itotal 5 l 5 l

0.5 Eqs. (31) and (33) also have the following shear deflection compo-
↑ Fully Composite nents, which are the major differences in the various composite
0 theories:
0 2 4 6 8 10 12 14 16 18 20
 2 " 2 2
#
Normalized Stiffness of Ties (Stiffness of PCS2 as Unit) qb If x b2 l
ys ¼  1 þ ð1  cosh axÞ ð51Þ
Fig. 9. Mid-span deflection of Allen’s and Holmberg’s theories @ q = 10.34 kPa (1.5
BG I 2 4h
psi).

1
1
0.9
0.9
0.8 0.8
Normalized Results
Normalized Results

0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
Discrete Model-Slip
0.2 Discrete Model-Deflection Continuous: Bending Moment
0.2
Continuous Model-Slip Continuous: Axial Force
0.1 Continuous Model-Deflection 0.1 Discrete: Axial Force
Discrete: Bending Moment
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/ (0.5l) x/ (0.5l)

Fig. 10. Comparison of different models.

204 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209

1 lim u ¼ 0 ð57Þ
ys ¼ K!1
16
4  2  2 " 2 !  2 !#
qbl a 2b 2b cosh vb x 1 2x
  1 þ 1  4.4.4. Upper and lower bounds of internal forces
Ew Itotal b vl vl cosh 2bvl 2 l
The upper and lower bounds in which the mid-span moment
ð52Þ and axial forces converge to are based upon Eqs. (37) and (41):
2
ðq=2Þbl
4.4.2. Upper and lower bounds of deflection lim M ¼ ð58Þ
K!0 8
The upper and lower bounds of Eq. (33) are plotted as non-
!
composite and fully composite in Fig. 9, which can be solved 2
qbl 2
theoretically as: lim M ¼ b =2 ð59Þ
K!1 8

24
x2 16
x4
4
5 qbl
lim y ¼   1 þ ð53Þ
K!0 384 Ew 2Isgl 5 l 5 l lim N ¼ 0 ð60Þ
K!0

4  !
5 qbl 24
x2 16
x4 2
qbl 2
lim y ¼ 1 þ ð54Þ lim N ¼ a =2r ð61Þ
K!1 384 Ew Itotal 5 l 5 l K!1 8
Eqs. (53) and (54) mean that when the shear stiffness of the
middle layer is sufficiently small, the wythes withstand local flexu- where a2 ¼ 2r 2 A=Itotal , b2 ¼ 1  a2 . a2 and b2 represent the ratios of
ral deformation independently. However, if the shear stiffness of middle layer bending stiffness, and wythes bending stiffness over
middle layer is sufficiently large, the wythes will act together as the overall bending stiffness respectively. The bending moment is
a fully composite beam. a result of the local bending in the wythes, and the axial force is
the result of composite action by force in the shear connectors.
4.4.3. Upper and lower bounds of slip Internal forces will distribute according to bending stiffness ratio
The bounds of relative slip can be calculated based on Eq. (55) only when fully composite action is achieved. For partially compos-
and Fig. 5: ite structures, internal forces will distribute according to

u ¼ u1 þ u2 ¼ 2r  y0 þ u2 ð55Þ
Table 2
For non-composite beams, u2 is zero since the axial force is zero Material properties.
for each wythe. y is obtained from Eq. (53). For fully composite
Young’s Modulus Stiffness of Stiffness of shear
beams, it is reasonable to assume that the relative slip is zero when of wythes, insulation foam, connectors,
considering the assumption made by Euler–Bernoulli beam theory GPa (ksi) kN/m3 (lb/in3) kN/m3 (lb/in3)
that the plane’s normal remains coincident with the beam center PCS 1 35.63 (5167) 84,517 (311.36) 3042 (11.21)
while bending. Finally, the upper and lower bounds of relative slip PCS 2 36.01 (5223) 84,517 (311.36) 196,723 (724.72)
can be obtained as: PCS 3 36.03 (5226) 84,517 (311.36) 196,723 (724.72)
PCS 4 36.91 (5353) 56,347 (207.58) 952 (3.51)
3 
5 rqbl 48
x 64
x3 PCS 5 36.86 (5346) 56,347 (207.58) 8490 (31.28)
lim u ¼  þ ð56Þ PCS 6 36.61 (5310) 56,347 (207.58) 374,678 (1380.30)
K!0 384 Ew Isgl 5 l 5 l

813

203 407 203

G

38
76

F E
76
38

B A D C A B

(a) Test Apparatus (b) Cross Secons (UNIT: mm)

Fig. 11. (a) Test apparatus and (b) cross sections (unit: mm).

Table 1
Cross section details.

A kN (kips) B C D E mm (in) F mm (in) G mm (in)

PCS 1 9.525 mm(3/800 ); @71.62(16.1) Steel C Shape WWR EPS 51 (2) 203 (8) 2@508 (20)
PCS 2 9.525 mm(3/800 ); @71.62(16.1) Carbon Mesh WWR EPS 51 (2) 203 (8) 2@508 (20)
PCS 3 2  200 mm2(#5) bars &9.525 mm(3/800 ); @71.62(16.1) Carbon Mesh WWR EPS 51 (2) 203 (8) 2@508 (20)
PCS 4 9.525 mm(3/800 ); @66.28(14.9) Metal C-pins 71 mm2(#3) @ 45.72 cm(1800 ) XEPS 76 (3) 229 (9) 3@305 (12)
PCS 5 9.525 mm(3/800 ); @66.28(14.9) Composite Pin 71 mm2(#3) @ 45.72 cm(1800 ) XEPS 76 (3) 229 (9) 2@508 (20)
PCS 6 9.525 mm(3/800 ); @71.62(16.1) Carbon Mesh WWR XEPS 76 (3) 229 (9) 2@508 (20)

Where: A is the longitudinal prestressing tendon; B is the shear connector; C is the reinforcement; D is the type of insulation foam; E is the thickness of insulation layer; F is
the total thickness of the sandwich beam; and G is the space between shear connectors.
 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209 205

equilibrium achieved between shear connector deformation and y ¼ yin ¼ yex ¼ y1 ð62Þ
wythe bending.

5.1. Consideration of prestressing forces

5. Experiment setup and validation
Holmberg’s solution is employed for the validation since
Discrete Model will yield the same results in the deflection and
slip. Also, the sandwich panels tested have very large vertical resis-
tances; therefore the second and third terms are trivial.
Prestressing forces are considered as a pair of axial forces plus a
pair of additional bending moments if eccentric prestressing ten-
dons are present. For this study, the prestressing tendons are in
the center of each wythe, therefore only a pair of axial forces is
considered. Since this study is limited to small displacement and

20
25
18

16
20
14

Load (KPa)
Load (KPa)

12 15
10

8 PCS1 Specimen A
10
PCS1 Specimen B
PCS2 Specimen A
6 PCS1 Specimen C
PCS2 Specimen B
Theoretical Solu. w/o Foam
4 Theoretical Solution w/o Foam
Theoretical Solu. w/ Foam 5
Theoretical Solution w/ Foam
Full Composite
2 Full Composite
Non-composite
Non-composite
0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8
Deflection (mm) Deflection (mm)

18

25 16

14
20
12
Load (KPa)

Load (KPa)

10
15
8
10 PCS3-G2 F 6 PCS4 Specimen A
PCS3-G2 DEF Average PCS4 Specimen C
Theoretical Solution w/o Foam 4 Theoretical Solution w/o Foam
5 Theoretical Solution w/ Foam Theoretical Solution w/ Foam
Full Composite
Full Composite 2
Non-composite Non-composite
0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Deflection (mm) Deflection (mm)

18 18

16 16

14 14

12 12
Load (KPa)

Load (KPa)

10 10

8 8
PCS5 Specimen A PCS6 Specimen A
6 PCS5 Specimen B 6 PCS6 Specimen B
PCS5 Specimen C PCS6 Specimen C
4 Theoretical Solution w/o Foam 4 Theoretical Solution w/o Foam
Theoretical Solution w/ Foam Theoretical Solution w/ Foam
2 Full Composite 2 Full Composite
Non-composite Non-composite
0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Deflection (mm) Deflection (mm)

Fig. 12. Deflection–resistance comparisons.

206 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209

linear elastic behaviors, curvature and deflection will not be
affected by this pair of axial forces.
5.2. Experiment setup
Data from experiments conducted by Naito [23,28–32] are used
for validation. Displacements were measured as a distributed load
that was applied by a load tree apparatus that simulated uniform
pressure. The test apparatus, and cross section properties of sand-
wich panels tested are illustrated in Fig. 11 and Table 1. The span of
all specimens is 3.05 m (120 in.).
5.3. Mid-span deflection validation
The upper and lower bounds of deflection are given in Eqs. (53)
and (54) of Section 4.4.2. Erroneous sensor data and deflection
associated with seating of the samples were eliminated.
Several characteristics of behavior can be observed. Firstly, the
testing results will usually fall between the theoretical solutions
with and without the middle layer foam, although it is yet to be
studied to what extent the foam shear stiffness is effective.
Secondly, the comparisons between the testing results and the
theoretical solutions confirm that the stiffness influences the
deflection in a nonlinear way, as can be also observed from
Fig. 9. For example, the comparison of PCS1 and PCS2 shows that
the same additional foam will help reduce the deflection more
effectively for specimen with small existing shear connecter
stiffness compared to the one with large existing shear connecter
stiffness. Furthermore, the behavior related to a common manufac-
turing flaw of concrete sandwich structures, poor embedment of
the shear connectors, can be observed from the PCS6 specimens.
Although, the shear stiffness of PCS6 is large, as defined in
Table 2, all three tested samples show non-composite behaviors
shortly after loading begins. For example, the theoretically pre-
dicted slopes of deflection are much larger than that demonstrated
by the testing results, and the slopes of the testing results are actu-
ally very close to the lower bound, non-composite action. This is
because when the poor bonding of the shear connecter embed-
ment is smaller than the demand exerted by the applied load, shear
connecter embedment failure occurs and the two concrete wythes
act independently as non-composite structures. Nonetheless, the
deflection comparisons of Fig. 12 demonstrate that the developed
methodology can capture the characteristic of sandwich structures
with reasonable accuracy considering the random characteristics
of concrete structures and uncontrollable factors associated with
the manufacturing process.
5.4. End slip validation
The upper and lower bounds of deflection are given in Eqs. (56)
and (57) of Section 4.4.3. It can be noted from Fig. 13 that the rela-
tive slip predicted at the ends of the specimens correlates well with
the test results. However due to manufacturing imperfections, the
slip at one end may differ slightly from the slip at the other end.
6. Pattern of internal forces and stresses
Eqs. (37) and (41) are employed. The dimensions of PCS1 under
pressure of 10.34 kPa is taken as an example. The internal forces

30 30

25 25

20 20
Load (KPa)

Load (KPa)

15 15

10 10 PCS3-G2 Specimen East

PCS1 Specimen East PCS3-G2 Specimen West
Theoretical Solution w/o Foam Theoretical Solution w/o Foam
5 Theoretical Solution w/ Foam 5 Theoretical Solution w/ Foam
Non-composite Non-composite
Full Composite Full Composite
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Slip (mm) Slip (mm)

20 22
18 20
16 18

14 16
14
Load (KPa)

Load (KPa)

12
12
10
10
8
8
PCS4 Specimen East PCS5 Specimen East
6 PCS4 Specimen West 6 PCS5 Specimen West
Theoretical Solution w/o Foam Theoretical Solution w/o Foam
4 4
Theoretical Solution w/ Foam Theoretical Solution w/ Foam
2 Non-composite Non-composite
2
Composite Composite
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Slip (mm) Slip (mm)

Fig. 13. Load-slip relationship.

 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209 207

1 1

0.9 ← Where Stress Plotted 0.9

0.8 0.8

Normalized Results
Normalized Results

0.7 0.7

0.6 0.6
Bending Moment
0.5 Axial Force 0.5

0.4 0.4

0.3 0.3 Constant Model: Exterior Top

0.2 Constant Model:Exterior Bottom
0.2 Discrete Model: Exterior Top
0.1 0.1 Discrete Model:Exterior Bottom

0 0
0 2 4 6 8 10 12 14 16 18 20 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
K/ (K@PCS1) x/ (0.5l)
(a) (b)
Fig. 14. (a) Mid-span internal forces and (b) stresses.

and stresses are plotted with an assumption of 60 times of original
shear connectors’ stiffness.
Fig. 14 shows how of shear connector stiffness influences the
internal forces and the related stress pattern. As expected, the
bending moment decreases and the axial force increases to their
bounds, respectively, when shear connectors become sufficiently
strong. Then, at the normalized stiffness of 2.1, exterior wythe
stresses are plotted with both Continuous and Discrete Models. It
should be noticed that the Discrete Model will yield larger stresses
near shear connectors and the maximum value exists at the first
one or two connectors from the mid-span.
for relatively thick insulation layers, it is more difficult to achieve
fully composite action than for thin insulation layers (the 3–4–3
panel would need about 3 times the number of connectors com-
pared to the 3–3–3 panel, and 5 times compared to 3–2–3). After
achieving fully composite action, the 3–4–3 panel would be stiffer
and have less deflection due to the larger moment of inertia, but
for average sandwich panel design, the deflection associated with
shear dominates and consequently a thicker insulation layer results
in a larger deflection. Therefore it is not sound to conclude that a
larger moment of inertia would help reduce deflection in sandwich
structures.

7. Influence of different insulation layer thickness on overall 4.5

behavior
4 ↑ Non-composite

Five insulation setups are considered as illustrated in Fig. 15. 3-4-3

The insulation layers’ thicknesses vary but the concrete wythes’ 3.5 3-3-3
3-2-3
Deflection (mm)

dimensions are the same. Also, the assumption is that the same
3 3-1-3
type of shear connectors and insulation foam are used, but their 3-0.5-3
heights can vary according to different middle layer thickness. 2.5
In order to study the effect of insulation layer thickness varia-
tion, the assumption that the shear connectors’ ends are built into 2
the wythes as fixed boundary conditions is made. In that case,
, 1.5
3
p pð2r  dÞ 12EI 1
K ¼ s=u ¼ ¼ ð63Þ
A 12EI A ð2r  dÞ3 1

It can be observed in Eq. (63) that K, as a function of thickness of the 0.5

middle layer 2r  d, changes exponentially. The deflection is plotted 0 1 2 3 4 5 6 7 8 9 10
in Fig. 16 as a function of middle layer stiffness for the five cross- Normalized Stiffness
sections. Parameters and dimensions of PCS6 are taken under
10.34 kPa pressure for demonstration. Fig. 16 demonstrates that Fig. 16. Influence of middle layer on mid-span deflection.

812.8 812.8
812.8 812.8 812.8

76.2 76.2
76.2 76.2 76.2 88.9
44.4 50.8 63.5 76.2
44.4 50.8 63.5 76.2 88.9
76.2 76.2 76.2 76.2 76.2

812.8 812.8 812.8 812.8 812.8

3:0.5:3 3:1:3 3:2:3 3:3:3 3:4:3

Fig. 15. Cross-section and dimension ratios considered (unit: mm).

208 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209

8. Conclusions Granholm’s and Holmberg’s Theories

This research reviews and compares various existing composite u2 slip between wythes due to axial
theories that could be used for analysis of insulated concrete sand- displacement
wich panels and derives a rigorous discrete analysis model for Ew moduli of elasticity of wythes
composite structures. Most importantly, it addresses the absence s shear stress in the middle layer
of reliable analysis methodology in the ICSP industry. Key findings N resultant axial force in wythe
and accomplishments of the work include: b wythe width
d thickness of wyth
1. Both Allen’s and Granholm’s theories yield the same results A = bd cross section area of wythe
when applied to ICSP structures (demonstrated in Fig. 9). K shear stiffness of the middle layer
2. The Discrete Model can improve the stress calculation accuracy u = u1 + u2 total slip between wythes
near shear connectors and capture the key characteristic of ICSP u1 relative slip due to bending
structures. y total deflection
3. Deflection of sandwich structures are decomposed and classi- r distance from neutral axis of wythe to the
fied as flexural and shear components. Furthermore, the upper overall neutral axis
and lower bounds of deflection, end slip and internal forces v2 2Kb/(Ew AÞ
are derived and verified against full-scale test data. Mex external bending moment
4. The deflection associated with shear deformation can be so internal bending moment of interior
M in
large that it can dominate the deflection when the middle layer wythe
stiffness is relatively small (demonstrated in Fig. 16). M ex internal bending moment of exterior
wythe
Future work will include the theoretical development of 3 moment of inertia of each wythe
Isgl ¼ bd =12
unsymmetrical wythe ICSPs, the development of composite beam
Itotal ¼ 2Isgl þ 2r 2 A moment of inertia for the whole cross-
‘‘elements’’ that can be used for finite element analyses of ICSPs,
section
and additional validation and evaluation of the practical implica-
a2 2r2 A=Itotal
tions of the developed theories for design optimization of ICSPs.
b2 1  a2
Acknowledgement k vertical compressive stiffness of the
middle layer
The financial support provided by the China Scholarship Council l span length
   
(CSC) is appreciated. The first author also benefits from the discus- W u01 u02 þ u04  u03 u02 þ u04
sion with Wenxiu Liu, Syracuse University and her AutoCAD u01 cos kcosh k
drawings. u02 cos ksinh k
u03 sin ksinh k
Appendix A. Notations and symbols u04 sin kcosh k
u1 cos ecosh e,
(In the order of appearance). u3 sin esinh e,
Allen’s Theory p
k 2/4(l/l0 Þ
p
e 2=4ðx=l0 Þ
ys shear deflection pffiffiffi
l0 4ðEw  i=2k  bÞ
c ¼ 2r  d middle layer thickness
i 2Isgl
d wythe thickness
yf flexural deflection
2r distance between centroids of the two wythes
ys shear deflection
c shear strain of middle layer
Q 1 ¼ Ef Iy000
f
internal shear force resultant of flexural
deflection in wythes and middle layer
Q 2 ¼ Ef If y000 internal shear force resultant of shear Discrete Model
s
deflection in wythes (Notations are the same with Granholm’s model except
b width of cross section following)
s shear stress in the middle layer
Ef modulus of elasticity for wythes
Kf the shear stiffness function of the whole structure
If bd3/6 = moment of inertia of two wythes
K in the shear stiffness of individual shear connecter
yf flexural deflection
t length of positive phase
G shear modulus of the middle layer
T period length
B b  ð2rÞ2 =c Z1 u0
I 3 2
bd =6 þ 2bdr = moment of inertia of the Z2 u
whole cross section Z3 y0
Q external applied shear force Z4 y
a2 BG=ðEf If ð1  If =IÞÞ k1 2bK f =ðb2 Ew AÞ
y total deflection k2 2rM0ex =ðb2 Ew Itotal Þ
h a  l=2 k3 rA=Itotal
b2 1
ðh  coshhÞ k4 M ex =Ew Itotal
l span length
 F. Bai, J.S. Davidson / Engineering Structures 91 (2015) 197–209 209

References [24] Pessiki Stephen, Mlynarczyk Alexandar. Experimental evaluation of the

composite behavior of precast concrete sandwich wall panels. PCI J 2003;
48(2):54–71.
[1] Granholm H. Om sammansatta balkar och pelare med särskild hänsyn till
[25] Salmon David C et al. Full scale testing of precast concrete sandwich panels.
spikade träkonstruktioner. (On composite beams and columns with particular
ACI Struct J 1997;94(4):354–62.
regard to nailed timber structures.) Transaction of Chalmer University of
[26] Bush Thomas D, Zhiqi Wu. Flexural analysis of prestressed concrete sandwich
Technical, No. 88. Gothenburg; 1949.
panels with truss connectors. PCI J 1998;43(5):76–86.
[2] Holmberg Ake, Plem Erik. Behaviour of load-bearing sandwich-type structures.
[27] Salmon David C, Einea Amin. Partially composite sandwich panel deflections. J
Handlingar nr 49 Transactions, Statens institut för byggnadsforskning; 1965.
Struct Eng 1995;121(4):778–83.
[3] Allen Howard G. Analysis and design of structural sandwich panels, vol.
[28] Frankl Bernard A, Lucier Gregory W, Hassan Tarek K, Rizkalla Sami H. Behavior
51. Oxford: Pergamon Press; 1969.
of precast, prestressed concrete sandwich wall panels reinforced with CFRP
[4] Hartsock John A. Design of foam-filled structures. CT: Technomic Publishing
shear grid. PCI J 2011;56(2):42–54.
Company; 1969.
[29] Naito Clay, Dinan Robert, Bewick Bryan. Use of precast concrete walls for blast
[5] Ha KH. Exact analysis of bending and overall buckling of sandwich beam
protection of steel stud construction. J Perform Construct Facilities
systems. Comput Struct 1992;45(1):31–40.
2011;25(5):454–63.
[6] Ha KH. Finite element analysis of sandwich plates: an overview. Comput Struct
[30] Naito Clay J, Hoemann John M, Shull Jonathon S, Saucier Aaron, Salim Hani A,
1990;37(4):397–403.
Bewick Bryan T, Hammons Michael I. Precast/prestressed concrete
[7] Davies J Michael. The analysis of sandwich panels with profiled faces. In:
experiments performance on non-load bearing sandwich wall panels. Air
Eighth international specialty conference on cold-formed steel structures.
Force Research Laboratory Report, AFRL-RX-TY-TR-2011-0021, Panama City
Missouri S&T (formerly the University of Missouri-Rolla), St. Louis, Missouri,
(FL): Tyndall Air Force Base; 2011.
November 11–12; 1986.
[31] Naito Clay, Beacraft Mark, Hoemann John, Shull Jonathan, Salim Hani, Bewick
[8] Gordaninejad, Faramarz, Bert Charles W. A new theory for bending of thick
Bryan. Blast performance of single-span precast concrete sandwich wall
sandwich beams. Int J Mech Sci 1989;31(11):925–34.
panels. J Struct Eng 2014;140(12):04014096.
[9] Goyal Vijay K, Kapania Rakesh K. A shear-deformable beam element for the
[32] Naito Clay J, Hoemann John M, Bewick Bryan T, Hammons Michael I.
analysis of laminated composites. Finite Elem Anal Des 2007;43(6):463–77.
Evaluation of shear tie connectors for use in insulated concrete sandwich
[10] Noor Ahmed K, Scott Burton W, Bert Charles W. Computational models for
panels. Interim report, AFRL-Rx-TY-TR-2009-4600; 2009.
sandwich panels and shells. Appl Mech Rev 1996;49(3):155–99.
[33] Pfeifer, Wayne Donald, Hanson JA. Precast concrete wall panels: flexural
[11] Goodman James Richard. Layered wood systems with interlayer slip. PhD
stiffness of sandwich panels. ACI Special Publication 1965;11.
thesis, University of California, Berkeley, Calif; 1967.
[34] Woltman Greg, Tomlinson Douglas, Fam Amir. Investigation of various GFRP
[12] Newmark Nathan M, Siess Chester P, Viest IvanM. Tests and analysis of
shear connectors for insulated precast concrete sandwich wall panels. J
composite beams with incomplete interaction. Proc Soc Exp Stress Anal
Compos Constr 2013;17(5):711–21.
1951;9(1):75–92.
[35] Tomlinson Douglas, Fam Amir. Experimental investigation of precast concrete
[13] Girhammar Ulf Arne, Gopu Vijaya KA. Composite beam-columns with
insulated sandwich panels with glass fiber-reinforced polymer shear
interlayer slip – exact analysis. J Struct Eng 1993;119(4):1265–82.
connectors. ACI Struct J 2014;111:1–6.
[14] Ranzi G, Bradford MA, Uy B. A direct stiffness analysis of a composite beam
[36] Benayoune A, Samad AAzizA, Trikha DN, Abdullah Abang Ali A, Ashrabov AA.
with partial interaction. Int J Numer Methods Eng 2004;61(5):657–72.
Structural behavior of eccentrically loaded precast sandwich panels. Constr
[15] Ranzi G, Gara F, Ansourian P. General method of analysis for composite beams
Build Mater 2006;20(9):713–24.
with longitudinal and transverse partial interaction. Comput Struct
[37] Gara Fabrizio, Ragni Laura, Roia Davide, Dezi Luigino. Experimental tests
2006;84(31):2373–84.
and numerical modelling of wall sandwich panels. Eng Struct 2012;37:
[16] Ranzi Gianluca. Locking problems in the partial interaction analysis of multi-
193–204.
layered composite beams. Eng Struct 2008;30(10):2900–11.
[38] Benayoune A, Abdul Samad AA, Abang Ali AA, Trikha DN. Response of pre-cast
[17] Reza Salari M et al. Nonlinear analysis of composite beams with deformable
reinforced composite sandwich panels to axial loading. Constr Build Mater
shear connectors. J Struct Eng 1998;124(10):1148–58.
2007;21(3):677–85.
[18] Sousa Jr João Batista M, da Silva Amilton R. Analytical and numerical analysis
[39] Benayoune A, Samad AA, Trikha DN, Ali AA, Ellinna SHM. Flexural behaviour of
of multilayered beams with interlayer slip. Eng Struct 2010;32(6):1671–80.
pre-cast concrete sandwich composite panel – experimental and theoretical
[19] Fabbrocino G, Manfredi G, Cosenza E. Analysis of continuous composite beams
investigations. Constr Build Mater 2008;22(4):580–92.
including partial interaction and bond. J Struct Eng 2000;126(11):1288–94.
[40] Bush Thomas D, Stine Gregory L. Flexural behavior of composite precast
[20] Xu Rongqiao, Wu Yu-Fei. Two-dimensional analytical solutions of simply
concrete sandwich panels with continuous truss connectors. PCI J 1994;
supported composite beams with interlayer slips. Int J Solids Struct 2007;
39(2).
44(1):165–75.
[41] Einea Amin. Structural and thermal efficiency of precast concrete sandwich
[21] Xu Rongqiao, Ding Haojiang. Two-dimensional solutions for orthotropic
panel systems. Paper AAI9308172, Lincoln: ETD collection for University of
materials by the state space method. Compos Struct 2007;78(3):325–36.
Nebraska, January 1; 1992.
[22] Hassan Tarek K, Rizkalla Sami H. Analysis and design guidelines of precast,
[42] Einea Amin, Salmon David C, Tadros Maher K, Culp Todd. A new structurally
prestressed concrete, composite load-bearing sandwich wall panels reinforced
and thermally efficient precast sandwich panel system. PCI J 1994;39(4):
with CFRP grid. PCI J 2010;55(2):147–62.
90–101.
[23] Naito C, Hoemann J, Beacraft M, Bewick B. Performance and characterization of
shear ties for use in insulated precast concrete sandwich wall panels. J Struct
Eng 2012;138(1):52–61.