Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP.

108-123

An Equivalent Thickness for Buckling Verification

of Laminated Glass Panels Under
In-Plane Shear Loads
Claudio Amadio1, Chiara Bedon*2
University of Trieste, Department of Civil Engineering and Architecture (DICAR)
Piazzale Europa 1, 34127 Trieste, Italy
1
amadio@units.it; 2bedon@dicar.units.it; 2c.bedon@libero.it

Abstract- Monolithic and laminated glass elements, as known, offer interesting opportunities in the realization of innovative
architectures of modern buildings. Nevertheless, similar structural elements are typically brittle and slender, thus frequently
subjected to buckling phenomena. In these hypotheses, the paper focuses on the load-carrying behavior of 2-layer and 3-layer simply
supported laminated glass panels subjected to in-plane shear loads. Analytical formulations based on the concept of equivalent
thickness are presented to describe with accuracy their typical behavior. As shown, predicted critical loads and load-transversal
displacement relationships are in good agreement with numerical data obtained by using sophisticated 3D-FE models, as well as
simplest but accurate geometrical simplified FE models. According to the suggestions that the Eurocodes give the verification of
traditional structural elements, a suitable verification criterion appropriately calibrated to numerical and experimental predictions
available in literature is suggested to guarantee the requisites of resistance, serviceability and durability typically imposed in the
design of conventional structural systems made of steel, concrete or timber. As a result, the proposed approach could be used in daily
practice to perform a suitable and rational buckling verification of such brittle load-bearing elements.
Keywords- Equivalent Thickness; Sandwich Theory; In-Plane Shear Loads; 2 Or 3-Layer Laminated Glass Panels; Buckling
Verification; Standardized Buckling Curve

I. INTRODUCTION
Structural glass elements are frequently adopted as stiffeners for roofs in modern buildings or in the construction of
innovative and futuristic architectures. Consequently in-plane or out-of-plane loads could represent the cause of possible
instability of these brittle and slender elements [1, 2, 3].
Several authors investigated the buckling behaviour of glass panels or beams in different loading conditions, providing
interesting experimental results and sophisticated numerical validations. Belis [4], for example, focused on the out-of-plane
bending of laminated glass beams and performed more than 300 tests on laminated glass (LG) beams having various
mechanical and geometrical properties. Luible [5] studied the buckling response of compressed columns and panels, as well as
beams in out-of-plane bending. Englhardt [6] recently analysed, through experimental and numerical predictions, the buckling
response of in-plane compressed monolithic and laminated glass panels. Also Mocibob [7] deeply investigated the buckling
response of glass panels supported at top and bottom sides and subjected to in-plane shear forces (e.g. lateral wind acting on a
façade of the building and transferred to bracing glass panels by floor slabs), out-of-plane distributed loads (e.g. perpendicular
wind) and in-plane compressive forces (e.g. self weight). Through numerical and experimental investigations, Mocibob
analysed the buckling response of these glass panels, highlighting the effects of point or linear connections, as well as the
interaction of simultaneous loads on their global behaviour. Similarly, Wellershoff and Sedlacek [8] performed tests on
monolithic and laminated glass panels simply supported on the four edges and subjected to in-plane shear loads, providing
interesting results and considerations. In addition, Wellershoff focused on the use of glass panels in space grid structures and
studied analytically, numerically and experimentally the shear buckling response of flat glass panels glued to grid members
along the four edges [9]. Further recent experimental investigations on buckled glass elements in various boundary and loading
conditions can also be found in [10, 11, 12]. Nevertheless, the knowledge on LG panels behavior under in-plane loads is still
limited and with constrained applications.
In this context, it should not be ignored that analytical models existing in literature are in general derived from classical
sandwich theory formulations, thus they well apply only to 2-layer composite elements and to specific loading and boundary
scenarios [13, 14]. At the same time, it is known that consolidate verification criteria available in literature for buckling
verification of traditional structural elements, realized by means of conventional materials as steel, concrete or timber [15, 16,
17], cannot be directly applied to LG elements, because they do not take into account a series of factors typical of glass
structures (e.g. influence of production tolerances, initial imperfections, brittle behavior of glass, viscoelastic behavior of
thermoplastic interlayers,…).
Based on these assumptions, an analytical approach which requires the contemporary check of maximum stresses,
deformations and simultaneous acting loads has recently been proposed by the authors for the buckling verification of LG
beams under in-plane compression or out-of-plane bending [18, 19]. Also a new analytical approach, based on the concept of

- 108 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

equivalent thickness, has been proposed for a rational buckling verification of LG panels under in-plane compression [20]. By
means of opportunely calibrated correction factors, the model estimates accurately the critical buckling load of these composite
panels in various boundary conditions (linear supports, point supports, etc.). In addition, the equivalent thickness approach can
be used to simplify numerical simulations, thus to predict with a good level of accuracy their critical load as well as the
corresponding load-transversal displacement relationship [20]. As shown in [21], the approach can be rationally applied to the
verification of 2 and 3-layer LG panels and columns under compression, by simply calibrating a series of opportune correction
factors.
Based on these last considerations, in this paper the same equivalent thickness approach is proposed to perform a rational
buckling verification of 2 and 3-layer simply supported LG panels under in-plane shear loads. As shown, predicted critical
shear loads and load-transversal displacement relationships are in good agreement with sophisticated numerical simulations.
As a result, the buckled response of a simply supported LG panel under in-plane shear can be predicted, for a well-defined
temperature and load-time scenario [18, 19, 20, 21]. At the same time, in accordance with the suggestions of Eurocodes 3, 4, 5
[15, 16, 17], a verification criterion is proposed to guarantee the requisites of resistance, serviceability and durability typically
imposed by standards in the design of conventional structural systems. As highlighted in the following sections, the proposed
verification criterion agrees with numerical results of performed simulations, as well as with experimental data collected in
literature, thus it could be used by designers in daily practice.

II. 2-LAYER LAMINATED GLASS PANELS UNDER IN-PLANE SHEAR

A. Existing Analytical Models for Monolithic Panels

As proposed in Fig. 1, let us consider a flat monolithic glass panel (height a and width b , thickness t ; Young’s modulus
E , Poisson’s coefficient  ) simply supported along the edges and subjected to a shear loading N xy  t  xy  V . The well-
known differential equation able to describe its load-carrying behavior is [22]:

2w ,
D 2 w  2 V (1)
xy
where D and  are respectively defined as:

D  E t 3 12 (1  2 ) (2)

the flexural stiffness of the element per unit width b and  the Laplace’s operator.

Fig. 1 Simply supported flat panel subjected to in-plane shear. (a)geometry; (b) deformed configuration

Assuming for the out-of-plane deflection w  w( x, y) of the panel an appropriate form and introducing it in Eq. (1), the
resulting critical shear load is defined as:

 2D
Vcr( E )  k . (3)
b2
In Eq. (2), the shear buckling coefficient k is commonly expressed as a function of the aspect ratio   a b (Fig. 1) and the
boundary conditions of the panel. In the case of four sides simply supported elements, for example, k can be evaluated as [23]:

 5.34
4.00   2  1
k   . (4)
5.34  4.00  1
 2

- 109 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

These formulations can be only applied to monolithic panels under in-plane shear. Nevertheless, with opportune attention,
similar considerations can be extended also to the analysis of 2 and 3-layer laminated glass panels under in-plane shear forces.

B. Existing Analytical Models for 2-Layer Panels

In general, the analysis of a 2-layer LG element is performed by directly taking into account the existing formulations
originally proposed for the calculation of sandwich structures. A similar approach often provides accurate results also for
laminated glass elements, as checked by Luible for in-plane compressed panels [5]. Nevertheless, the theoretical background of
sandwich models does not agree well with the mechanical behavior of LG elements consisting in very thin layers and
extremely soft middle films, as highlighted in the following sections. Consequently, particular attention is generally required.
For this purpose, let us examine a 2-layer LG panel (height a , width b , Fig. 1) composed of two monolithic glass sheets
(thicknesses t1 and t2 ; Young’s modulus E , Poisson’s ratio  ) and a middle interlayer (thickness tint ; Young’s modulus Eint ,
shear modulus Gint , Poisson’s ratio  int ), simply supported along the four edges.

As noted by Mocibob [7], Kuenzi et al. solved the problem of rectangular isotropic sandwich plates, simply supported or
clamped along the edges, subjected to in-plane shear loads [24]. By applying Kuenzi’s expression to the examined LG panel,
the equilibrium equation of forces and moments due to in-plane shear force V in the deformed configuration is:

Di Dlam  2 w  w 1 
3 w  2 w  2 V   , (5)
Sint D0  Dc xy  Sint D0  Dc 
where:

E (t13  t23 )
Di  (6)
12 (1  2 )
is the flexural stiffness of the glass sheets around their neutral axes;

E (t1  t2  tint ) d 2
D0  (7)
2 (1  2 )
is the flexural stiffness of the glass sheets around the centroidal axis of the total cross section;
t1 t2
d   tint (8)
2 2
is the distance between the centroidal axes of the glass sheets;
3
Eint tint
DC  (9)
12 (1  int
2
)
is the flexural stiffness of the middle interlayer;
Dlam  Di  D0  DC (10)
represents the flexural stiffness of the LG panel;

Gint d 2
Sint  (11)
tint
is the shear stiffness of the interlayer.
In this hypothesis, the critical shear load Vcr( E,lam
)
is:

 2 Dlam
Vcr( E,lam
)
 k ,lam , (12)
b2
where the buckling coefficient k ,lam is defined by Kuenzi as a function of the aspect ratio  , the shear stiffness Sint of the
adopted interlayer and the applied boundary conditions. For a panel simply supported along the edges its value is:
16 4

k ,lam  3 2 . (13)
 2 Dlam  13 3 
1 2   
b Sint  3  2 

- 110 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

Similarly, for a clamped panel k ,lam should be estimated as:

17
9
k ,lam  32 . (14)
 Dlam  23 13 
2
1 2   
b S int  3 3 2 

C. Equivalent Thickness Approach for 2-Layer LG Panels

An alternative formulation for the analytical evaluation of the critical buckling load of simply supported 2-layer LG panels
under in-plane shear can be derived from the simplified approach based on the concept of equivalent thickness originally
formulated by Wölfel for the analysis of sandwich structures [25]. The original procedure, actually proposed by the American
[26] and Australian Standards (AS1288) for the verification of LG beams, requires the evaluation of a monolithic beam of
“effective” thickness with equivalent bending properties to a sandwich beam.
In accordance with this theoretical model, the level of connection effectively offered by the adopted interlayer can be
expressed by means of a shear transfer coefficient Γ , comprised between 0 (layered limit) and 1 (monolithic limit), defined as:
1
 , (15)
EJ s tint
1   2

Gintt s2 2
where:
  min( a, b) (16)

is a scale factor (minimum dimension of the panel) and the equivalent parameters J s , t s ,1 , t s , 2 and t s are respectively defined
as (Fig. 2):

J s  t1 t s2, 2  t 2 t s2,1 ; (17)

t s ,1  t s t1 (t1  t 2 ) ; (18)

t s , 2  t s t 2 (t1  t 2 ) ; (19)

t s  0.5 (t1  t2 )  tint . (20)

In addition,  is a coefficient defined as a function of the specific boundary and loading conditions (in the original
formulation,   1 [25]).

Fig. 2 Schematic view of 2-layer laminated glass panel (cross section)

Based on the original analytical approach, the deformation w of the composite element can be evaluated referring to an
equivalent thickness defined as:

t eq,w  3 t13  t 23  12 Γ J s , (21)

in which  and Js are respectively given by Eq. (15) and Eq. (17).
Similarly, the calculation of the maximum bending stresses in each glass sheet can be performed by taking into account two
additional effective thicknesses (one for each glass pane [27]).

- 111 -
 Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

In the specific case of simply supported 2-layer LG beams composed of two monolithic glass sheets and a middle interlayer,
subjected to an uniformly distributed load q acting orthogonally to their plane, assuming   1 [25] and evaluating the
corresponding equivalent thickness t eq ,w (Eq. (21)), it is possible to accurately describe the associated load – midspan
maximum displacement relationship. Undoubtedly, a similar approach only applies to 2-layers LG beams in bending in well-
defined conditions of temperature and load duration. Nevertheless, as recently highlighted by Bennison et al. [27], it represents
an important simplification for the analysis of similar composite elements. Also Calderone et al. [28] demonstrated that the
equivalent thickness formulation adequately calculates stresses and deflections in each glass layer, thus it could represent a
useful approach for designers in daily practice.
At the same time, it should not be ignored that Wölfel’s formulation was originally proposed for the analysis of sandwich
structures characterized by a very soft core and metallic faces. In particular, Wölfel’s analytical model has been developed
assuming that:
 the external layers are characterized by noticeable axial stiffness but negligible bending stiffness;
 the middle interlayer can be defined only in terms of shear stiffness, whereas its axial and flexural rigidities can be
ignored.
Evidently, these assumptions do not apply to 2-layer LG elements. Because of this reason, a series of analytical calculations
has been performed by the authors to detect if Wölfel-Bennison’s approach can be used for the analysis of LG panels simply
supported along the four edges and subjected to in-plane shear forces. Specifically, Vcr( E,lam
)
has been evaluated, for different LG
panels, both using the linear elastic sandwich theory (Eq. (12)) proposed by Kuenzi [24] and the equivalent thickness approach,
that is by substituting t eq , w (Eq. (21), with   1 ) in Eq. (3) and

Deq  E t eq , w 12 (1   )
3 2
(22)

in Eq. (3).
Analytical calculations have been carried out highlighting the effects on Vcr( E,lam
)
of mechanical and geometrical parameters
characterizing a typical 2-layer LG panel, that is the value of Gint ( 10 4 N/mm 2  Gint  10 4 N/mm 2 ), the aspect ratio 
( 1    10 , with a fixed width b  1m ), the thicknesses of glass sheets and interlayer ( 6/1.52/6mm , 8/1.52/8mm ,
10/1.52/10m m ). However, since the aim of this work is to provide useful criteria for the buckling verification of simply
supported LG panels under in-plane shear, only the first buckling load was considered in these comparisons ( m  1 , Fig. 1).
Results presented in Fig. 3, for example, refer to a squared (1 m x 1 m) 2-layer LG panel ( 8/1.52/8mm ) simply supported
along the four edges. As known, depending on the effective level of connection offered by the adopted interlayer, that is
depending on the value of the shear modulus Gint , the critical shear load of a generic 2-layer LG panel is always comprised
between the layered limit (abs, no connection between the glass sheets) and monolithic limit (full, rigid connection between the
glass sheets), which can be rationally estimated as:
 layered limit (abs, Gint  0 ):

 2 Dabs
Vcr( E,lam
)
 Vcr( E,abs
)
 k , (23)
b2
with k given by Eq. (4) and Dabs by Eq. (6);

 monolithic limit (full, Gint   ):

 2 D full
Vcr( E,lam
)
 Vcr( E, full
)
 k , (24)
b2
With
D full  E t 3full 12 (1  2 ) (25)

and
t full  t1  t int  t 2 . (26)

- 112 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

3500
8/1.52/8mm
a= 1m x b= 1m
3000
Full (Eq.(24))
Kuenzi (Eq.(12))
2500 teq,w (Eq.(3), with = 1)
Abs (Eq.(23))

V(E)cr,lam [kN]
2000

1500

1000

500

0
-4 -3 -2 -1 0 1 2 3 4
10 10 10 10 10 10 10 10 10
Gint [N/mm2]
Fig. 3 Critical shear load Vcr( E,lam
) for 2-layer simply supported laminated glass panels under
in-plane shear. Analytical and numerical comparisons (= 1)

As shown in Fig. 3, the equivalent thickness approach (with = 1), allows estimating these limit shear buckling loads for
the examined 2-layer LG panel. Specifically, Vcr( E,lam
)
 Vcr( E, full
)
in presence of a very stiff interlayer ( Gint  1000 N/mm 2 , in this
example), whereas Vcr( E,lam
)
 Vcr( E,abs
)
if the connection between glass sheets is soft ( Gint  0.01N/mm 2 for the examined LG
panel).
Performed analytical comparisons highlighted that also Kuenzi’s approach correctly estimates the full critical shear load
(E)
V of 2-layer LG panels, but strongly underestimates the effective buckling strength of similar composite elements,
cr , full

especially if the adopted interlayer is extremely soft (in this example, Vcr( E,lam
)
 0 for Gint  0.1N/mm 2 , Fig. 3).

The intermediate predicted critical loads Vcr( E,abs

)
 Vcr( E,lam
)
 Vcr( E, full
)
, as shown in Fig. 3, do not agree. Nevertheless, no
alternative analytical formulations are available in literature to check the accuracy of Kuenzi’s approach and the equivalent
thickness formulation. Based on these assumptions, numerical simulations were performed to validate both the analytical
approaches.

D. Numerical Validation and Analytical Comparisons for 2-Layer Laminated Glass Panels
An accurate three-dimensional numerical finite element model was developed with the non linear code ABAQUS [29]. In
this FE-model (3D + shell), glass sheets (thickness t1  t2  8mm ) have been modeled by means of shell elements (S4R),
whereas PVB-interlayer (thickness tint  1.52mm ) has been described through 3D-8 node elements (C3D8H, hybrid
formulation, compatible modes). For the examined LG panel, having dimensions a  1m x b  1m , a mesh based on 20 mm x
20 mm module was assumed. Over the depth of the interlayer, two 3D elements have been realized (Fig. 4).

Fig. 4 Detail of 3D+shell FE-model for a simply supported 2-layer laminated

glass panel subjected to in-plane shear (ABAQUS)

- 113 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

Accordingly with Luible and Crisinel [30] and previous investigations [20, 21], 3D elements and shell elements were
connected together using the same nodes. Moreover, to describe the effective geometry of the examined panel, a section offset
toffset  4mm from the centroidal axis of each glass sheet was applied to monolithic shell elements. In-plane shear was
introduced in the model in the form of shear nodal loads acting on the middle nodes of interlayer. To constrain all four edges of
the simply supported LG panel and to avoid possible eccentricities, also boundary conditions have been applied to the central
nodes of the interlayer.
Glass has been defined as an isotropic and linear elastic material (Young’s modulus E  70000 N/mm2 , Poisson’s ratio
  0.23 , density   2490 Kg/m3 ). Also the interlayer was considered to behave linear-elastically, assuming a Poisson’s
ratio  int  0.49 and a density int  660 Kg/m3 typical of PVB-films [18, 19, 20, 21].

Since the aim of these numerical simulations consisted in validating the proposed analytical approach for the estimation of
the critical load Vcr( E,lam
)
, buckling analyses were performed in ABAQUS to predict the critical load of the examined LG panel in
a series of well-defined conditions of temperature and load duration. As a result, the interlayer has been characterized in each
of these analyses by a different value of shear modulus Gint , estimated in a pre-established range ( 1  Gint  1000 N/mm 2 ).

Numerical and analytical results presented in Fig. 5, referred to the squared 8/1.52/8mm panels previously investigated
( a  b  1m ), confirm that Kuenzi’s formulation constitutes a useful criterion for the evaluation of the critical shear load
Vcr( E,lam
)
of layered “stiff” plates. Nevertheless, this approach cannot be directly applied to the analysis of LG panels typically
characterized by the presence of stiff external faces (the glass sheets) bonded together by a soft and thin middle interlayers.
3500
8/1.52/8mm
a= 1m x b= 1m
3000
Full (Eq.(24))
Kuenzi (Eq.(12))
2500 teq,w (Eq.(3), with  given by Eq.(27))
ABAQUS (3D+shell)
Abs (Eq.(23))
V(E)cr,lam [kN]

2000

1500

1000

500

0
-4 -3 -2 -1 0 1 2 3 4
10 10 10 10 10 10 10 10 10
2
Gint [N/mm ]
Fig. 5 Critical shear load Vcr( E,lam
) for 2-layer simply supported laminated glass panels under in-plane shear.
Analytical and numerical comparisons (= (), Eq. (27))

At the same time, it should be noticed that the equivalent thickness approach well applies to the prediction of the shear
critical load Vcr( E,lam
)
of 2-layer simply supported LG panels, but accurate results can be obtained only assuming for the
correction factor  a series of appropriately calibrated values [20, 21]. Specifically, parametric numerical simulations and
analytical comparisons highlighted that Vcr( E,lam
)
can be evaluated with Eq. (3), referring to an opportune equivalent thickness
t eq ,w (Eq. (21)), by simply assuming in Eq. (15) a series of coefficients    ( ) , numerically calibrated as a function of the
aspect ratio  (Table 1) and well expressed by the fitting curve (Fig. 6):
5.25
  7.32 . (27)
2

- 114 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

TABLE 1 CALIBRATED NUMERICAL VALUES FOR COEFFICIENT =()

2-LAYER LAMINATED GLASS PANELS UNDER IN-PLANE SHEAR

= a/b
1 2 3 4 5
 13.15 9.10 8.42 8.15 7.80

14
 coefficient for in-plane shear
Simply supported 2-layer LG panels
13 Numerical values (Table 1)
Fitting curve (Eq.(27))

12

11


10

7
0 2 4 6 8 10

Fig. 6 Coefficient  for 2-layer laminated glass panels under in-plane shear
Calibrated numerical results (Table 1) and fitting curve (Eq. (27))

Calibrated values for  summarized in Table 1 were obtained in this work assuming for the examined LG panels a fixed
width b and various heights a ≥ b, thus  ≥ 1. However, additional numerical simulations highlighted that the fitting curve of
Eq. (27) well applies, with k given by Eq. (4), also for the analysis of LG elements characterized by an aspect ratio  ≤ 1
(comparisons were performed in the range 0.5 ≤  ≤ 1).
90
TEST [9]
cr [N/mm2] - Equivalent thickness approach

FEM [9]

75

60

45

30

15

0
0 15 30 45 60 75 90
cr [N/mm2] - Experimental and numerical predictions

Fig. 7 Critical shear stress cr of monolithic and laminated glass panels Comparison between analytical
results (equivalent thickness approach) and experimental/numerical predictions [9]

Further comparisons were performed to check the validity of the proposed approach and the accuracy of the presented
correction factors. Comparisons proposed in Fig. 7, for example, refer to experimental tests and numerical predictions
performed by Wellershoff [9] on monolithic and laminated squared glass panels characterized by various dimensions (a=b=
1.2 m, 1.6 m), nominal thicknesses of glass sheets (t1=t2= 3.85 mm, 5.85 mm, 7.7 mm, 9.7 mm, 11.7 mm, 14.7 mm) with tint=
1.52 mm the thickness of PVB-film, and shear stiffness of interlayer (0.4 N/mm2 < Gint < 100 N/mm2). Results are proposed in
terms of critical shear stress cr, therefore analytical data were obtained, for each panel, dividing the critical shear load Vcr( E,lam
)

- 115 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

(Eq. (3), with t eq ,w (Eq. (21)) and   13.15 , Table 1) for the total thickness of glass sheets.

Also in this circumstance, it is possible to notice that analytical predictions are in good agreement with data available in
literature.

E. Load-Transversal Displacement Relationship

It is known that studying in a realistic manner the stability problem of a laminated glass panel under in-plane shear loads,
the estimation of the critical buckling load Vcr( E,lam
)
does not constitute a useful criterion to define its ultimate strength, since the
panel, due to possible post-buckled membrane effects, could be able to sustain greater loads than Vcr( E,lam
)
. In this context, to
appropriately analyze the load-carrying behaviour of the composite plate, it is thus important to describe precisely its
characteristic load V – transversal displacement w relationship. Generally, this aspect is investigated by means of
sophisticated tridimensional numerical models based on the Finite Element Method. Nevertheless, also in this specific
circumstance the proposed equivalent thickness approach could constitute a useful criterion for the simplification of numerical
modeling and simulation phases.
For this purpose, let us consider the squared (height a= 1 m x width b= 1 m) simply supported LG panel (8/1.52/8 mm)
previously investigated. The panel is affected by an initial imperfection proportional to its first modal shape and characterized
by a maximum amplitude w0= a/500 [7, 20]. As known, the thermoplastic materials typically used to bond together the glass
sheets are strongly temperature and load-time dependent, especially PVB-films [20, 31]. Consequently, the effective level of
connection offered by the adopted interlayer should be accurately taken into account.
In Fig. 8, the results of incremental static analyses performed in ABAQUS with the 3D+shell FE model are compared.
These curves refer to the well-known limit conditions for the examined LG panel (monolithic limit – full – and layered limit –
abs) and to a specific condition of temperature and load duration (T=20°C, 3 seconds of applied load).
Assuming for the interlayer the equivalent shear modulus Gint of a PVB-film subjected to the examined temperature and
loading condition ( Gint  8.06 N / mm 2 [20]), the corresponding load V-transversal displacement w relationship can be suitably
obtained, as shown in Fig. 8. Nevertheless, the accurate modeling of a 2-layer LG panel, as well as the execution of static
incremental analyses, could request long processing times. At the same time, the presence in the composite element of
extremely thin and soft layers (e.g. the interlayer) could compromise the accuracy of results, as well as the convergence of
performed simulations [21].
3000

2500
8/1.52/8mm
a= 1m x b= 1m
Temperature: 20°C; w0= a/500
2000 Full
3s, ABAQUS (Shell-teq)
V [kN]

3s, ABAQUS (3D+shell)

1500 Abs

1000

500

0
0 5 10 15 20
w [mm]
Fig. 8 Load V-transversal displacement w relationship for simply supported2-layer LG panels under in-plane shear.
Numerical comparisons (ABAQUS, 3D+shell and Shell-teq FE-models)

Because of these reasons, the proposed equivalent thickness approach could be taken into account in the modeling of a
geometrically simplified, but equivalently accurate, Shell-teq FE-model consisting in monolithic glass S4R shell elements
having a total thickness t eq ,w given by Eq. (21). As shown in Fig. 8, the obtained V-w curve is in good agreement with the
corresponding curve given by the sophisticated 3D+shell FE-model. In addition, the Shell-teq FE-model is quickly
implementable and it well applies, in general, for the analysis of LG panels bonded together with very soft and thin interlayers
or subjected to various temperature and load-time conditions [20, 21].

- 116 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

III. BUCKLING VERIFICATION OF 2-LAYER LAMINATED GLASS PANELS UNDER IN-PLANE SHEAR

A. Traditional Verification Criterion

In accordance with the Limit State design approach, the buckling verification of a flat 2-layer LG panel under in-plane
shear, simply supported along its edges, could be reasonably developed by contemporarily considering two different conditions,
respectively referred to requisites of deformability and durability [18, 19, 20, 21].
Concerning the deformability, the maximum transversal displacement wmax of the panel should be limited as a function of
its aspect ratio, by posing for example the condition:
a,
wmax  (28)
k
with a the height of the panel and k an appropriate coefficient to be defined for the specific limit state considered. Based on
results of a series of performed numerical and analytical simulations, the coefficient k should be assumed equal to k  300 .
Nevertheless, a reasonable check of the maximum deformation for the composite element should be carried out taking into
account also an initial imperfection proportional to the first modal shape of the panel, having a maximum amplitude w0,tot
defined as:
w0,tot  w0,imp  w0,e  w0,b (29)

and representative of:

 initial imperfections, as for example geometrical tolerances in product standards, structural imperfections due to
fabrication, residual stresses ( w0,imp );

 possible load eccentricities ( w0,e );

 possible boundary eccentricities ( w0 ,b ).

As suggested by Mocibob [7], a minimum amplitude w0,min  a 1000 should be taken into account for the initial
imperfection. In this context, as proposed in the previous section, the maximum deformability of a generic LG panel could be
rationally checked by taking into account a geometrically simplified but precise Shell-teq FE-model.
Contemporarily, the design shear load VEd should be compared with the buckling resistant value Vb, Rd of the composite
panel, defined as:

Vcr( E,lam
)

VEd  Vb , Rd  , (30)
 M1
where Vcr( E,lam
)
is the critical buckling load, obtained from Eq. (3) by posing t  teq,w (Eq. (21) and    ( ) , Eq. (27)), and
 M 1 is an appropriate safety coefficient. Based on numerical simulations and experimental tests, Wellershoff and Sedlacek
suggest for the safety factor a value  M 1  1.40 [8].

B. Alternative Verification Criterion

As known, for the buckling verification of traditional structural elements made of steel, concrete or timber, consolidate
verification criteria are available in literature. The Eurocode 3 [15], for example, estimates the buckling resistance of
compressed steel members by taking into account a series of appropriate buckling curves, opportunely calibrated to take into
account the effects in their buckled response of possible initial imperfections of different amplitude, as well as different
residual stresses.
The suggestions of Eurocode 3 cannot be directly applied to the buckling verification of 2-layer LG panels under in-plane
shear. However, since extremely practical, similar buckling curves could be opportunely calibrated and used to predict the
buckling resistance of 2-layer simply supported laminated glass panels subjected to in-plane shear loads and affected by initial
imperfections or boundary/loading eccentricities.
The main advantage of buckling curves proposed by Eurocodes, in fact, consists in their theoretical background. In them,
as originally proposed by Ayrton-Perry [32] for the analysis of geometrically imperfect columns loaded by uniform
compressive loads, the initial imperfections, as well as other effects (residual stresses, possible eccentricities) are efficiently
described through an equivalent initial sine-shape imperfection, that is through a generalized imperfection factor  . In the

- 117 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

specific case of a compressed monolithic column affected by a sine-shape imperfection of maximum amplitude w0, for
example, the generalized non-dimensional imperfection factor proposed by Ayrton and Perry is defined as [32]:
A
  w0 , (31)
W
where A and W respectively individuate the total cross-section area and the elastic resistant modulus of the examined column.
As a result, the governing parameters able to describe the buckling resistance of an imperfect compressed column are:
 the normalized slenderness of the examined column
A Rk ,
 (32)
N cr( E )

with  Rk the characteristic tensile strength of glass and N cr(E ) the Euler’s critical buckling load, and

 a buckling reduction factor

N cr ,
 (33)
A Rk
defined as a function of the axial load N cr associated to the reaching in the mid-span cross section of the tensile strength  Rk .

In this context, the buckling verification of a 2-layer simply supported LG panel subjected to in-plane shear could be still
performed by satisfying the condition given by Eq. (30), in which the design buckling resistance Vb, Rd of the layered panel
could be expressed as:
Vb, Rd   A  Rd , (34)

with  an opportune reduction factor, and:

A  b (t1  t 2 ) , (35)

the glass resistant cross-section area of the composite panel. In Eq. (34), conservatively, the design shear strength  Rd of glass
should be assumed equal to the design tensile strength  Rd , as recommended by Wellershoff and Sedlaceck [8].

In addition, the Eurocode 3 suggests for the reduction factor  the expression:

1 , with   1
 (36)
2
   2

and:
2
Φ  0.5 [1  α imp (λ  α 0 )  λ ] , (37)

A Rk
 the slenderness of the composite panel, (38)
Vcr( E )

 Rk the characteristic shear strength of glass (  Rk   Rk [8]),

α imp and α 0 appropriate imperfection coefficients.

Specifically, the coefficient α 0 individuates the values of slenderness λ associated to a reduction factor equal to   1
(thus   1 for λ   0 ). At the same time, the coefficient α imp is representative of the maximum allowable imperfection for
the panel. In this work, the imperfection factors α imp  0.49 and α 0  0.50 have been appropriately calibrated on the basis of
numerical (ABAQUS) and experimental data available in literature for simply supported monolithic or laminated glass panels
under in-plane shear. Specifically, numerical results have been obtained by performing in ABAQUS (3D+shell FE-model) a
series of static incremental analyses on 2-layer simply supported LG panels under in-plane shear characterized by different
geometrical properties. An initial imperfection proportional to the first modal shape of the examined panels, having maximum

- 118 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

amplitude w0= a/1000, was taken into account [7]. Additional numerical comparisons allowed noticing that a similar limitation
(Eq. (34), with α imp  0.49 and α 0  0.50 ) approximately coincides, for λ  1.10 , with the assumption of k  300 in Eq. (28)
or  M 1  1.40 in Eq. (30). In the same Figure, additional numerical data are proposed for monolithic and laminated simply
supported glass panels [9]. Finally, experimental results obtained by Wellershoff by performing tests performed on glass
panels are taken into account [9].
1.4
M1= 1.40 (Eq.(30))
ABAQUS, w0= a/1000
1.2 TEST [9]
FE-Monolithic [9]
FE-Laminated [9]
1.0

0.8


0.6

0.4

0.2
Euler's Curve
"EC Curve" (imp= 0.49, 0= 0.50)
0.0
0 1 2 _
3 4 5

Fig. 9 Buckling “EC Curve” for the verification of simply supported 2-layer laminated glass panels under in-plane shear
Comparisons with numerical (ABAQUS) and experimental data [9]

As a result, data collected in Fig. 9 to validate the proposed verification approach are in good agreement with the obtained
buckling curve (called “EC Curve”).

IV. 3-LAYER LAMINATED GLASS PANELS UNDER IN-PLANE SHEAR

As known, special applications in modern architectures require the use of 3-layer (or even more) laminated glass elements.
A similar technological solution guarantees an important increasing of resistance to high impact impulsive loads and
significant anti-theft functions. Nevertheless, existing analytical approaches generally apply only to 2-layer LG elements, as
well as to specific loading and boundary scenarios. At the same time, numerical simulations performed on similar structural
elements typically require onerous modeling and computing times.
Based on these assumptions, additional analytical calculations were performed to extend the proposed equivalent thickness
approach to the analysis of 3-layer LG panels under in-plane shear. In this specific context, only simply supported panels were
examined. However, a similar approach could be easily applied to LG elements with various boundary and loading conditions,
as for example recently proposed for LG columns and panels under in-plane compression [20, 21]. As a result, noticeable
simplifications could be provided in the investigation of the buckled response of 2 and 3 (or more) layer laminated glass
elements. At the same time, a generalized unified buckling verification approach could be assumed for them.
For this purpose, let us consider a 3-layer simply supported LG panel obtained by assembling 3 glass sheets (thickness t1
for the external faces and thickness t2 for the middle sheet) and two interlayers of thickness tint, as proposed in Fig. 10.

Fig. 10 Schematic view of 3-layer laminated glass panel (cross section)

- 119 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

Based on simple geometrical considerations, the critical shear load Vcr( E,lam
)
can be still estimated by means of Eq. (12),
assuming for the equivalent thickness the expression [21]:

t eq,w  3 2t13  t 23  12 Γ J s , (39)

where  and  are respectively given by Eq. (15) and Eq. (16). In addition, with reference to Fig. 10, the following parameters
should be taken into account [21]:

J s  2t1 t s2,1 ; (40)

t s ,1  0.5t1  t int  0.5t 2 ; (41)

t s  t1  2tint  t 2 . (42)

Finally, the correction factor  can be estimated by means of Eq. (27), as previously done for the analysis of 2-layer LG
panels.
Also in this circumstance, depending on the shear stiffness of the adopted interlayer, the predicted critical shear load Vcr( E,lam
)

(Eq. (3), with t eq , w given by Eq. (39)) is always comprised between the limit values of buckling load Vcr( E,abs
)
( Gint  0 ) and
Vcr( E, full
)
( Gint   ), given respectively by Eq. (23) and Eq. (24), where:

E (2t13  t 23 )
Dabs  (43)
12 (1  2 )
is the layered flexural stiffness of the 3-layer LG panel and D full (Eq. (25)), with

t full  2t1  2tint  t 2 , (44)

is the monolithic flexural stiffness.

Since no existing analytical formulations can be used to validate the accuracy of the proposed equivalent thickness
approach, also in this specific circumstance parametric buckling analyses were performed in ABAQUS to check the accuracy
of the model.

A. Numerical Validation for 3-Layer Laminated Glass Panels

At first, a further 3D+shell FE-model was developed in ABAQUS to estimate the critical load of 3-layer LG panels. In it,
the middle glass sheet (thickness t2) and the interlayers (thickness tint) were described through 3D-8 node elements, whereas the
external glass sheets were modelled in the form of monolithic S4R shell elements (thickness t1) with offset (toffset= t1/2). Also in
this specific circumstance, a second Shell-teq FE-model consisting in monolithic S4R shell glass elements of thickness teq,w (Eq.
(39)) was taken into account for further comparisons.
Buckling analyses were performed on both the FE-models to check the accuracy of the presented analytical approach.
Several geometrical and mechanical parameters were taken into account in performing simulations, as for example the shear
stiffness of the adopted interlayers ( 10 4 N/mm 2  Gint  10 4 N/mm 2 ), the aspect ratio of the panels ( 1    10 , with
b  1m ), the thicknesses of each layer (6/1.52/6/1.52/6 mm, 6/0.76/6/0.76/6 mm, 6/0.38/6/0.38/6 mm).
The main numerical results are proposed in Figs. 11-13 and compared with analytical predictions (Eq. (3), with teq,w given
by Eq. (39)). As shown, analytical critical loads are in good agreement with numerical results obtained by the sophisticated
3D+shell FE-model. Clearly, the 3D+shell FE-model is the more accurate, but the modeling of the LG panel and the
performance of buckling analyses generally require rather long processing time. Furthermore, as noticed in previous works [19,
20], the 3D+shell FE-model could overestimate the real critical buckling load Vcr( E,lam )
in presence of extremely soft
thermoplastic films (Figs. 11-13, Gint  101 N/mm 2 ), or could present convergence problems in performing simulations.

These aspects should not be ignored, especially in the verification of LG panels assembled with PVB-films or subjected to
elevated temperatures, as well as long-term loads, and particular attention should be dedicated to their modeling and analysis.
In contrast, the Shell-teq FE-model can be quickly implemented and buckling analyses can be performed in a very short time,
thus a similar modeling approach could represent a major simplification in the verification LG panels. Moreover, the Shell-teq
FE-model has no convergence problems associated with the presence of extremely thin layers or very soft films, as would
happen by using for example multilayer shell models [21].

- 120 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

In conclusion, based on the proposed comparisons, the equivalent thickness approach could represent a useful expedient to
be used in the analysis of 2 and 3-layer simply supported LG panels under in-plane shear. The buckling verification of 3-layer
LG panels, also in this specific context, could be rationally performed by contemporarily satisfying the deformability and
resistance conditions expressed by Eq. (28) and Eq. (30). Equivalently, by simply replacing in Eq. (35) the glass cross-section
area A of the examined 3-layer LG panel, the buckling “EC Curve” proposed in Fig. 8 should be taken into account in
performing a suitable verification.
6000
6/1.52/6/1.52/6mm
a= 1m x b= 1m

5000 Full (Eq.(24))

teq,w (Eq.(3), with  Eq.(27))
ABAQUS (3D+shell)
4000 ABAQUS (Shell-teq)
V(E)cr,lam [kN]

Abs (Eq.(23))

3000

2000

1000

0
-2 -1 0 1 2 3 4
10 10 10 10 10 10 10
2
Gint [N/mm ]
Fig. 11 Critical shear load Vcr( E,lam
) for 3-layer simply supported laminated glass panels under in-plane shear
(6/1.52/6/1.52/6 mm, a= 1 m x b= 1 m). Analytical and numerical comparisons (= (), Eq. (27))
5000
6/0.76/6/0.76/6mm
a= 1m x b= 1m

4000 Full (Eq.(24))

teq,w (Eq.(3), with  Eq.(27))
ABAQUS (3D+shell)
ABAQUS (Shell-teq)
V(E)cr,lam [kN]

3000 Abs (Eq.(23))

2000

1000

0
-2 -1 0 1 2 3 4
10 10 10 10 10 10 10
2
Gint [N/mm ]
Fig. 12 Critical shear load Vcr( E,lam
) for 3-layer simply supported laminated glass panels under in-plane shear
(6/0.76/6/0.76/6 mm, a= 1 m x b= 1 m). Analytical and numerical comparisons (= (), Eq. (27))

- 121 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

4000

3500 Full (Eq.(24))

teq,w
(Eq.(3), with  Eq.(27))
3000
ABAQUS (3D+shell)
ABAQUS (Shell-teq)
V(E)cr,lam [kN] 2500 Abs (Eq.(23))

6/0.38/6/0.38/6mm
2000 a= 1m x b= 1m

1500

1000

500

0
-2 -1 0 1 2 3 4
10 10 10 10 10 10 10
2
Gint [N/mm ]
Fig. 13 Critical shear load Vcr( E,lam
) for 3-layer simply supported laminated glass panels under in-plane shear
(6/0.38/6/0.38/6 mm, a= 1 m x b= 1 m). Analytical and numerical comparisons (= (), Eq. (27))

V. CONCLUSIONS
Because of their characteristic high slenderness, LG panels subjected to in-plane shear can be affected by stability problems.
Existing analytical formulations derived from the theory of sandwich panels in different boundary and loading conditions, due
to the hypotheses they have been formulated on, frequently cannot be directly applied to LG panels.
Because of these reasons, an analytical model, based on the concept of equivalent thickness, is proposed for the analysis
and verification of simply supported 2 and 3-layer LG panels. As shown, the proposed approach allows designers to simply
and realistically evaluate their critical buckling load, by taking into account the effective level of connection associated to well-
defined temperature and load-time conditions. As highlighted by the proposed numerical validations, the presented analytical
formulation provides realistic results In addition, numerical computation can be simplified by realizing geometrical simplified
but still accurate monolithic shell models having a glass cross-section of equivalent thickness.
In this way, according to the State Limit approach, a rational buckling verification of laminated glass panels under in-plane
shear could be carried out by contemporary satisfying deformability and resistance criteria. Equivalently, as usually suggested
by the Eurocodes for the buckling verification of conventional structural elements made of steel, concrete or timber, a buckling
“EC curve” opportunely calibrated to numerical and experimental data could be taken into account in performing an extremely
suitable and rational buckling verification.

REFERENCES

[1] J. O’Challaghan, and C. Bostick, “The Apple Glass Cube: Version 2.0,” Proceedings of Challenging Glass 3 – Conference on
Architectural and Structural Applications of Glass, TU Delft, IOS Press, ISBN 978-1-61499-060-4, 2012.
[2] T. Henriksen, “Future Application of Structural Use of Glass,” Proceedings of Challenging Glass 3 – Conference on Architectural and
Structural Applications of Glass, TU Delft, IOS Press, ISBN 978-1-61499-060-4, 2012.
[3] R. Nijsse, “Glass Walls Carrying the Roof and Withstanding the Wind Load on the Façade: Conservatory of the Museum in Dordrecht
and Raaks Glass Cube in Haarlem,” Proceedings of Challenging Glass 3 – Conference on Architectural and Structural Applications of
Glass, TU Delft, IOS Press, ISBN 978-1-61499-060-4, 2012.
[4] J. Belis, “Kipsterkte van monolitische en gelamineerde glazen liggers,” Universiteit Gent, Laboratorium voor Modelonderzoek,
Doctoraatsthesis, ISBN: 90-8578-034-9, 2006 (available: http://lib.ugent.be).
[5] A. Luible, “Stabilität von Tragelementen aus Glas,” Ecole Polytechnique Fédérale de Lausanne, Thése EPFL 3014, 2004 (available:
http://icom.epfl.ch/publications).
[6] O. Englhardt, “Flächentragwerke aus glas – Tragverhalten und Stabilität,” Universität für Bodenkultur Wien, Dissertation zur Erlangugn
des Doktorgrades, 2007.
[7] D. Mocibob, “Glass panel under shear loading – Use of glass envelopes in building stabilization,” Ecole Polytechnique Fédérale de

- 122 -
Journal of Civil Engineering and Science Sept. 2013, Vol. 2 Iss. 3, PP. 108-123

Lausanne, Thése EPFL 4185, 2008 (available: http://icom.epfl.ch./publications).

[8] F. Wellershoff, and G. Sedlacek, “Stabilization of Building Envelopes with the use of the Glazing,” Proceedings of Glass Processing
Days, 2005.
[9] F. Wellershoff, “Nutzung der Verglasung zur Aussteifung von Gebäudehüllen,” RWTH Aachen, Dissertation, 2006.
[10] M. Feldmann, and K. Langosch, “Knickfestigkeit und einheitliche Knickkurven für scheibenförmige Glasstützen mit
Monoglasquerschnitt aus TVG und ESG,” Ernst & Sohn Verlag für Architektur und technische Wissenschaften GMbH & Co. KG,
Berlin – Stahlbau Spezial 2010-Konstruktiver Glasbau, 2010.
[11] S. Aiello, G. Minafò, and N. Scibilia, “Compressive behaviour of laminated structural glass members,” Eng. Struct., vol. 33, pp. 3402-
3408, 2011.
[12] P. Foraboschi, “Buckling of a laminated glass column under test,” The Structural Engineer 87(1), 2009.
[13] J. Blaauwendraad, “Buckling of laminated glass columns,” Heron, 52(1-2).
[14] K. Stamm, and H. Witte, “Sandwichkonstructionen – Berechnung, fertigung, ausführung,” Springer Verlag, Germany, 1974.
[15] UNI EN 1993-1-1: 2005. Eurocode 3 – Design of Steel Structures – Part 1-1: General rules and rules for buildings, May 2005.
[16] UNI EN 1994-1-1: 2005. Eurocode 4 – Design of Composite Steel and Concrete Structures – Part 1-1: General rules and rules for
buildings, May 2005.
[17] UNI EN 1995-1-1: 2005. Eurocode 5 – Design of Timber Structures – Part 1-1: General rules and rules for buildings, May 2005.
[18] C. Amadio, and C. Bedon, “Buckling of laminated glass elements in out-of-plane bending,” Eng. Struct., vol. 32, pp. 3780-3788, 2010.
[19] C. Amadio, and C. Bedon, “Buckling of laminated glass elements in compression,” J. Struct. Eng., vol. 137(8), 2011.
[20] C. Bedon, and C. Amadio, “Buckling of flat laminated glass panels under in-plane compression or shear,” Eng. Struct., vol. 36, pp. 185-
197, 2012.
[21] C. Amadio, and C. Bedon, “Analytical approaches for buckling verification of in-plane loaded laminated glass columns and panels,”
Proceedings of Challenging Glass 3 – Conference on Architectural and Structural Applications of Glass, TU Delft, IOS Press, ISBN
978-1-61499-060-4, 2012.
[22] S. Timoshenko, “Theory of Plates and Shells,” McGraw-Hill Book Company, 1989.
[23] R. Southwell, and S. Skan, “On the stability under shearing forces of a flat elastic strip,” Proceedings of Royal Society of London,
Series A, 1924.
[24] E.W. Kuenzi, W.S. Ericksen, and J.J. Zahn, “Shear stability of flat panels of sandwich construction,” Armed services technical
information agency, Virginia 1975.
[25] E. Wölfel, “Nachgiebiger Verbund eine Näherungslösung und deren Anwendungsmöglichkeiten,” Stahlbau 6/1987, pp.173-180, 1987.
[26] ASTM E1300 – 9a, Standard Practice for Determining Load Resistance of Glass in Buildings, ASTM, 2009.
[27] S.J. Bennison, M.H.X. Quin, and P.S. Davies, “High-Performance Laminated Glass for Structurally Efficient Glazing,” Innovative
Light-weight Structures and Sustainable Façades, Hong Kong, 2008.
[28] I. Calderone, P.S. Davies, S.J. Bennison, H. Xiaokun, and L. Gang, “Effective Laminated Thickness for the Design of Laminated
Glass,” Proceedings of Glass Performance Days, 2009.
[29] ABAQUS® version 6.9, Simulia, Pawtucket, R.I. 02860 USA, 2009.
[30] A. Luible, and M. Crisinel, “Plate Buckling of Glass Panels,” Proceedings of Glass Processing Days, 2005.
[31] Meissner, M., and Sackmann, V., “On the effect of artificial weathering on the shear bond and the tear strength of two different
interlayers of laminated glass,” in Porc. ISAAG, 2006.
[32] W.E. Ayrton, and J. Perry, “On Struts”, The Engineer, pp. 137-153, 1886.

- 123 -