a r t i c l e
i n f o
Article history:
Received 16 March 2014
Accepted 17 May 2014
Available online xxxx
Keywords:
Glass design
Laminated Glass
Minimum thicknesses
Optimal design
Polymeric interlayers
Sacricial ply
a b s t r a c t
This paper focuses on glass used in buildings and presents a criterion for fail-safe optimal design of glass
plates.
First, the paper shows that the glass plate is fail-safe only if the load-bearing system is composed of two
glass layers bonded to one another with an elastomeric interlayer (Laminated Glass), and if the live loads
act upon a sacricial glass ply (tri-layer system).
Then, activity was directed at analyzing the simply-supported fail-safe glass plate loaded out-of-plane,
and carrying out research targeted at reducing the incidence of weight and cost (optimal design). The
results, obtained using an analytical exact model, show that the limit states are always dictated by the
maximum deection and not by the load-carrying capacity. Thus, optimal design requires nding
the thickness of the glass layers and the stiffness of the interlayer that provide the plate with exactly
the minimum allowable stiffness, while the stress verications are fullled automatically.
Finally, for every span and load that is found in building applications of glass, the paper provides the
thicknesses and the materials of the glass layers and interlayers that adjust capacity to match demand.
These results may also replace structural analysis and assessment of Laminated Glass plates.
2014 Elsevier Ltd. All rights reserved.
1. Introduction
To optimize the performance and cost of a structure, the task of
a structural designer is to determine the material(s) and geometry
of component(s) under some functional requirements and constraints [115]. To select the optimal structure among possible
similar structures is called optimal design.
This paper is devoted to the optimal design of plates made of
glass, loaded predominantly out-of-plane, and used as primary or
secondary structures, or as non-bearing elements. Primary structures include oors and roofs [1621]; secondary structures
include overhead glazing supported by primary structures and elements of staircases [16,20]; non-bearing elements include faades
and partitions [22,23].
Not only do glass elements have to withstand the design loads,
but also they have to fulll fail-safe design [16,21,22,2428]. Fail
safe design of glass is based on a fundamental evaluation namely,
the ratio between the load-carrying capacity and ultimate load
demand, and the ratio between the span and maximum deection
do not provide an exhaustive description of safety. That these
ratios are adequate is a necessary but not a sufcient condition.
http://dx.doi.org/10.1016/j.matdes.2014.05.030
0261-3069/ 2014 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
The greatest enhancements in mechanical properties of interlayers have been provided by IPs, which currently hold an appreciable fraction of the world market share [24,26,28]. The key
mechanical property difference between an IP interlayer and a
PVB interlayer is the visco-elastic behavior for high temperatures
and long-term load durations, which provides an IP interlayer with
a signicant stiffness advantage over a PVB interlayer. In particular,
for high temperatures and long loading durations, the shear elasticity modulus of IP may be more than two orders of magnitude
greater than that of PVB [3944].
Since the newer interlayers are signicantly stiffer than the traditional interlayers, these recent developments in technology have
split the thermoplastic family for lamination of glass into two classes namely, the utterly compliant interlayers, such as PVB, and
the relatively stiff interlayers, such as IP [44].
When a glass ply is broken, almost all the glass shards adhere to
the polymeric interlayer rather than scatter, avoiding injury to
people or pets from broken glass. Thus, LG fullls part of the
demand of the second fail-safe condition, whereas LG does not fulll automatically the other part of this condition.
More specically, when the glass is broken, the polymeric interlayer holds in place almost all the pieces of glass, but some pieces
of glass can fall down anyway. Thus, the layer that can give rise to
ying, shattering or falling glass (e.g., the inferior layer of horizontal glazing) has to disintegrate into pieces that are small and not
sharp (i.e., small dice shapes, instead of dangerous shards). However, this last demand of the second fail-safe condition is not satised automatically by LG, but it requires that specic glass types are
used, according to the structural function of the glass. In so doing,
the second fail-safe condition is satised completely.
LG also provides the system with some post-breakage load-carrying capacity. In fact, the common boundary between the glass
ply and interlayer keeps the fracture in the glass ply where it has
started; i.e., the interface prevents the crack from propagating into
the whole system. Accordingly, when a glass layer is broken, the
other glass layer may bear the applied load at least as an individual
element. It follows that a LG system guarantees a residual capacity
at least equal to the strength of the intact glass layer [43,45,46].
Hence, the loaddeection curve of LG descends slowly from
the peak up to the failure of the other glass layer. Thus, LG satises
automatically two of the demands posed by the third fail safe condition namely, the ultimate displacement is fairly large and the
rst branch of the post-peak loaddeection curve is not steep.
Conversely, LG does not satisfy automatically the last demand
posed by the third fail safe condition namely, the collapse of
LG must not convert excessive potential energy into kinetic energy.
In order to avoid that shortcoming, one of the layers has to be
made of a glass type that, in the event of breaking, fractures into
large chunks and slivers. The interlayer can connect these large
pieces of broken glass to each other, which provides the laminated
system with a certain residual capacity. As a result, the system can
carry the dead load even when all glass layers are broken. In so
doing, the last branch of the post-peak loaddeection curve is
not steep and the system dissipates almost all the potential energy
before reaching the collapse. This system cuts down the kinetic
energy released during the breakage, and it meets completely the
third fail-safe condition.
Ultimately, only LG can fulll the second and third fail-safe conditions; however, LG does not fulll automatically all these conditions. To be fail-safe LG also has to fullls the requirements of
Section 2.3.
The rst fail-safe condition can be satised only by using the
sacricial ply design criterion [26,43,4749]. This criterion can
be synthesized as follows: The glass ply that collects the live loads
is allowed to fracture in service (sacricial ply). According to the
sacricial ply design criterion, hence, the layer that collects the live
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
Fig. 1. LG plate loaded on the top face, designed according to the sacricial ply
concept. Tri-laminated system (three glass plies and two interlayers): upper glass
layer, central glass layer, and lower glass layer. The glass ply that collects the live
loads (upper layer) is sacricial.
Fig. 2. LG plate loaded on the top and bottom faces, designed according to the
sacricial ply concept. Quadri-laminated system (four glass plies and three
interlayers). The glass plies that collect the live loads are sacricial.
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
Fig. 4. Stress prole in the load-bearing system (central and upper layers, together
with the connecting interlayer; Fig. 3). The stress prole is anti-symmetric with
respect to the middle plane of the load-bearing system (shown in the gure).
Shadowed areas: tension stresses. The gure shows the symbols used for the
stresses throughout the paper.
Fig. 5. Stress prole in the load-bearing system, in the case of high value of the
interlayer stiffness, k. The central layer is completely in compression and the lower
layer is completely in tension.
Fig. 3. Load-bearing system: bi-layer system composed of the central and lower
layers of the tri-laminated system, together with the connecting interlayer
(shadowed). Prole of the longitudinal displacement, u, due to anti-clockwise
rotation angle of the cross-section, h. The points P, R, S and, respectively, P0 , R0 , S0 are
anti-symmetric with respect to the middle plane of the load-bearing system. Thus,
the u-displacements of P, R, S are equal in magnitude and opposite in direction with
respect to P0 , R0 , S0 .
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
demand, but these glass types never fulll the failure mode
demand.
The central glass layer is subjected to compression stresses and
usually also to tension stresses (Fig. 3). This layer bears the maximum compression stress of the tri-layer system, rgo (Figs. 4 and
5), which occurs at the center of the surface connected to the sacricial ply through the interlayer. Conversely, the tension stress in
the central glass layer is always less than in the lower glass layer.
The maximum tension stress in the central layer, rgi, approaches
rgo when the stiffness of the interlayer approaches the stiffness
of the layers.
Fail-safe design of glass requires that at least one layer of the
laminated system is made of either heat-strengthened or annealed
glass, in order to provide the system with adequate post-breakage
capacity when all glass layers are broken. In fact, a heat-strengthened glass layer (sometimes even an annealed glass layer) can
carry the entire dead load of the plate when the load-bearing system is broken as well as the sacricial glass ply, since the interlayer
connects the broken pieces of heat strengthened glass.
Considering that the lower layer has to be made of toughened
glass and that the central layer has to bear tension stresses less
than the lower layer, the central layer has to be made of either
heat-strengthened glass or annealed glass, according to the maximum tension stress in this layer rgi (Figs. 1, 4 and 5).
The fact that the central lass layer breaks into large chunks and
slivers with razor sharp edges poses no risk of injure to people,
since this layer is inner.
3. Optimal design of fail-safe Laminated Glass plates
The criterion that allows glass to optimize the structural performance, maximize transparency, and minimize the cost (optimal
design criterion) is that the thicknesses of the central and lower
layers and the stiffness of the interlayer produce a behavior that
matches the most severe limit of the plate.
The load-bearing system (Section 2.3.2; Figs. 1 and 2) must have
adequate strength to resist the most severe loads and adequate
stiffness to resist all the behaviors that can compromise the serviceability of the plate. The rst requirement is satised if and only
if the maximum tension stress in each glass layer (i.e., rgo and rgi;
Figs. 4 and 5) does not exceed the tension strength of the glass that
the layer is made of. The second requirement is satised if and only
if the maximum deection of the plate (i.e., wmax) does not exceed
the maximum allowable deection.
Consistently with the optimal design criterion, the glass type
and thickness of the central layer and the lower layer, together
with the stiffness of the interlayer have to produce either a maximum tension stress in one of the two layers that is equal to the
tension strength of that layer or a maximum deection of the plate
that is equal to the maximum allowable deection. One condition
is matched exactly while the other two conditions with a certain
margin.
Tension strength of glass depends on the glass types used for
the layers and also on the duration of load, which inuences glass
strength [23,24,3639,42]. Moreover, this limit has to include a
partial factor that allows for the class of consequences associated
to the intended use of the plate as well as the uncertainties in
the material strength.
The load to be used in the ultimate limit state verications in
accordance with the considered code has to produce stresses that
do not exceed the strength of each type of glass that the system
is composed of. If this limit is respected, the strength of the glass
plate is adequate.
For the dead loads (i.e., long-duration load), the limits that fulll
the main codes on glass [43,45,46,5263] are 5.5 N/mm2 for the
annealed glass, 25.0 N/mm2 for the heat-strengthened glass,
65.0 N/mm2 for the tempered glass, and 80.0 N/mm2 for the chemically-strengthened glass.
For the wind loads (i.e., short-duration load), the limits that fulll the main codes on glass [43,45,46,5263] are 19.0 N/mm2 for
the annealed glass, 40.0 N/mm2 for the heat-strengthened glass,
80.0 N/mm2 for the tempered glass, and 105.0 N/mm2 for the
chemically-strengthened glass.
The maximum deection of the plate depends on the intended
use of the plate and the lifespan of the building. The load to be used
in the serviceability limit state verications in accordance with the
considered code has to produce a ratio between the maximum
deection and span that is less than a maximum value. If this limit
is respected, the stiffness is greater than the minimum required by
the code.
The deections must not damage the construction or cause
alarm to users, and the vibrations must not cause ailments to users
or generate or propagate noises. In order to guarantee these serviceability conditions, two ratios have to be considered. Firstly,
the ratio that results from the deection due to the total load (dead
plus live load) and secondly, the ratio that results from the deection due to the live load only. The rst ratio measures the stiffness
against the damage of the construction, and ensures adequate
durability. The second ratio measures the stiffness against the
vibrations, and prevents the occurrence of deections that could
cause alarm to users.
The maximum allowable deection-to-span ratio prescribed for
glass depends on the intended use of the element [21,22].
If the intended use of the plate is a oor, the maximum allowable deection-to-span ratio recommended by almost all the codes
is 1/200 under the live loads. If the intended use of the plate is a
roof, it is still 1/200 but under the total loads [43,45].
If the intended use of the plate is a faade or a partition, the
maximum allowable deection-to-span ratio recommended by
the most important codes is 1/65 under the live load
[43,45,46,5255].
It is to note that the maximum allowable deection-to-span
ratio used for glass faades and partitions is very high, and that
used for glass oors is double than the limit commonly used for
other materials. The reason is that the elasticity modulus of glass
is low, i.e. approximately 70,000 N/mm2. If the common limits
were used, transparent architecture would be impossible. Hence,
the deection limits were calibrated so as to foster the use of glass
in architecture. Nevertheless, the deection beyond which a glass
element no longer fullls the relevant design criteria is usually
well represented by the values adopted by codes. It being understood that designers should check the limit used case by case.
4. Analytical modeling of the Laminated Glass plate
Until a short time ago, the sandwich plates, including LG plates,
were modeled by using either nite element models or practical
formulas.
The layer-to-interlayer elastic modulus ratio and layer-to-interlayer thickness ratio may be very high. For LG, the former ratio may
reach one million and the latter may be greater than fty. These
values impinge upon all the numerical solutions [23,36,42,47,64
68]. Thus, the numerical results need to be checked and calibrated
against exact results. However, no exact solution existed; therefore, numerical results could be neither checked nor calibrated.
Moreover, no closed-form formula is available with a numerical
model; therefore, numerical models allow optimal design to be
performed only with a trial and error approach.
The practical formulas of LG, which were derived either empirically or from numerical models, replace the laminated system with
an equivalent monolithic system. However, not only are these formulas rough, but above all they provide the maximum stress and
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
deection only, while they do not provide the complete stress eld
in the LG plate. Thus, practical formulas are not viable tools for optimal design of LG plates.
Ultimately, numerical models and practical formulas do not
allow the structural designer to select the optimal LG plate among
the fail-safe plates. In order to determine the thickness and material
of layers and interlayer under any functional requirements and constraints, the design has to use an analytical closed-form (exact)
model.
Now, the LG plate can be modeled analytically, since a closedform exact model was recently obtained [19,6971], which provides the stresses and displacements of the sandwich plate. The
modeling assumptions are that the glass layers are governed by
the KirchhoffLove hypotheses and that the polymeric interlayer
has an elastic behavior.
Hence that model is linear. The deection-to-span limit within
which geometrical non-linearity of a plate is always negligible is
lower than 1/65. However, the simple-support is a restraint that
reduces geometrical non-linearity. As a result, geometrical non-linearity is not substantial in LG [72], and it may be neglected. However, geometric non-linearity increases the safety margin.
The interlayer materials behave in a visco-elastic way, but they
can be modeled in a linear-elastic way by means of the modulus of
elasticity in shear Gt, provided that Gt is related to temperature and
loading duration [18,23,25,36,37,4044,64]. Ultimately the results
of the analytical model do not suffer from any constraining
assumption.
When the structural designer commits to certain materials (i.e.,
glass type and a polymer for the interlayers), that analytical model
[19,6971] may provide a technique to optimally determine the
geometry of the sandwich plate. Likewise, when the geometry is
chosen, that model together with the fail-safe conditions may provide a technique to select the best materials for the layers and
interlayers.
Fixing one in general inuences the optimality of the other, but
concurrently determining the optimal geometry and selecting the
best materials remained an open issue in LG plate design. Therefore, activity was directed at carrying out research targeted at providing a general framework for optimal design of LG plate based on
a criterion suitable for LG plates, and at reducing the incidence of
weight in the design of LG plates.
To this end, the aforesaid model was applied to the laminated
system that fullls the fail safe condition, in the framework of
the optimal design.
Since the upper layer is sacricial, analytical modeling does not
consider the tri-layer system (Fig. 1), but the sandwich system
(Fig. 6). Hence, analytical modeling took into account the LG plate
whose cross-section is formed by two glass layers, each one of thickness h (Figs. 36), called central and lower layer respectively, plus an
interlayer of thickness t. This research considered the rectangular
plate with sides L B (L 6 B), simply-supported at the four edges,
subjected to a uniformly distributed lateral load p (Figs. 1 and 6).
The material properties of the layers are the elastic modulus,
shear (elastic) modulus, and Poissons ratio of glass, Eg, Gg, and mg,
respectively. For structural glass, commonly: Eg = 70,000 N/mm2,
Gg = 28,689 N/mm2, and mg = 0.22, independently of the glass type
[43,45,46,5263].
The interlayer material properties are the shear modulus Gt and
the elasticity modulus Et.
The behavior of the interlayer can be represented by its stiffness, k (Fig. 6):
2 Gt
t
Fig. 6. Diagram of the load-bearing system of the laminated glass plate, with the
geometric and mechanical symbols. The diagram shows the glass types of the load
bearing system. The system together with the sacricial glass ply (plies) satises
the fail-safe conditions.
The dependence of the shear elasticity modulus of the interlayer Gt on temperature and loading duration is dictated by the
interlayer material, which is a design option. Moreover, t is a
design parameter as well. Hence, k is a design parameter whose
values, within certain limits, can be chosen so as to obtain the optimal design.
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
More specically, the maximum values of rgo, rgi, and wmax used in
the limit analysis, and the parameters used in the models were the
following:
The maximum allowable stresses (strength) considered in ultimate limit state verications of LG plates used for oors or covering roofs were 5.5, 25.0, 65.0, and 80.0 N/mm2 for annealed,
heat-strengthened, tempered, and chemically-strengthened
glass, respectively.
These stress values include all the strength-reduction factors of
glass, barring the factor that takes into account the type of edge
work, since the maximum tension stress is away from the edges.
In particular, these values consider that the load duration is equal
to the lifespan of the element. Moreover, these values include the
material partial factors for the ultimate limit state prescribed by
the codes.
The maximum allowable stresses (strength) of each glass type
considered in ultimate limit state verication of LG plates used
for faades and partitions were 19.0, 40.0, 80.0, and 105.0 N/
mm2 for annealed, heat-strengthened, tempered, and chemically-strengthened glass, respectively.
Also these stress values (and for the same reason) include all the
strength-reduction factors of glass, barring the factor that takes
into account the type of edge work. These values consider that
the stresses are produced by the wind action or impacts, whose
load duration was assumed to be 5 s. These stress values also
include the material partial factors for the ultimate limit state prescribed by the codes.
The maximum allowable displacement considered in serviceability limit state verication of LG plates used for oors and
covering roofs was 1/200 of the minor span of the plate.
The maximum allowable displacement considered in serviceability limit state verication of LG plates used for faades
and partitions was 1/65 of the minor span.
For every glass type, Eg = 70,000 N/mm2 and mg = 0.22.
The analysis considered two interlayer materials namely, PVB
and IP (i.e., the most common utterly compliant and relatively
stiff types, respectively) and two types of glass members
namely, primary or secondary bearing members, as oors or
covering roofs, and non-bearing members, as faades and
partitions.
The conditions considered for oors and roofs (bearing members) were a loading duration of more than 50 years and a temperature of at least 50 C. The interlayer shear elasticity modulus used
for this combined condition was Gt = 0.052 N/mm2 in the case of
PVB material and Gt = 1.50 N/mm2 in the case of IP materials.
The conditions considered for faades and partitions (non-bearing members) were a loading duration of 30 s (the wind gust can
act on a plate already loaded by the mean wind) and a temperature
of 50 C. The interlayer shear elasticity modulus used for this combined condition was Gt = 0.300 N/mm2 for PVB material and
Gt = 105.00 N/mm2 for IP materials.
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
Table 1
Laminated Glass plate congurations that match the most severe limit state of
bearing glass members. L = span of the plate; B = width of the plate; p = load
uniformly distributed on the plate (on the sacricial glass ply, which is the upper
layer); h = thickness of each glass layer of the bi-layer system that carries the stresses
(i.e., of the central layer and the lower layer); k = stiffness of the interlayer between
the central and lower layer, dened by Eq. (1); rgo = maximum tension stress in the
lower glass layer; rgi = maximum tension stress in the central glass layer; wmax =
maximum deection of the plate.
Data
Design
Behavior
L
mm
B
mm
p
kN/m2
h
mm
k
N/mm3
rgo
rgi
N/mm2
N/mm2
L/wmax
mm/mm
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3500
3500
3500
3500
3500
1500
1500
1500
1500
2000
2000
2000
2000
2500
2500
2500
2500
2000
2000
2000
2000
2500
2500
2500
2500
3000
3000
3000
3000
2500
2500
2500
2500
3000
3000
3000
3000
3500
3500
3500
3500
4000
4000
4000
4000
3000
3000
3000
3000
3500
3500
3500
3500
4000
4000
4000
4000
4500
4500
4500
4500
5000
5000
5000
5000
3500
3500
3500
3500
4000
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
3
4
5
6
4
5
6
8
4
5
8
8
5
5
8
10
5
6
8
10
5
8
10
10
5
8
8
10
6
8
10
12
6
8
10
12
8
8
12
12
6
8
10
12
8
10
12
15
8
10
12
15
8
10
15
15
8
10
15
15
8
10
12
15
8
3.6512
3.4571
5.5368
4.5128
1.4723
2.2415
4.7642
1.8212
2.9663
4.9663
1.0563
3.2632
0.6855
7.4374
1.1913
0.7281
1.6479
3.7532
3.0116
1.8108
3.8705
0.7448
1.0369
3.4539
4.0558
0.7600
3.4964
3.3766
1.7650
1.5023
2.1618
1.7737
4.9142
3.1047
4.9194
3.3684
0.5216
7.8622
1.3590
6.7982
4.3127
2.7732
4.3107
2.9583
0.6256
0.8733
1.6805
1.0136
0.9958
1.4670
3.2853
1.6689
1.5933
2.4221
0.7350
2.5796
2.553
4.3150
0.9655
3.9043
0.8780
1.2522
2.7321
1.4377
1.8864
7.32
9.01
11.62
12.92
7.58
9.70
12.81
13.43
8.59
11.08
12.32
15.06
6.56
9.76
10.52
11.81
9.99
10.02
12.41
13.45
8.73
9.56
12.15
15.04
7.68
8.34
12.47
13.19
7.77
9.50
12.14
13.50
8.78
10.61
13.60
14.98
7.77
11.61
12.81
16.30
7.78
9.39
12.05
13.27
7.16
9.13
15.04
13.01
15.07
15.01
15.03
14.14
8.43
10.79
12.24
15.20
8.96
11.50
12.77
16.16
6.99
8.94
11.80
12.65
7.91
1.88
3.28
3.68
5.43
2.83
3.34
3.20
6.26
1.56
1.68
5.07
4.20
4.61
1.40
7.30
9.92
7.68
2.22
3.91
6.12
1.16
4.20
5.20
3.89
1.31
5.93
1.76
5.06
2.25
3.66
4.39
5.92
0.84
2.06
2.28
3.73
3.09
8.25
4.25
2.07
1.10
2.71
3.04
5.00
3.63
4.46
11.92
7.05
7.84
9.99
13.20
5.14
1.83
2.10
5.17
3.75
1.21
1.25
4.37
2.69
3.43
4.13
3.84
6.80
1.88
200.4
200.0
199.7
199.8
199.8
200.7
200.5
201.3
200.6
200.3
199.6
199.4
200.2
200.2
200.4
200.6
200.2
199.7
199.3
199.4
199.9
200.4
200.0
200.3
200.1
199.8
199.2
200.0
200.2
199.4
199.0
199.6
199.9
200.2
199.8
200.2
200.1
199.4
199.3
200.4
200.5
200.6
199.7
199.5
200.6
199.7
199.5
199.7
199.1
199.9
199.6
200.1
200.2
199.7
199.5
200.1
200.0
199.7
200.3
199.2
200.1
199.3
199.4
199.6
199.1
Table 1 (continued)
Data
Design
Behavior
L
mm
B
mm
p
kN/m2
h
mm
k
N/mm3
rgo
rgi
N/mm2
N/mm2
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
5000
5000
5000
5000
6000
6000
6000
6000
4000
4000
4000
4500
4500
4500
4500
5000
5000
5000
5000
5500
5500
5500
5500
4000
4000
4000
4000
4500
4500
4500
4500
5000
5000
5000
5000
6000
6000
6000
6000
5000
5000
5000
5000
6000
6000
6000
6000
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
10
15
15
10
12
15
19
10
12
15
19
10
12
15
19
8
12
15
15
10
12
15
19
10
12
15
19
10
15
19
19
10
15
19
19
12
19
22
25
3.1368
0.8134
3.0526
0.4792
0.8984
1.2644
0.6858
0.6706
1.3285
1.9211
0.9641
0.8951
1.9685
2.9474
1.2772
4.5208
0.7796
1.0842
6.3684
0.6966
1.4169
2.0573
0.9958
1.1326
2.9274
4.8447
1.5450
4.2421
0.6770
0.8453
4.1179
5.1868
0.6508
0.8158
4.2921
5.3184
0.3693
0.9239
1.1618
10.12
11.30
14.23
7.40
9.63
12.24
13.33
7.84
10.28
13.01
14.07
8.25
10.87
13.89
14.74
7.92
8.68
11.05
14.23
7.30
9.60
12.25
13.09
7.92
10.51
13.44
14.13
9.06
9.97
12.57
16.00
8.00
8.71
10.98
14.11
8.07
8.50
11.29
13.00
2.00
6.13
3.97
3.48
3.81
4.62
6.92
2.80
2.89
3.42
5.68
2.28
2.13
2.45
4.73
0.88
4.90
5.98
2.41
3.24
3.23
3.80
6.69
2.27
1.79
1.88
5.00
0.68
3.56
4.53
2.36
0.64
4.86
6.18
2.86
0.55
5.67
5.41
5.93
L/wmax
mm/mm
200.4
200.3
199.8
199.9
199.3
199.5
199.4
200.1
199.6
199.7
199.6
200.2
200.3
199.9
200.2
199.5
200.0
199.6
199.6
200.2
199.8
199.5
199.5
200.1
200.0
200.0
199.1
200.4
200.2
200.1
200.4
200.3
200.0
200.1
199.7
199.4
199.7
199.4
200.0
For each plate, these tables provide the minimum thickness of the
glass layers and the minimum stiffness of the interlayer that allow
the plate to satisfy the serviceability and ultimate limit states.
Hence, each LG plate with the relevant geometry and stiffness in
Tables 1 and 2 satises both the deection verication and the
strength verication; one verication exactly, the other verication
with a certain margin.
Moreover, Tables 1 and 2 provide, for each case, the maximum
stress in the lower glass layer rgo, the maximum stress in the central glass layer rgi (Fig. 4), and the ratio between the span L and
maximum deection wmax. The shear stress in the interlayer can
be calculated from those values, using Eg and Gt.
Tables 1 and 2 allow the design to select the optimal combinations of thicknesses of the glass layers and spans of the plate, for a
given design load.
6. Discussion
The analysis considered all the spans and loads of LG plates
used in civil engineering, for bearing (Table 1) and non-bearing
(Table 2) members. Moreover, the analysis considered only real
glass plies and interlayers, which restricted the domain in which
the minimum values had to be found. In fact, the minimum thickness of each glass ply and the minimum stiffness of the interlayer
were found in a discrete domain, with a substantial increment
from one value to the next. The discrete domain implied that the
minimum thickness found among the glass plies available on the
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
Design
Behavior
L
mm
B
mm
p
kN/m2
h
mm
k
N/mm3
rgo
rgi
N/mm2
N/mm2
L/wmax
mm/mm
1500
1500
1500
1500
1500
1500
1500
1500
1500
2000
2000
2000
2000
2000
2000
2000
2000
2000
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
1500
1500
1500
2000
2000
2000
2500
2500
2500
2000
2000
2000
2500
2500
2500
3000
3000
3000
2500
2500
2500
3000
3000
3000
3500
3500
3500
4000
4000
4000
3000
3000
3000
3500
3500
3500
4000
4000
4000
4500
4500
4500
5000
5000
5000
3500
3500
3500
4000
4000
4000
4500
4500
4500
5000
5000
5000
5500
5500
5500
4000
4000
4000
4500
4500
4500
5000
5000
5000
6000
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
2
3
3
3
3
4
3
4
4
3
4
5
3
4
5
4
4
6
4
4
6
4
5
6
4
5
8
5
6
8
4
5
8
5
6
8
5
6
8
5
8
8
6
8
10
5
6
8
5
8
10
6
8
10
6
8
10
6
8
10
6
8
10
6
8
10
6
8
10
8
1.9394
1.0369
21.6921
0.5022
3.6685
3.2616
0.8289
0.8289
15.5658
0.8951
0.9185
1.3544
3.1372
2.6109
4.6053
0.5886
24.2728
1.7303
0.5577
100.5886
1.6133
1.2140
1.9737
7.2368
3.3603
9.2625
0.7288
0.6105
1.1776
1.0511
2.9040
8.1671
0.6679
0.7450
1.6535
1.3158
1.3978
5.0498
2.7714
3.1422
0.4189
8.9258
0.5900
0.5299
0.7532
1.1374
3.8640
2.3196
6.0815
0.4621
0.6457
0.8252
0.7065
1.0112
1.3960
1.0652
1.6467
2.8421
1.5942
2.5901
0.6788
0.6087
0.8933
1.5269
1.0970
1.8027
10.0763
2.6609
4.9477
0.3777
15.16
17.60
27.10
15.11
22.07
27.18
16.70
20.93
31.5
14.44
17.93
23.01
17.63
21.60
27.97
16.21
24.95
27.21
14.15
22.05
23.94
16.47
21.27
28.52
18.64
24.17
25.70
16.91
22.40
27.41
16.52
21.40
22.64
15.74
20.92
25.54
17.49
23.57
28.37
19.10
20.53
31.00
17.46
21.54
27.40
15.66
21.03
25.31
18.04
18.94
24.15
16.80
20.55
26.26
18.14
22.02
28.11
19.34
23.40
29.98
15.18
18.53
23.58
17.05
20.66
26.31
18.89
22.63
29.10
16.76
4.10
10.49
0.56
7.53
3.30
6.26
5.26
7.49
0.70
6.88
10.13
12.20
1.82
4.17
4.22
6.01
0.48
7.52
8.19
1.74
10.12
4.38
4.85
2.32
1.42
0.58
10.81
5.26
5.51
8.43
1.86
0.70
14.45
5.69
5.15
9.07
3.30
1.56
4.92
1.35
8.43
1.36
4.88
7.22
8.78
4.55
2.39
6.62
0.36
9.90
12.15
4.60
7.41
8.93
2.90
5.43
6.16
1.35
3.86
4.17
6.13
9.55
11.32
3.00
6.13
6.61
0.08
2.76
2.55
5.78
64.7
65.0
64.6
65.5
65.2
64.7
65.3
64.7
65.0
65.0
65.1
64.8
64.9
64.8
64.5
65.5
64.6
65.4
65.3
64.43
64.6
64.7
64.5
64.6
64.7
64.5
64.3
65.4
64.3
64.4
64.7
64.9
64.7
65.4
65.1
64.5
65.1
64.3
64.7
64.7
64.8
64.4
64.7
64.4
64.5
64.7
64.7
64.9
64.4
65.1
64.9
65.0
64.7
64.5
64.8
64.8
65.1
64.8
64.5
64.5
64.5
64.8
65.3
64.3
64.1
65.1
64.0
64.8
64.3
64.8
Table 2 (continued)
Data
Design
L
mm
B
mm
p
kN/m2
h
mm
4000
4000
5000
5000
5000
6000
6000
6000
6000
6000
5000
5000
5000
6000
6000
6000
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
10
12
8
10
12
8
12
15
Behavior
k
N/mm3
0.5315
1.0799
0.3714
0.5216
1.0631
16.8526
0.4580
0.6395
rgo
rgi
N/mm2
N/mm2
21.40
28.04
14.59
18.66
24.58
17.37
18.73
23.89
7.06
7.14
7.70
9.39
9.28
0.40
9.25
11.24
L/wmax
mm/mm
64.6
64.6
65.3
64.9
64.6
64.9
65.0
64.7
market did not depend on the choice of the interlayer; and vice
versa, the minimum stiffness of the interlayer obtainable using
products available on the market did not depend on the choice of
the glass plies. This result facilitated the analysis and simplies
this discussion.
When the maximum deection is equal to 1/200 of the span, the
maximum tension stress in the lower glass layer rgo is always less
than one-quarter of the allowable tension stress for tempered glass
and it is always less than two-thirds of the allowable tension stress
for heat-strengthened glass (Table 1). Simultaneously, the maximum tension stress in the central glass layer rgi is always less than
30% of the allowable tension stress for heat-strengthened glass and
it exceeds the allowable tension stress for annealed glass only in
approximately one-fth of the cases (Table 1).
When the maximum deection is equal to 1/65 of the span, the
maximum tension stress in the lower glass layer rgo is always less
than 40% of the allowable tension stress for tempered glass, it is
always less than approximately three-quarters of the allowable
tension stress for heat-strengthened glass, and in approximately
two fth of the cases it is less than the maximum allowable tension
strength of annealed glass (Table 2). Simultaneously, the maximum
tension stress in the central glass layer rgi is always less than
approximately one-third of the allowable tension stress for heatstrengthened glass and it is always less than the allowable tension
stress for annealed glass (Table 2). Note that the allowable tension
stress (strength) of non-bearing members is greater than that of
bearing members, since in the former case the duration of the
design load is much shorter than in the latter case.
Hence, the design and structural assessment of the simply-supported LG plates whose lower layer is made of toughened glass and
whose central layer is made of heat-strengthened glass are always
dictated by the allowable deection. The design and structural
assessment can disregard the stress eld and strength of materials;
the limit state analysis can ignore resistance verication, while it
has to consider only the stiffness verication. Glass tension stresses and strength, as well as resistance verication have to be considered only when the central layer is made of annealed glass, but
only in few cases (some spans and loads), which are shown in
Tables 1 and 2.
Another result was that the choice of the interlayer does not
inuence the optimal selection for the glass layers, and vice versa.
Accordingly, the optimal design of the glass layers and of the interlayer can be separated from one another, which simplies the optimal design process.
7. Simultaneous selection of geometry and materials
This section is devoted to the optimal design of the simply-supported LG plate.
Tables 1 and 2 provide the design solutions with the minimum
weight of glass and the minimum stiffness of the interlayer to
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
10
Table 3
Materials and thicknesses to obtain the optimal congurations found in Table 1 for bearing Laminated Glass plates. This table uses the same nomenclature as Table 1. Moreover,
type refers to the glass and material to the interlayer, where A = annealed glass, H = heat-strengthened glass, T = toughened glass (tempered or chemically strengthened),
IP = Ionoplast Polymers.
Central glass layer
L
mm
B
mm
p
kN/m2
Type
Label
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
1500
1500
1500
1500
2000
2000
2000
2000
2500
2500
2500
2500
2000
2000
2000
2000
2500
2500
2500
2500
3000
3000
3000
3000
2500
2500
2500
2500
3000
3000
3000
3000
3500
3500
3500
3500
4000
4000
4000
4000
3000
3000
3000
3000
3500
3500
3500
3500
4000
4000
4000
4000
4500
4500
4500
4500
5000
5000
5000
5000
3500
3500
3500
3500
4000
4000
4000
4000
4500
4500
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
A
A
A
A
A
A
A
H
A
A
A
A
A
A
H
H
H
A
A
H
A
A
A
A
A
H
A
A
A
A
A
H
A
A
A
A
A
H
A
A
A
A
A
A
A
A
H
H
H
H
H
A
A
A
A
A
A
A
A
A
A
A
A
H
A
A
H
A
A
A
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Interlayer
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
h
mm
Material
Label
t
mm
Type
Label
3
4
5
6
4
5
6
8
4
5
8
8
5
5
8
10
5
6
8
10
5
8
10
10
5
8
8
10
6
8
10
12
6
8
10
12
8
8
12
12
6
8
10
12
8
10
12
15
8
10
12
15
8
10
15
15
8
10
15
15
8
10
12
15
8
10
15
15
10
12
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
0.76
0.76
0.38
0.38
1.52
1.14
0.38
1.52
0.76
0.38
1.52
0.76
1.52
0.38
1.52
1.52
1.52
0.76
0.76
1.52
0.76
1.52
1.52
0.76
0.38
1.52
0.76
0.76
1.52
1.52
1.14
1.52
0.38
0.76
0.38
0.76
1.52
0.38
1.52
0.38
0.38
0.76
0.38
0.76
1.52
1.52
1.52
1.52
1.52
1.52
0.76
1.52
1.52
1.14
1.52
1.14
1.14
0.38
1.52
0.76
1.52
1.52
0.76
1.52
1.52
0.76
1.52
0.76
1.52
1.52
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
h
mm
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
H
T
T
T
T
T
T
T
T
T
T
T
3
4
5
6
4
5
6
8
4
5
8
8
5
5
8
10
5
6
8
10
5
8
10
10
5
8
8
10
6
8
10
12
6
8
10
12
8
8
12
12
6
8
10
12
8
10
12
15
8
10
12
15
8
10
15
15
8
10
15
15
8
10
12
15
8
10
15
15
10
12
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
11
B
mm
p
kN/m2
Type
Label
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
5000
5000
5000
5000
6000
6000
6000
6000
4500
4500
5000
5000
5000
5000
5500
5500
5500
5500
4000
4000
4000
4000
4500
4500
4500
4500
5000
5000
5000
5000
6000
6000
6000
6000
5000
5000
5000
5000
6000
6000
6000
6000
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
0.50
1.00
2.00
3.00
A
H
A
A
A
H
A
A
A
A
A
A
H
A
A
A
A
H
A
A
A
A
A
A
A
A
A
A
H
A
A
H
A
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Interlayer
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
h
mm
Material
Label
t
mm
Type
Label
15
19
10
12
15
19
10
12
15
19
8
12
15
15
10
12
15
19
10
12
15
19
10
15
19
19
10
15
19
19
12
19
22
25
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
0.76
1.52
0.38
1.52
1.52
0.38
1.52
1.52
1.14
1.52
1.52
0.76
0.38
1.52
0.38
1.52
1.52
0.38
0.38
1.52
1.52
0.38
0.38
1.52
1.52
1.52
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
h
mm
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
15
19
10
12
15
19
10
12
15
19
8
12
15
15
10
12
15
19
10
12
15
19
10
15
19
19
10
15
19
19
12
19
22
25
Only the Ionoplast Polymers allow the interlayer to reach the stiffness shown in Table 1. In many cases, it is possible to choose between two glass types for the central layer. It
is always possible to choose between two glass types for the lower layer. The table presents all the glass types that can be used for each layer to satisfy the limit states of the
plate; however, these glass types do not all necessarily satisfy the fail-safe conditions.
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
12
Table 4
Materials and thicknesses to obtain the optimal congurations found in Table 2 for non-bearing Laminated Glass plates. This table uses the same nomenclature as Tables 1 and 3;
moreover, PVB = Poly Vinyl Butyral.
Central glass layer
L
mm
B
mm
p
kN/m2
Type
Label
1500
1500
1500
1500
1500
1500
1500
1500
1500
2000
2000
2000
2000
2000
2000
2000
2000
2000
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
2500
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
3500
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
4000
1500
1500
1500
2000
2000
2000
2500
2500
2500
2000
2000
2000
2500
2500
2500
3000
3000
3000
2500
2500
2500
3000
3000
3000
3500
3500
3500
4000
4000
4000
3000
3000
3000
3500
3500
3500
4000
4000
4000
4500
4500
4500
5000
5000
5000
3500
3500
3500
4000
4000
4000
4500
4500
4500
5000
5000
5000
5500
5500
5500
4000
4000
4000
4500
4500
4500
5000
5000
5000
6000
6000
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
2.00
0.50
1.00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Interlayer
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
h
mm
Material
Label
2
3
3
3
3
4
3
4
4
3
4
5
3
4
5
4
4
6
4
4
6
4
5
6
4
5
8
5
6
8
4
5
8
5
6
8
5
6
8
5
8
8
6
8
10
5
6
8
5
8
10
6
8
10
6
8
10
6
8
10
6
8
10
6
8
10
6
8
10
8
10
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
IP
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
PVB
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52/0.76
1.52/0.38
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
1.52
Type
Label
0.38
1.14
0.38
0.38
0.38
0.38
0.38
0.76
0.76
0.38
0.76
0.76
0.38
0.38
0.76
0.76
0.38
0.38
1.14
0.76
1.14
0.76
0.38
1.14
0.76
0.38
0.38
0.38
0.38
0.76
0.76
0.38
0.38
0.38
1.52
1.14
A
A
H
A
H
H
A
H
H
A
A
H
A
H
H
A
H
H
A
H
H
A
H
H
A
H
H
A
H
H
A
H
H
A
H
H
A
H
H
H
H
H
A
H
H
A
H
H
A
A
H
A
H
H
A
H
H
H
H
H
A
A
H
A
H
H
A
H
H
A
H
h
mm
H
H
H
H
H
H
H
T
H
H
T
H
H
H
H
H
H
H
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
8
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
10
T
T
T
T
A
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
2
3
3
3
3
4
3
4
4
3
4
5
3
4
5
4
4
6
4
4
6
4
5
6
4
5
5
6
8
4
5
8
5
6
8
5
6
8
5
8
8
6
8
5
6
8
5
8
10
6
8
10
6
8
10
6
8
10
6
8
10
6
8
10
6
8
10
8
10
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
13
B
mm
p
kN/m2
Type
Label
4000
5000
5000
5000
6000
6000
6000
6000
5000
5000
5000
6000
6000
6000
2.00
0.50
1.00
2.00
0.50
1.00
2.00
A
A
A
A
A
A
A
H
H
H
H
H
H
H
Interlayer
T
T
T
T
T
T
T
h
mm
Material
Label
12
8
10
12
8
12
15
IP
IP
IP
IP
IP
IP
IP
1.52
1.52
1.52
1.52
1.52
1.52
1.52
Type
Label
0.38
1.52
1.14
0.38
1.14
0.76
H
A
A
H
A
A
H
h
mm
H
H
H
H
T
T
T
T
T
T
T
12
8
10
12
8
12
15
In many cases, it is possible to choose between two interlayer materials. It is always possible to choose between two glass types for the central layer. It is always possible to
choose between two glass types, and in many cases between three glass types, for the lower layer. The table presents all the glass types that can be used for each layer to
satisfy the limit states of the plate; however, these glass types do not all necessarily satisfy the fail-safe conditions.
ultimate limit state, the central and lower glass layers that
compose the load-bearing system have to produce stresses less
than the strength of each glass layer. Conversely, the combination
of actions for the serviceability limit state is carried by the whole
tri-layer system (load-bearing system plus sacricial ply); i.e., the
deection verication includes the upper layer. In brief, the upper
layer is sacricial for the stress verication while can be considered
for the stiffness verication.
Since the combination of actions for the ultimate limit state is at
least 1.4 times greater than the combination of actions for the serviceability limit state, this design option signicantly reduces the
thickness of the LG plate. Design and assessment can still be
accomplished using Tables 1 and 2, since the behavior of the LG
plate is linear.
7.4. Updating the fail-safe design concept: Second proposal
The results show that heat-strengthened glass would provide
the lower glass layer with enough strength to resist the loads
(Tables 1 and 2). However, this layer is external; in the event that
this glass layer breaks, therefore, no sharp shards can fall, according to the second fail-safe condition. Thus, the lower layer has to be
made of toughened glass, which shatters into small, blunt pieces.
Conversely, heat-strengthened glass breaks into large, razor-sharp
slivers.
On one hand, to replace the toughened glass with heatstrengthened glass would not provide any economic benet, since
the latter costs no less than the former. On the other hand, however, this result suggests another updating of the fail-safe concept.
There are glass plates whose collapse does not involve building
occupants and pedestrians. These glass plates can be fail-safe also
without the toughened glass at the lower layer. For these plates,
the design can use a bi-layer system instead of a tri-layer system:
An upper layer made of toughened glass and a lower layer made of
heat-strengthened (or annealed) glass, together with the polymeric
interlayer between these glass layers. The layers should have the
same thickness. Since falling glass fragments, although neither
small nor blunt, cannot create potential injury or risk of death to
users and pedestrians, this LG plate fullls the second fail-safe condition. Moreover, the upper layer is devoted to collecting the live
loads (toughened glass), while the lower layer is devoted to providing the plate with adequate post-breakage capacity (heat-strengthened or annealed glass). Thus, this system is completely fail-safe.
There are glass plates whose live loads may act on both the
faces. These glass plates can be fail-safe only if the laminated system uses two sacricial ply, one at the top and one at the bottom.
For these plates, the design can use the above-described bi-layer
system plus a bottom sacricial ply made of toughened glass (and
the interlayer), instead of a quadri-layer system. Since this
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
14
bottom layer shatters into small, blunt pieces, this LG plate fullls
the second fail-safe condition. Thus, this system is always failsafe, while the bi-layer system is fail-safe only if collapse does
not endanger users and pedestrians.
Since the sacricial ply added at the bottom of the bi-layer system has no structural role, the structural behavior of this system is
identical to that of the bi-layer system it derives from. Hence, only
the bi-layer system is considered in this subsection, which nevertheless includes both the design solutions.
The upper layer of the bi-layer system is not sacricial for the
serviceability verications. Thus, the deection verication due
to the combination of actions for the serviceability limit state is
performed using the bi-layer system. Hence, wmax is calculated
by considering the composite behavior of the LG plate.
Conversely, the upper layer of the bi-layer system may be considered sacricial for the ultimate verications. If it is sacricial,
the strength verication due to the combination of actions for
the ultimate limit state is performed using the lower layer only.
Hence, the maximum tension stress in the system is calculated
considering the monolithic behavior of the lower plate.
However, it is not mandatory to consider the upper layer sacricial. In many cases, the depth of the neutral axis is greater than
the depth of the load-induced cracks (Fig. 7). E.g. the stress proles
produced by rgo and rgi in Tables 1 and 2 show that, at the midspan, the compression zone is often greater than t/2 (while the
length of a load-induced crack is substantially less than t/2). The
neutral axis positions provided by the analytical model showed
that, in many cases, this condition remains true for the entire LG
plate. This condition is directly connected to the stiffness of the
interlayer, as dened by Eq. (1).
In those cases, the load-induced cracks are subjected to compression stresses and not to tension stresses (Fig. 7). Therefore,
the magnitude of the tension stresses is amplied by a stress intensity factor that is dictated, not by the load-induced cracks, but by
the cracks due to the aws, whose size is shorter.
When that condition is guaranteed, it is no longer mandatory
to consider the upper layer of the bi-layer system sacricial for
the ultimate verications; conversely, the strength verication
can be performed using the LG system. In this case, thus, rgi
and rgo are calculated considering the composite behavior of
the plate. However, the composite behavior can be considered
only for the load combinations that imply a depth of the neutral
axis greater than the depth of the load-induced cracks (Fig. 7). For
the other load combinations, the maximum stresses have to be
calculated considering the lower layer only.
According to Tables 1 and 2, the maximum tension stress in the
lower glass layer, rgi, is much less than the tension strength of the
heat-strengthened glass. More specically, under the loads that
produce a deection of 1/200 of the span, the maximum stress in
the bi-layer system is always less than two-thirds of the allowable
tension stress for heat-strengthened glass. Moreover, under the
loads that produce a deection of 1/65 of the span, the maximum
stress in the bi-layer system is always less than three-quarters of
the allowable tension stress for heat-strengthened glass.
Those differences prompt the use of the above-described bilayer system in lieu of the tri-layer system of Section 2.3, where
it is possible. In fact, this design solution saves weight with
respect to the tri-layer system and reduces the number of laminations from three to two layers. Moreover, if the compression
zone of the upper layer is deeper than the load-induced cracks
(Fig. 7), the amount of glass weight saved by the bi-layer system
is substantial, which drastically increases transparency and
reduces the costs. Also in this case, design and assessment can
still be accomplished using Tables 1 and 2, since the behavior
of the LG plate is linear.
8. Conclusions
The paper focuses on the Laminated Glass (LG) plate and presents criteria and a method for concurrent design and material
selection, by using analytical exact modeling. The paper provides
information that allows the structural designer not only to dimension the LG plate, but also to choose the best geometry-material
combination for a LG plate used as bearing member (oor, covering
roof, stair) or non-bearing member (faade, partition).
The consequences classes that those glass members are categorized in require that they are fail-safe. To this end, resort shall be
made to the fail-safe conditions presented in Section 2, which
allow the design to obtain adequate failure modes of the various
components and the whole laminated system, necessary to ensure
suitable robustness and damage tolerance, and to avoid both brittle
collapses and that dangerous shards can fall down. According to
these conditions, the plate has to be composed of at least three layers and two interlayers (tri-layer system). The bi-layer system can
be accepted only if the design satises in another way those failsafe conditions that are not met.
This paper uses a recently published analytical exact model of
the sandwich plate, which was derived in closed-form from the
KirchhoffLove assumptions for the layers and considering an elastic interlayer. The model provides analytically the stress eld in the
layers and interlayer, and the displacements.
When a structural designer commits to certain materials (types
of glass, interlayer material), that model can be used to determine
the best geometry of the LG plate. Likewise, when architectural
design has dened the geometry, that model can be used to select
the best materials. However, xing the geometry and then choosing the material or xing the material and then determining the
geometry are not truly optimal, since xing one in general inuences the optimality of the other. Concurrently determining the
optimal geometry and selecting the best materials remained an
open issue in the design of LG plates; this paper has lled this
gap for the simply-supported glass plate.
The rst result is the hierarchy of resistance. A LG plate that
matches the serviceability limit state always satises the ultimate
limit state (furthermore, with a great margin). This result holds
true for faades and partitions, although the conditions in which
these structures become unt for the intended purpose is when
the maximum displacement exceeds 1/65 of the span, which is a
great deection. This result holds true even if the deection verication considers the serviceability load while the strength verication considers the ultimate load.
Moreover, this result also holds true for LG plates whose lower
layer is made of heat-strengthened glass instead of toughened
glass; however, this solution can be used only if the design satises
in another way the fail-safe condition that a heat-strengthened
outer layer does not meet.
Ultimately, the design and assessment of the LG plate are dictated by the stiffness and not by the load-carrying capacity.
The second result achieved by this research, which follows logically from the rst result, is that only the choice of the glass type
used for the central layer is dictated by glass strength, while the
choices of the glass types used for the upper and lower layers are
dictated by the fail-safe conditions only.
The glass ply that collects the loads (upper layer) has to be
made of toughened glass and has to be sacricial, since an adequately small probability of failure cannot be guaranteed for a
glass ply exposed to live loads. If both the external faces are
exposed to live loads, including the environmental actions (windborne debris), fail-safe design calls for another sacricial ply (four
glass layers). The lower layer of the tri-layer system is external;
therefore, in the event of breaking, it has to shatter into small,
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
15
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030
16
[49] Saxe T, Behr R, Minor J, Kremer P, Dharani L. Effects of missile size and glass
type on impact resistance of sacricial ply laminated glass. J Arch Eng ASCE
2002;8(1):2439.
[50] Gy R. Ion exchange for glass strengthening. Mater Sci Eng: B
2008;149(2):15965.
[51] Morse S, Norville H. Design methodology for determining the load resistance
of heat-treated window glass. J Arch Eng ASCE 2012;18(1):4251.
[52] ASTM E1300 Standard Practice for Determining Load Resistance of Glass in
Buildings.
[53] ASTM C1422-99, Standard specication for chemically strengthened at glass,
American Society for Testing Material (ASTM); 1999.
[54] ASTM E1300-09a, Standard practice for determining load resistance of glass in
buildings, American Society for Testing Material (ASTM); 2009.
[55] ASTM C158-02, Standard test methods for strength of glass by exure
(Determination of Modulus of Rupture); 2012.
[56] EN 572-1, Glass in building Basic soda lime silicate glass products Part 1:
Denitions and general physical and mechanical properties; 2012.
[57] EN 572-2, Glass in building basic soda lime silicate glass product 2012 Part
2: Float glass; 2012.
[58] EN 1863-1, Glass in building heat strengthened soda lime silicate glass Part
1: Denition and description; 2011.
[59] EN 1863-2, Glass in building heat strengthened soda lime silicate glass Part
2: Evaluation of conformity/product standard; 2004.
[60] EN 12150-1, Glass in building thermally toughened soda lime silicate safety
glass Part 1: Denition and description; 2000.
[61] EN 12150-2, Glass in building THERMALLY toughened soda lime silicate
safety glass Part 2: Evaluation of conformity/Product standard; 2004.
[62] EN 12337-1, Glass in building chemically strengthened soda lime silicate
glass Part 1: Denition and description; 2000.
[63] EN 12337-2, Glass in building chemically strengthened soda lime silicate
glass evaluation of conformity/Product standard; 2004.
[64] Kerr A, Smith MJ, Cowling MJ, Hodgkies T. The biofouling resistant properties
of six transparent polymers with and without pre-treatment by two
antimicrobial solutions. Mater Des 2001;22(5):38392.
[65] Awad ZK, Aravinthan T, Zhuge Y, Manalo A. Geometry and restraint effects on
the bending behaviour of the glass bre reinforced polymer sandwich slabs
under point load. Mater Des 2013;45:12534.
[66] Lee MG, Yoon JW, Han SM, Suh YS, Kang KJ. Bending response
of sandwich panels with discontinuous wire-woven metal cores. Mater Des
2014;55:70717.
[67] Mostafa A, Shankar K, Morozov EV. Inuence of shear keys orientation on the
shear performance of composite sandwich panel with PVC foam core:
numerical study. Mater Des 2013;51:100817.
[68] Zhen W, Wanji C. A higher-order displacement model for stress concentration
problems
in
general lamination congurations.
Mater
Des
2009;30(5):145867.
[69] Foraboschi P. Layered plate with discontinuous connection: exact
mathematical model. Compos B Eng 2013;47:36578.
[70] Foraboschi P. Three-layered sandwich plate: exact mathematical model.
Compos B Eng 2013;45(1):160112.
[71] Foraboschi P. Three-layered plate: elasticity solution. Compos B Eng
2014;60(April):76476.
[72] Bedon C, Amadio C. Buckling analysis of simply supported at glass panels
subjected to combined in-plane uniaxial compressive and edgewise shear
loads. Eng Struct 2014;59:12740.
Please cite this article in press as: Foraboschi P. Optimal design of glass plates loaded transversally. J Mater Design (2014), http://dx.doi.org/10.1016/
j.matdes.2014.05.030