-
Available online at www.sciencedirect.com
aCIENCE@DIRECT Construction and Building
ELSEVIER Construction and Building Materials 20 (2006) 761 - 768
MATERIALS www.elsevier.comlIoca te/con b ui Idma t
Structural performance of laminated and unlaminated tempered
glass under monotonic transverse loading
Amir Fam a,., Sami Rizkalla b a Queen's University. Kingston.
Ont .. Canada K7L 3N6
I> North Carolina SW(e University. Centennial Campus.
Raleigh. NC 27695-7533. USA Received II August 2004; received in
revised form 3 November 2004; accepted 31 January 2005
Available online 7 April 2005
Abstract
A total of thirty six bending tests have been conducted on 1220
x 460 mm sheets of glass, 9.5-, 12.7- and t5.9-mm thick, using
slow-rate monotonic loading. Twenty four specimens were laminated
on one side using either one or two 0.36-mm thick polyester
transparent laminates. The study showed that lamination has
significantly changed the failure mode of glass from a catastrophic
failure, where fragments of glass shatter in different directions,
to one which is still brittle yet safer, as the fractured glass
remains fully intact. The average gains in flexural strength,
stiffness and strain energy, as a result of lamination, were 20%,
10% and 34%, respectively, while the maximum gains in flexural
strength, stiffness and strain energy were 36%, 33% and 52%,
respectively. Because of the scatter of data, no specific
correlation between the gains and reinforcement ratio (expressed as
the ratio of laminate-to-glass thickness) could be established. The
load-deflection behaviour of both laminated and unlaminated glass
was linear up to failure. No rupture or delamination of the
laminates were observed. 2005 Elsevier Ltd. All rights
reserved.
Keywords: Glass; Retrofit; Laminated; Shatter; Monotonic
loading; Pressure; Plexure
L Introduction
Glass is increasingly being used in the construction industry.
Tempered glass, in particular, is being used in roofing
applications. Glass, which is quite a brittle material, is
generally vulnerable to failure due to a num-ber of reasons,
including excessive wind loading in hur-ricanes, accumulation of
snow in overhead roofing applications or due to terrorism and
vandalism acts. An additional problem with broken tempered glass in
overhead applications is its tendency not to break into small
parts. It may rather fall in large clumps, which could lead to
serious human injuries and possible life threats [1].
Corresponding author. Tel.: + I 6135336352; fax: + I 613 533
2128. E-mail address:[email protected] (A. Fam).
0950-06 18/$ see front matter 2005 Elsevier LId. All rights
reserved . doi: 10.1016Jj.conbuildmat.2005.01 .05 I
Tempered glass, sometimes referred to as toughened glass, is
produced by heating ordinary annealed glass to just below its
softening point, to about 650C, and then cooling it rapidly with
blasts of air. This causes the surface of the glass to cool more
rapidly than the in-ner core, which in turn causes the outer zones
of the glass to be under compressive stresses, while the inner core
is under tensile stresses. These stresses are in a state of
equilibrium and are generally not less than 70 MPa [2], This
'prestressing' effect results in increasing the bending strength of
glass by four to five times, compared to ordinary glass. It also
changes the failure mode of the glass, from shattering into few
large and sharp pieces; to small pieces (diameter less than 10 mm),
without sharp edges [3]. However, clusters of the small broken
pieces are often lumped, and when falling from a height, could
induce severe injuries as indicated earlier.
-
762 A. Fam. S. Rizkalla I Construction and Building Materials 20
(2006) 761- 768
The term " laminated glass" or "sandwich glass" of-ten refer to
two or more glass plies bonded together with an elastomeric
interlayer such as polyvinyl butyral (PYB) to improve the
post-breakage characteristics of the glass. This type of glass is
usually prefabricated in this form before installation and commonly
used in automotive vehicle windshields. In this paper, however, the
term "laminated glass" is used within a different con-text to
indicate a regular single shect of tempered glass, which is
retrofitted, either under service conditions or before
installation, using a special polymeric transparent lamina attached
to the cxternal surface of the glass to enhance its performance and
failure mode. This simple technique is quite useful and economical,
compared to sandwiched glass. While the structural performance and
failure modes of standard tempered glass and sand-wiched glass have
been studied experimentally and numerically under transverse
loading [4- 6], the behav-iour and failure mode of externally
laminated tempered glass have not been studied. In this paper,
tempered sheets of glass of different thicknesses have been
exter-nally laminated using transparent polyester laminate and
tested in flexure. The study is focused on examining the effects of
lamination on failure mode, flexural strength, stiffness, and
strain energy, as compared to unlaminated glass.
2. Experimental program
In this section, the transparent polymeric laminate and the
installation procedure are described. The test specimens, along
with the test setup and procedure are also described.
2.1. Laminate material and installation
The laminate used in this study consists of three lay-ers of
polyester films, bonded together using acrylic
Table 1 Summary of test matrix and results
Specimen Number of Glass Thickness Lamination J.D. specimens
product (mm)
Al A2 A3 A4 A5 BI B2 B3 B4 B5 B6 B7
3 each
3 each
A 12.7 12.7 12.7 15.9 15.9
B 9.5 9.5 9.5
12.7 12.7 12.7 15.9
No I-Laminated 2-Laminated No I-Laminated
No I-Laminated 2-Laminated No I-Laminated 2-Laminated No
Fig. 1. Typical polyester O.36-mm laminate, showing removal of
protective layer.
pressure-sensitive adhesive. The cold-lamination process under
high pressure provides high shock absorption per-formance and
superior optics at the same time. The lam-inate has a total
thickness, including adhesive, of 0.36 mm, a tensile strength of
193 MPa and a Young's modulus of 3.8 GPa, determined in accordance
with ASTM (D 882-75, 1004-76 and D 1938-67) [7,8]. The laminate has
a visible light transmittance capacity of 92%, a total solar energy
rejection of 17% and an ultra-violet light transmittance of 0-5%.
The side of the lam-inate, which is bonded to glass, has a layer of
acrylic pressure-sensitive adhesive, coated with a thin protective
film. This film can easily be peeled off, prior to installa-tion,
in a similar fashion to wall paper, as shown in Fig. I. For
retrofit application, the surface of ordinary glass is cleaned and
dried, followed by installation of the lam-inate (or multiple
laminates) under high pressure. For glass replacement applications
or new glass installations, the laminate is pre-installed on
typical standard size an-nealed or tempered glass, using the same
process, and shipped to the site. The peel strength of the laminate
is 3.86 MPa.
2.2. Description of test specimens and parameters
The experimental program included a total of 36 tests conducted
on both unlaminated and laminated
Reinforcement Ratio C% age) Average test results at maximum load
Load Deflection Stiffness Energy (kN) (mm) (kN/m) (kNm)
0 9.25 30.9 300 0.143 2.83 9.53 31.4 303 0.150 5.67 9.05 31.0
292 0.140 0 14.00 24.3 576 0.170 2.26 17.37 29.5 588 0.257
0 3.80 37.5 101 0.072 3.79 4.90 44.0 III 0. 109 7.58 4.59 46.0
100 0.106 0 6.86 30.5 225 0.105 2.83 8.95 34.6 293 0. 155 5.67 9.36
31.1 300 0.146 0 14.49 22.8 637 0.167
-
A. Ftt/H, S. Ri=kttlla I Construction and Building MateriaJ~' 20
(2006) 761- 768 763
.:;;:;::=3t.~~ Steel rollers
330
Potentia meter
-1< Elevation
330 -I
Fig. 2. Test setup for monotonic loading.
tempered glass specimens. Two different glass products, provided
by two suppliers from the United States and Canada were used in
this study and referred to, herein, as Glass A (provided by
Virginia Glass) and Glass B (provided by Laurier Glass) . All
specimens consisted of 1220 x 460 mm sheets of glass with polished
edges. Three different thicknesses were investigated in this study,
namely 9.5, 12.7, and 15.9 mm. Table I provides summary of test
specimens. Specimens AI - AS are of Glass A, whereas BI- B7 are of
Glass B. Three identical specimens were tested for each parameter
to provide reliable average values. Specimens AI , A4, BI, B4 and
B7 were unlaminated, and were used as control speci-mens. Specimens
A2, AS, B2 and B5 were laminated with a single laminate applied to
one face . Specimens A3, B3 and B6 were laminated using two
laminates (O.72-mm thick), both applied to the same face. The
laminates were applied to the entire surface area of glass. This
scheme has resulted in reinforcement ratios, defined as the ratio
of thickness of laminate to glass, ranging from 2.26% to 7.58%.
The behaviour of specimens (A I, B4) and (A4, B7) was used to
compare the two types of glass (A and B),
20 A4
16
Z 12 e. "0 ro 0 8
...J
4
10 20 30
15.9mm AS UIlLam
40 50 10 Deflection (mm)
(a) Glass A.
20 30
for the same thickness. The behaviour of specimens (AI - A3),
(A4, AS), (BI- B3), and (B4-B6) was com-pared to examine the effect
of lamination, including the number of laminates, on the behaviour.
Finally, specimens (AI, A4), and (BI, B4, B7) were com-pared to
examine the effect of thickness of glass on the behaviour.
2.3. Test setup and procedure
The experimental program described in this paper was intended to
simulate lateral pressure applied gradu-ally to glass at a slow
rate, over a period of time. As such, all specimens were tested to
failure under mono-tonic loading, using a four-point bending
configuration as shown in Fig. 2. An MTS hydraulic actuator was
used to apply the load using stroke control at a rate of 0.5- 1.0
mmlmin. This setup is a modified version of the one used to
determine the strength of glass in flexure (ASTM C 158-95) [9],
mainly to accommodate full scale specimens in this case. The span
between supports was 990 mm, while the distance between the loads
was 330 mm. Steel rollers of 50-mm diameter were used at
20 15.9mm B7 15.9mm
ILam UIILAm 16
Z12 e. "0 ro 0 8
...J
4
40 10 20 30 40 50 Deflection (mm)
(b) Glass B. Fig. 3. Load-deftection behaviour of 15.9mm
glass.
-
764 A. Fam, S. Rizkafla I Construction and Bllilding Materials
20 (2006) 761- 768
20 A1 12.7mm A2 12.7mm A3 J2.7mm
16 Un-LAm I-lAm 2-um
Z 12 C " '" .3 8
4
10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 Deflection (mm)
(a) Glass A .
20 64 12.7 mm 65 12.7mm 6 6 12.7 mm
Ull-Lam I-Lam 2-Lam 16
Z 12 C " '" 0 8
...J
4
0 0 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50
Deflection (mm) (b) Glass B.
Fig. 4. Load-defleclion behaviour of 12.7-mm glass.
20 61 9.Smm 62 9.Smtn 6 3 9.5mm
UIl-lAm I-Lam 2-Lam 16
Z 12 C " '" 0 8
...J
4
0 0 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50
Deflection (mm) Fig. S. Load-deflection behaviour of9. 5-mm
Glass B.
both loading and support points. Rubber pads, 12 mm thick, were
placed between the steel rollers and surfaces of glass in order to
avoid stress concentrations. Deflec-tions were measured using two
electric potentiometers
at mid span, at both sides, and also at the center of each
support to account for the settlement resulting from the rubber
pads. This configuration allows for measuring the net
deflection.
-
A. Fam, S. Rizkalla I Construction and Building Materials 20
(2006) 761- 768 765
24
22
20
16 ( 1 layer) (2~) (ll1yw) (2I1)'1f)
16
Z 14 ~
" ..,
I (a) Maximum load capacity I ~ 10 0 -'
8
6
4
2
0 60
54
48 I (b) Maximum deflection I E
42
.s 36 0 30
.Q 1:1
~ 24 = ID Cl 18
12
6
0 800
no
640
:f 560 I (e) Stiffness I Z C 480 0 0 400 ID
"' ij5 320
240
160
80
0.3
0.25 I (d) Energy I f 0.2 Z
~ 0.15 '" e>
::1 ID 0 W 0.1
0.05 :' .,1 :,
Fig. 6. Comparison of unlaminated and lami nated specimens o f
Glass A and B.
3. Test results and failure modes
The load--
-
766 A. F(lIIl, S. Ri:kalla I Comtrue/iull and Building Materials
20 (2006 ) 76/- 768
3.1. Effect of type of glass 3.3. Effect of lamination
Specimens A I and B4 (of Glass A and B) were both 12.7-mm thick
and un laminated. The ultimate load and stiffness of Al were 35%
and 33';', higher than B4, respec-tively. Specimens A4 and B7 were
both 15.9-mm thick and unlaminated. The ultimate load and stiffness
of A4 were about 3% and 9% lower than B7, respectively. Therefore,
the strength and stiffness ratios of types A and B glass do not
appear to be consistent for all thick-nesses. As will also be
confirmed later, specimen B7 seems to have higher strength and
stiffness than expected.
3.2. Effect of thickness of glass
Specimens Al and A4 of Glass A were 12.7- and 15.9-mm thick,
respectively. The ratios of strength and stiff-ness of A4-Al were
1.51 and 1.92, respectively, which are In good agreement with
thickness ratios [(15.9:12.7)' = 1.57] for strength and
[(15.9:12.7)3 = 1.96] for stiffness, using basic mechanics
principals. Specimens BI , B4, and B7 of Glass B were 9.5-,
12.7-and 15.9-mm thick, respectively. The ratios of strength and
stiffness of B7 to B4 were 2.11 and 2.82, respec-tively, which are
not in good agreement with thickness ratios [(15.9:12.7)' = 1.57]
for strength and [(15.9:12.7)' = 1.96] for stiffness. The ratios of
strength and stiffness of B4 to BI were 1.81 and 2.23,
respec-tively, which are in good agreement with thickness ratios
[(12 .7:9.5)' = 1.79] for strength and [(12.7:9.5)3 = 2.39] for
stiffness. These results confirm that an inconsistency, perhaps
related to quality control and tempering pro-cess, appears to be
associated with specimens B7 (15.9-mm Glass B) and results in
higher strength and stiffness than expected.
60 Load
50 0 Deflection
6 Stiffness .. 40 '"
Energy
'" i" "
30 E II .. co 20 '"
0
.... 0 10
0 6 ~ -10
o 2 3
0
6
Fig. 6 shows a comparison of all test specimens, both laminated
and unlaminatcd, in terms of maximum load , deflection , stiffness
and elastic strain energy. The stiffness is defined as the slope of
the load-deflection curve in this case, whereas strain energy is
defined as the area under the load- deflection curve. Fig. 7 shows
a relation between the reinforcement ratio, represented as the
ratio of thickness of laminate to that of glass, and the percentage
increase in ultimate load , deflection, stiffness and the strain
energy. While Figs. 6 and 7 show that gains in strength, stiffness,
and strain energy, up to 36%, 33%, and 52%, respectively, were
observed as a result of the lamination, Fig. 7 clearly shows a wide
scatter of data, which makes it very difficult to establish a
correlation between the reinforcement ratio and the gain in
strength, stiffness, or strain energy. This could be attributed to
the very brittle nature of glass and its sensitivity to any slight
difference in quality control during the tempering process. Based
on test re-sults, however, the average gains in strength, stiffness
and strain energy were 20%, 10% and 34%, respec-tively. It is also
clear from Fig. 6 that using two lami-nates does not provide
additional gains, compared to one laminate. This is possibly
attributed to the lack of composite action (horizontal shear
transfer provided by adhesive) between the layers at high loads.
Also, the structural contribution of lamination is low because of
its low stiffness. The ultimate strain of glass, based on the
measured loads, is less than 0.0019, which corre-sponds to very low
stress in the lamina, due to its low modulus. Perhaps the most
pronounced advantage of lamination is its effect on failure mode as
will be dis-cussed next.
6 ~
>l !! u
4 5 6 7 8 Reinforcement ratio (% age)
Fig. 7. Effect of laminate reinforcement ralio on ultimate load.
deflection, stiffness and strain energy.
-
A. Fam, S. Rizkalla I Construction alld Building Materials 20
(2006) 761- 768 767
3.4. Failure modes
A distinct difference in failure mode between the lam-inated and
un laminated glass was observed for all range of thicknesses and
for both Glass A and Glass B. Fig. 8(a) shows an un laminated
IS.9-mm Glass A, specimen A4, with high deflection, moments before
failure . Fig. 8(b) shows the same specimen A4 and also specimen B7
immediately after failure. The unlaminated glass shatters
violently, once the modulus of rupture of glass is attained. Fig.
8(b) clearly reflects the very brittle and catastrophic nature of
the failure. Fragments of glass were scattered and have travelled
more than 6 m away
from the specimen. All unlaminated specimens failed in this
manner. It is envisioned that serious injuries and panic would have
definitely resulted, had a similar sce-nario been encountered in a
real structure. Careful examination of the failure of both un
laminated Glass A and B shows that, while both were extremely
brittle, Glass B shatters into smaller fragments compared to Glass
A as shown in Fig. 8(b). This could be attributed to slight
differences in the tempering processes of both types.
Fig. 8(c) shows the laminated glass after failure . Although the
glass itself was completely fractured in every direction,
throughout its entire surface, all pieces
(a) Glass sped mens with large deflection just before
failure.
(b) Unlaminated glass just after failure.
(c) Laminated glass just after failure. Fig. 8. Deflection and
failure modes of unlaminated and laminated glass.
-
768 A. Fum. S. Ri:kuffa I Con;yirtlctioll and Bllilding Mmeria/s
20 (2006 ) 761 -768
were contained together as one unit due to the presence of the
laminate. All laminated specimens failed in this manner. Fig. 8(c)
also shows that the specimen main-tains the same deflected shape
after fai lure. No signs of failure of the laminates were observed.
It appears that, once the bottom fibre of glass reaches the
modu-lus of rupture and cracks, the bonded lamina fails to sustain
the tensi1e stresses and maintain the applied load (unlike other
composite systems, such as rein-forced concrete for cxample). This
is mainly due to the very low modulus of the lamina (3.8 GPa,
com-pared to 66 GPa of glass), which leads to excessive
deformations without rupture of the lamina. This fail-ure mode was
distinctly different from that of unlami-nated glass and was
certainly quite safer and less catastrophic. A failed laminated
specimen was easily removed from the test setup as one unit.
Similarly, in an actual structure, failed laminated panels would be
easily removed and replaced.
4. Summary and conclusions
This study included a total of 36 tests conducted on both
unlaminated and laminated clear float tempered glass sheets of
thrcc different thicknesses (9.5, 12.7, and 15.9 mm). Monotonic
loading was applied at a slow rate to simulate a rather gradual and
sustained pressure over a short period of time. The lamination used
in-volved either one or two 0.36-mm thick polyester lami-nates
attached to one side of the glass. Based on this study, the
following conclusions are drawn:
I. The most distinct advantage of lamination is that it
significantly changes the failure mode from a brittle and
catastrophic failure, where small fragments of glass shattcr in
different directions, potentially caus-ing serious injuries, to one
which is still biittle yet safer, as the fractured glass remain
intact and can eas-ily be removed and replaced.
2. The average gains in flexural strength, stiffness and strain
energy as a result of lamination were 20%, 10% and 34%,
respectively. Although gains as high as 36%, 33% and 52% in
strength, stiffness and strain energy were observed, the wide
scatter of data made it difficult to establish a specific
correlation between the reinforcement ratio (thickness of
laminates-
to-thickness of glass) and the gains. The scatter of data could
be attributed to the very brittle nature of glass and its
sensitivity to slight variations in temper-ing process, especially
for different thicknesses, where the cooling rate across the
thickness affects the level of residual stresses.
3. Adding a second laminate may have an insignificant effect on
strength, stiffness, strain energy, and failure mode.
4. The load-deflection behaviour of both laminated and
unlaminated glass is generally linear up to failure.
5. No rupture or delamination of the laminates were
observed.
6. The lamination process is quite suitable for both ret-rofit
of glass in existing structures as well as for new structures,
where pre-laminated glass can be installed in the field.
Acknowledgements
The authors acknowledge Clear Defense Inc., for supplying the
test specimens, Mrs. Jerry Atkinson, Ro-berto Nunez and Jeremy
Bloom for their assistance dur-ing the experimental work and the
Constructed Facilities Laboratory of NC State University.
References
(IJ Wheeler I. Glass failure: the unacceptable risk. Environ
Healt h J 2002: 196-9.
[21 Shepard CL, Cannon BD. Khalecl MA. Measurement of in ternal
stress in glass articles. J Am Ceram Soc 2003;86(8): 1353- 9.
[3J 851 , BS 6206. Impact performance requirements for flat
safety glass and safety plastics for use in buildings: 198 1.
[4] Pantel ides CP, Sallee GP, Minor JE. Edge strength of wind
ow glass by mechanical test. J Eng Mech 1994;120(5):1076-90.
[5] Carre H. Daudcville L. Loadbearing capacity of tempered
structuml glass. J Eng Mech 1999; 125(8):914--21.
[6] Norville HS, King KW. Swofford Jl. Behavior and strength of
laminated glass. J Eng Mech 1998: 124( I ):46-53.
(7J ASTM D882. Standard test method for tensile properties of
thin plastic sheeting; 2002.
[8] ASTM D1938. Standard test method for tcarpropagation resis
tance (trouser tear) of plastic fi lm and thin sheeting by a
singlc+tear method; 1994.
[9] ASTM CI58-95. Sta nd ard test methods for strength of glass
by flexure (determination of modulus of ru pture); 1995.
1111111111111111111111111111111111111111111111 I