Equibiaxial Flexure Strength of Glass - Influence of Glass Plate Size and Equibiaxial Ring Ratio

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

International Journal of Applied Glass Science, 1–9 (2014)

DOI:10.1111/ijag.12094

Equibiaxial Flexure Strength of Glass: Influence of Glass


Plate Size and Equibiaxial Ring Ratio
Jeffrey J. Swab,†,‡, Parimal J. Patel, Xuan Tran,1 Luke Gilde,2 Ernest Luoto,3
Mariel H. Gaviola,1 Ashley Gott,4 Blake Paulson,5 and Steven Kilczewski†,6
US Army Research Laboratory, Weapons and Materials Research Directorate, Aberdeen Proving
Ground, Maryland

Glass plays an important role in many engineering applications including transparent armor. Determining the actual
strength of a glass is problematic because typical glass strength numbers are presented in the literature as ranges. Minor
changes in surface finish, composition, test method employed, and processing all influence the mechanical strength of glass. A
standardized method needs to be utilized to ensure that strength data can be compared universally. The use of a standard
methodology is vital when comparing glass compositions or multiple sample lots of the same composition in order to attri-
bute any real strength variations to the glass itself. While the procedures established in ASTM C1499 provide a good frame-
work for glass strength measurement, flexibility in the experimental parameters of the standard allow for substantial strength
variations when testing glass materials. The ratio between the load and support ring diameters, as well as the specimen dimen-
sions, is shown to lead to variations in both the measured strength, as well as the number of valid/invalid tests based on
where fracture initiates. The goal of this study is to quantify these effects in order to establish optimized testing conditions
for determining the strength of float glass.

Background optical windows, filters and mirrors, and as a compo-


nent in bulletproof glass systems. This glass has a wide
Float glass is a commercial product used in a vari- range of uses because the float process produces large,
ety of structural applications including architectural and flat pieces of glass cheaply, and in large quantities.


Members, The American Ceramic Society.

jeffrey.j.swab.civ@mail.mil
© 2014 The American Ceramic Society and Wiley Periodicals, Inc
1
Work performed as part of the Senior Capstone Project for the Science and Mathematics Academy at Aberdeen High School, Aberdeen, MD.
2
Work performed through support by an appointment to the Research Participation Program at the US Army Research Laboratory (USARL) administered by the Oak Ridge Institute for Science
and Education through an interagency agreement between USARL and the US Department of Energy.
3
Work performed while an undergraduate student at Cecil County Community College and with support by an appointment to the Research Participation Program at the US Army Research Labo-
ratory (USARL) administered by the Oak Ridge Institute for Science and Education through an interagency agreement between USARL and the US Department of Energy.
4
Work performed while an undergraduate student at the University of Maryland and with support by an appointment to the Research Participation Program at the US Army Research Laboratory
(USARL) administered by the Oak Ridge Institute for Science and Education through an interagency agreement between USARL and the US Department of Energy.
5
Work performed while an undergraduate student at Harford Community College and with support by an appointment to the Research Participation Program at the US Army Research Laboratory
(USARL) administered by the Oak Ridge Institute for Science and Education through an interagency agreement between USARL and the US Department of Energy.
6
Work performed while on contract with Bowhead Science and Technology, LLC, Belcamp, MD.
2 International Journal of Applied Glass Science—Swab, et al. 2014

These pieces can quickly and easily be cut to the neces- fractures, the ratio of the load ring-to-support ring diam-
sary shape and size for the respective application. Float eter, and the testing of either circular or square plates.8
glass can be strengthened through thermal tempering or The foundation for this standard was a paper by Salem
chemical processes1–3 which in turn expands the range and Powers9 that delved into many of the issues and gen-
of applications. erated the equations for Weibull effective area that
As float glass is flat and can be easily cut, it would enables the comparison of strength data generated using
appear that the equibiaxial flexure strength test method, specimens of different sizes. A review of the available
where a thin plate is placed between two concentric international standards found a German standard (DIN
rings of different diameters and subjected to an increas- 52-292 part 1, “Testing of Glass and Glass Ceramics –
ing load until fracture occurs, would be a convenient Determination of Bending Strength – Coaxial Double
means of determining its strength or the effect of a Ring Bending Test on Flat Specimens with Small Test
strengthening process. This does not appear to be the Area”) was available in the early 1980’s for testing glass
case. An examination of the standards for glass available plates under equibiaxial loading conditions. This has
through the American Society for Testing and Materials since been replaced by European Standard EN 1288
(ASTM) reveals only one standard for strength testing, which is virtually identical to DIN 52-292.10
and this standard uses flat beam specimens subjected to While developed respectively to determine the
four-point flexural loading.4 Additionally, most of the equibiaxial flexure strength of different classes of mate-
literature shows that rods and fibers are the most com- rials, a review of ASTM C1499 and EN 1288 reveals
monly used test specimens for determining the strength many similarities. Both standards prescribe the use of
of glass or the effect of different treatments investigated essentially the same formula for calculating the
to increase glass strength.5–7 strength of a circular plate, Table I, with the only dif-
The equibiaxial flexure method is finding favor in ference that C1499 uses ring and specimen diameters
the advanced ceramics community over four-point flex- while EN 1288 uses radii. These equations can be
ure because of the equibiaxial stress state that is created, modified to determine the strength of a square or
the large amount of material that is examined and rectangular specimen by calculating the diameter or
because the influence of the specimen edges, specifically radius of a circle that expresses the characteristic size
the quality of the edges, on the strength is minimized. A of the plate and then inserting this value into the
standard test methodology (ASTM C1499) has been strength equation.
developed and approved by ASTM specifically for Both standards also provide similar requirements
advanced ceramics based on studies that addressed issues and information related to the loading machine, allow-
related to mid-point deflection of the specimen, edge able specimen deflection, specimen alignment, type of

Table I. Comparison of ASTM C1499 and EN1288


Parameter ASTM C1499 (Advanced ceramics) EN 1288 (Glass)
h i h i
Ds2 Dl2 ð1lÞr22 r12 F
Strength rf ¼ 2ph
3F
2 ð 1  l Þ 2D 2 þ ð 1 þ l Þ ln Ds
Dl rT ¼ 3ð1þlÞ
2p ln r2
r1 þ ð1þlÞ2r 2 h2
3
calculation
Specimen type Circular or square plates
Specimen size Dimensions determined using equations to ensure Diameter or side length of 66 or 100 mm
response follows simple plate theory
Load/support 0.2–0.5 except for a material with E ≤ 100 GPa Fixed at 0.2 for both sizes
ring ratio and r ≥ 1 GPa then the ratio is 0.2
Plate overhang (D  Ds)/h ≥ 2, h = plate thickness Fixed
Plate edges Provides some edge preparation guidelines + schematic No details provided
of valid and invalid fracture patterns
F, fracture load; h, plate thickness; , Poisson’s ratio; Ds, diameter of the support ring; Dl, diameter of the load ring; D, diameter of the circular
specimen or the diameter of a circle that expresses the characteristic size of a square plate; r2, radius of support ring; r1, radius of load ring; r3,
radius of the circular specimen or the radius of a circle that expresses the characteristic size of a square plate.
www.ceramics.org/IJAGS Equibiaxial Flexure Strength of Glass 3

load, and support ring material, the use of thin layers tribution and can result in substantially different
of material between the specimen and ring contact strength values compared to tests performed with sig-
areas to minimize stress concentrations and friction, as nificantly less overhang. The amount of overhang is
well as a number of other related issues. Although there specified in EN 1288 since the test plate and support
are many similarities between the two standards, there ring diameters are fixed. In ASTM C1499, an over-
are a number of items which are addressed differently. hang of (D  Ds)/h = 2, where h is the plate thick-
These similarities and differences are summarized below ness, is considered sufficient but it is noted that if the
and in Table I, and they may have an impact on deter- plates have poor edge finishes or are cut from larger
mining the strength of float glass. plates, a value of 12 may be required. It also states
that a larger value may be necessary when testing opti-
cal materials.
Specimen Size
ASTM C1499 does not specify the size of the test
Plate Edges
specimen, instead the standard provides equations that
can be used to calculate the plate dimensions, based on One of the advantages of the equibiaxial flexure
specimens thickness, to ensure that the plate response test is that the plate edges are not exposed to the maxi-
to the applied load is described by simple plate theory. mum stress and therefore, the likelihood of fracture ini-
EN 1288, on the other hand, specifies a plate diameter tiating at the edge is reduced. The stresses at the edge
or side length of 66 or 100 mm depending on the ring are circumferential, hoop stresses that are only a small
ratio employed. fraction of the stresses in the center of the plate, but if
flaws are sufficiently large on the edge, they can initiate
fracture. ASTM C1499 devotes a section to edge prepa-
Load and Support Ring Sizes (i.e., load-to-support
ration for advanced ceramics and provides a schematic
ring ratio)
of valid and invalid (edge-initiated fracture) fracture
ASTM C1499 does not provide specifications but patterns. EN 1288 Part 110 does not discuss edge issues
an equation to determine the support ring diameter other than to state in section 6.1.1 “The effect of the
based on the test specimen size. It also states that the specimen edge condition is for the most part sup-
ratio of the load-to-support ring diameter must be pressed.”
between 0.2 and 0.5, but in the case of a material with A review of the literature located several studies
an elastic modulus <100 GPa and fracture strength that determined the strength of float glass using ASTM
above 1000 MPa, the ratio should be 0.2. EN 1288 C149912–15 even though this standard was specifically
specifies a load/support ring ratio of 0.2 with load and designed for structural ceramics and possibly optical
support ring diameters respectively of 18 and 90 mm ceramics which are typically tougher than glass. In all
for the larger test specimen and 12 and 60 mm for the of these instances, the load-to-support ring ratios were
smaller test specimen. 0.4–0.5. The only paper that referenced the German or
European standard was reference 16, but they did not
follow the specified plate size or ring ratio instead using
Plate Overhang
plates with 50 mm edge lengths and a ring ratio
The amount of material that extends beyond the of 0.5.
edge of the support ring is known to have an influ- The information that follows summarizes an
ence on the strength. Powers et al.11 showed that the effort to examine the influence of the glass test speci-
stress at the edge of a plate decreased with increasing men size and the load-to-support ring ratio on the
overhang and increasing support ring to plate thick- measured strength of float glass. Results will be dis-
ness ratio. As the amount of overhang is reduced, the cussed in relation to minimizing the percentage of
stress at the plate edges increases and, depending on invalid strength tests (edge-initiated fracture) and the
the quality of the edges, can lead to plates where frac- consistency of the average strength to identify an opti-
ture commences at the edge. The data from a plate mum combination of glass test specimens size and
that fractures in such a fashion is considered invalid. load-to-support ring ratio to determine the strength of
Alternatively, too much overhang alters the stress dis- float glass.
4 International Journal of Applied Glass Science—Swab, et al. 2014

Materials of each plate to prevent glass fragments from flying


around and to assist with fractography. The side with
Three commercial float glass materials were part of the low-tack adhesive layer was then deemed the com-
this study: a soda-lime-silicate (SLS) from Guardian pression surface. A compliant layer (standard manila file
(Guardian Industries, Auburn Hills, MI), a borosilicate folder) was placed between load ring and the adhesive
(BF) from Schott (BOROFLOAT 33â Schott North plastic to reduce point contact and frictional stresses.
America, Duryea, PA) and a low-iron soda-lime-silicate Plates were subjected to a stressing rate of 30–35 MPa/
from PPG (Starphireâ PPG Industries, Inc, Pittsburgh, s (crosshead displacement rate of 7.6 mm/min for the
PA). The nominal thickness of both soda-lime glasses 0.5 ring ratio and 9.6 mm/min for the 0.2 ratio) until
was 5.6 mm while the BoroFloat had a thickness around fracture occurred. After fracture, the fracture initiation
6.6 mm. Plates 100-mm, 125-mm or 150-mm square location was identified, and each datum classified as a
were obtained from the heart of the respective sheet of valid—fracture initiated inside or near the load ring
glass by scoring and bending (Swift Glass Co., Elmira, area, or as invalid—fracture initiated outside the sup-
NY). All scoring was done on the air side of the glass port ring diameter typically at the edge of the plate. A
sheet, and all specimen edges were then swiped by hand Poisson’s ratio of 0.2 was used for all three glasses to
using a 100-grit SiC grinding belt to reduce the edge calculate the strength according to the equation in
sharpness for increased safety during handling. No lubri- ASTM C1499 (Table II).
cant was used during the cutting and swiping operation.
Each piece of glass was then individually wrapped in
Results
craft paper to minimize any damage during shipping.
The surface condition of the glass was considered
The strength testing data, as well as the number of
“as-received,” and no coatings or special treatments were
valid test results, are summarized in Table II for all
performed on the glass plates prior to testing. No
three glasses examined. All strength data shown is only
attempt was made to reduce the number of processing
for test plates that fractured in a valid fashion, and all
or handling flaws present on the glass surface.
data was generated by multiple investigators over sev-
eral years. Even with multiple investigators and slight
variations in test temperature and humidity, all data
Experimental Procedure
converged within an individual data set.
Ring-on-ring equibiaxial flexure testing was per-
formed on a Zwick/RoellTM Z030 (Zwick USA, Kennesaw, Some general observations
GA) test frame following the general guidelines of
(1). BF glass is weaker than the SLS or low-iron SLS,
ASTM C1499.8 Two steel load/support ring ratios
while the SLS and low-iron SLS have similar
were used: A 0.5 ratio which consisted of a 42.5-mm
strength values,
diameter load ring, and a 85-mm diameter support
(2). A significant amount of scatter exists in the
ring, and a 0.2 ratio with a 18-mm diameter load ring,
strength values highlighted by standard deviations
and a 90-mm diameter support ring. Prior to flexure
that were, in many cases, 20–40% of the average
strength testing the air and tin sides of each glass plate
strength. There is no trend relating these high
were identified using a tin detector (Model TS2300;
standard deviations to glass type, air or tin side,
EDTM, Inc, Toledo, OH). Strength data sets were
plate size or ring ratio,
generated for the air and tin side of each of the three
(3). Strength determined with the 0.2 ring ratio was
glasses using both ring ratios. In most cases, a mini-
always higher than the value obtained with the
mum of 25 plates were tested for each data set. Plate
0.5 ratio for all conditions,
thickness was determined by measuring the thickness
(4). The number of valid strength tests tended to
near the four corners of each plate with a micrometer,
increase with increasing plate size and/or a
and the average plate thickness was calculated for use
decrease in the ring ratio. Examples of valid and
in the subsequent strength calculation. A thin
invalid fracture patterns in these glasses are shown
(0.05 mm) low-tack adhesive plastic, nominally the
in Fig. 1.
same size as the glass plate, was then affixed to one side
www.ceramics.org/IJAGS Equibiaxial Flexure Strength of Glass 5

Table II. Summary of Equibiaxial Flexure Strength Data


SLS Low-Iron SLS BF

0.5 0.2 0.5 0.2 0.5 0.2


100 mm 9 100 mm plates
TIN
r (MPa) 106.2 153.2 132.7 160.1 93.6 133.8
StD 14.3 29.5 24.0 56.3 19.1 19.5
Valid 30 30 27 25 6 14
Edge-invalid 10 0 11 5 24 14
T (°C) — 23 23 25 23 26
%H — 50 42–52 43 26 43
AIR
r (MPa) 114.3 180.8 — 174.6 — 177.6
StD 44.7 21.6 — 45.5 — 19.1
Valid 4 15 0 10 0 8
Edge-invalid 26 10 15 20 15 20
T (°C) — 23 23 24 23 23–24
%H — 36 47 32 22 26–28
125 mm 9 125 mm plates
TIN
r (MPa) 154.4 185.9 111.8 145.8 113.2 151.6
StD 30.6 51.2 28.7 35.4 18.3 32.7
Valid 24 25 30 30 29 27
Edge-invalid 1 0 0 0 1 3
T (°C) 23 23 23 23 23 22–23
%H 45 46 47 32 45–52 55–31
AIR
r (MPa) 193.8 267.0 171.3 295.7 141.9 189.6
StD 23.0 57.9 31.2 44.1 29.0 46.4
Valid 15 20 18 20 23 24
Edge-invalid 10 5 12 9 7 5
T (°C) 23 23 23 23 23 22–23
%H 49 41 47 27 50–42 38–47
150 mm 9 150 mm plates
TIN
r (MPa) 150.1 199.4 142.7 147.3 124.6 166.7
StD 53.3 34.1 30.6 44.6 38.1 41.4
Valid 18 19 30 30 30 28
Edge-invalid 0 0 0 0 0 2
T (°C) 22 21 23 23 23–25 23
%H 41 62 21–27 31 48–28 23–29
AIR
r (MPa) 203.4 250.5 249.1 319.4 164.0 237.5
StD 42.2 42.9 58.4 56.0 36.5 58.2
Valid 17 20 28 21 29 30
Edge-invalid 1 0 1 8 2 0
T (°C) 22 21 23 23 23–25 23
%H 46 51 21 22 23–54 34–26
StD is the standard deviation associated with each strength value.
6 International Journal of Applied Glass Science—Swab, et al. 2014

Fig. 1. Examples of valid (left) and invalid (right) fracture patterns observed. Arrow in the invalid image on the right shows that frac-
ture initiated at the edge of the plate.

Initial testing conducted on the air and tin sides of strength was especially noted in the SLS where irrespec-
100-mm plates of the SLS glass with a 0.5 ring ratio tive of which side was tested or which ring ratio was
showed virtually no strength difference between these used, the strength data generated using the 125-mm
sides which was unexpected as previous studies have plates was essentially the same as the data from the
shown that the air side of float glass is consistently 150-mm plates. The data from the 100-mm plates was
stronger.14–18 None of the 100-mm low-iron SLS or consistently lower. This is also the case for the air side
BF plates, with the air side in tension, yielded valid test of the low-iron SLS tested with the 0.2 ring ratio. A
data, so no comparison to the tin side strength could definitive reason for this difference is unknown. How-
be made for these two glasses using this plate size/ring ever, the SLS 100-mm plates tested using the 0.5 ring
ratio combination. When the ring ratio was 0.2, valid ratio were purchased at a different time that than all
strength data were collected for the air side of the SLS the rest of the SLS glass plates that were tested. It is
and BF glasses and the air/tin side strength difference possible that there are subtle compositional differences
became apparent in BF glass, but this difference was between these different glass batches or that the scoring
not as apparent in the low-iron SLS due the extremely and handling was done differently leading to larger
high standard deviations associated with the average flaws on these initial plates. Even though the average
strength. In general, decreasing the ring ratio to 0.2 for strength was reproducible with the larger plates, the
the 100-mm plates improved the percentage of valid standard deviations remained quite high for all condi-
tests for both the air and tin sides, but the percentage tions making a statistical comparison of the strength of
of valid test data on the air side was at best slightly bet- the different sides difficult.
ter than 50%. Strength data collected using the 0.2 The percentage of valid strength data for the tin
ring ratio is expected to be higher than the strength side increased dramatically for all three glasses when
collected with the 0.5 ratio because the effective area the plate size was increased to plates with 125 or
with the 0.2 ratio is smaller and with that comes a 150 mm edge lengths. The percentage of valid fractures
reduced likelihood of largest flaw on the surface on the tin side jumped to at least 90% when the 125-
exposed to the maximum stress. While the strength is a mm square plates were tested and to 100%, for all
function of the effective area (ring ratio), strength val- 150 mm plates, except the BF with the 0.2 ring ratio
ues from different ring ratios can be compared using but even then, 93% of the plates exhibited a valid frac-
the Weibull effective area equations in Reference 9. ture pattern. This improvement in the percentage of
Increasing the plate edge length to 125 or valid tests occurred for both ring ratios with both plate
150 mm had a positive effect on the reproducibility of sizes. The percentage of plates that exhibited a valid
the average strength and the percentage of plates that fracture pattern also increased with increasing plate size
exhibited a valid fracture pattern. The effect on for the air side of all glasses; however, the increase was
www.ceramics.org/IJAGS Equibiaxial Flexure Strength of Glass 7

not as substantial as what was observed on the tin side, using the 125- and 150-mm plates and the 0.5 ring
but it was more substantial for the 150-mm square ratio, see Table III, but this value is not statistically dif-
plates than the 125-mm square plates. ferent than the average obtained with the 100-grit SiC.
Placing a c-bevel on the edges appears to lead to a
higher strength compared to the value obtained with
Discussion the 400-grit SiC belt, but again these values are not sta-
tistically different. However, the strength for the plates
It is clear from the strength data that the plate with c-bevels is statistically different than the strength
edge length coupled with the size of the equibiaxial obtain when the edges were treated with the 100-grit
ring configuration has an effect on the measured equi- SiC belt. The reasons for the difference are unknown,
biaxial flexure strength of glass and the occurrence of but a plausible explanation could be how the plates
edge-initiated fractures in the test plate. As the stress at were held during the edge treatment process. Plates
the edge is much lower than the maximum stress at the with the c-bevel were held by a vacuum chuck while
surface, a plate that has an edge-initiated fracture is an the edge was prepared; however, the 100-grit edge
indication that the edge damage is larger and more det- treatment was carried out by hand. It is possible, that
rimental that any of the inherent damage on the glass in the latter case, larger strength-limiting flaws were
surface. A recent publication15 showed a similar low introduced on the 100-mm square surfaces. Irrespective
percentage of valid test data for the air side of large of how the edge was treated there was essentially no
soda-lime-silicate glass when a 0.5 ring ratio was used. reduction in the percentage of invalid edge fractures.
None of the other publications containing equibiaxial In a majority of cases (8 of 12 combinations),
flexure strength data report the number of specimens increasing the edge length and keeping the ring ratios
that had edge-initiated fracture. Reducing the ring ratio fixed yielded strength data that were essentially the
to 0.2 improved the percentage of valid test data, but same for the 125- and 150-mm plates. It also increased
this improvement was not significant for the 125-mm the likelihood that the plate would fracture in a valid
and 150-mm plates. The data from the 100-mm plates manner. These plates are much larger than the support
raises concerns about the appropriateness of the plates ring diameter, so there is a fair amount of material over-
and ring sizes specified in EN 1288 for determining hang. ASTM C1499 recommends an overhang of <12 to
flexure strength of glass. (It should be noted that the avoid altering the stress distribution in the plate during
thickness of the plate can also influence the test. These testing, but it also states that an overhang up to 24 is
float glasses had a nominal thickness of 6 mm. This is allowed. The overhang of the 150-mm plates with the
in the middle of the thickness range recommended by 0.2 ring ratio is exactly (D  Ds)/h = 12, and it is
C1499, but it is about twice as thick as the minimum almost 13 when the 0.5 ring ratio is used. In fact, these
thickness recommended by EN1288). results validate the note in ASTM C1499 that an over-
Modifying the cutting and swiping procedure to hang in excess of 12 may be needed for optical materials.
minimize edge damage may mitigate some of these The high standard deviations associated with all of
issues for glass strength testing, but it might not be suf- the valid strength data collected in this effort was a
ficient for transparent armor applications. To examine
the influence of the edge damage, three additional sets Table III. Strength Data for 100-mm Square Plates
of strength tests were conducted with twenty-five 100- with Tin Side in Tension Using the 0.5 Ring Ration
mm square plates of SLS glass in each set. The differ- after Various Edge Treatments
ence between each set was the edge treatment. The first
two sets were swiped by hand with a 100-grit or 400- 100-SiC 400-SiC 400-grit
grit SiC belt sander running parallel to the plate edges grit w/H2O grit w/H2O c-bevel
with water as a lubricant. The third set had a c-bevel
r (MPa) 105.5 141.2 178.9
(rounded edge) placed on all four sides using a 400-grit
StD 15.4 31.5 28.5
tool bit. All specimens were tested with the tin side in
Valid 10 13 7
tension using the 0.5 ring ratio configuration. Reducing
Edge 15 11 18
the SiC grit size of the belt yields an average strength
value very similar to the average strength obtained StD is the standard deviation associated with each strength value.
8 International Journal of Applied Glass Science—Swab, et al. 2014

surprise. As flaw size and distribution control strength, biaxial flexure standards that exist, ASTM C1499 seems
the high scatter indicates a wide distribution of flaw to be the more appropriate than EN 1288 as the proce-
sizes on the plate surface. Weibull statistics were not dures and requirements in the former are flexible to
applied to these data sets as the high standard devia- accommodate issues related to glass, specifically plate
tions would result in extremely low Weibull values ren- size, load/support ring ratio, and the amount of mate-
dering the analysis meaningless. Recent work by rial overhang. It was determined that an increased plate
Wereszczak et al.19 used a nondestructive technique to size improved the number of valid fractures but did
map and indentify the flaw sizes and distributions on not improve the standard deviation associated with each
the surface of the same low-iron SLS and BF glasses strength value. An investigation to improve the edges
examined in this effort. Their results showed a signifi- during plate fabrication, using finer grit abrasives, was
cant number of potential flaws on the glass surface. unsuccessful in reducing the frequency of invalid frac-
The flaw density (number of large flaws per square tures. However, edge preparation procedures for glass
centimeter) was significantly higher for the BF (>7) might need to be developed as edge quality has a sig-
compared to low-iron SLS (<3) and that the flaws on nificant influence on the validity of the resulting
the BF surfaces tended to be larger. This difference in strength data.
flaw size and density accounts for the lower strength of
the BF and the large amount of scatter in the strength
data. Acknowledgments
Finally, the air side consistently had a higher per-
centage of invalid edge fractures than the associated tin The authors would like to acknowledge George
side. This can be attributed to the fact that all plates Quinn from the National Institute of Standards and
were scored on the air side which introduced larger Technology for providing background information and
flaws on these edges. This coupled with the higher flex- valuable discussions on equibiaxial flexure testing. Jared
ure strength and hence higher stress at the plate edge, Wright from Army Research Laboratory is also acknowl-
increased the likelihood that edge fractures would edged for training the students on the equibiaxial flexure
occur. testing procedure and the load frame operation.
The results of this effort indicate that the proce-
dures and specifications outlined in ASTM C1499 are
more appropriate for determining the flexure strength
References
of float glass than those in EN 1288. Based on these
results, it is proposed that for float glass plates approxi- 1. H. P. Kirchner, Strengthening of Ceramics. Treatment, Tests and Design
mately 6 mm thick, the edge length should be no Applications, Marcel-Dekker, New York, NY, 1979.
<150 mm, the ring ratio shall be 0.5 or less, and the 2. M. E. Nordberg, E. L. Mochel, H. M. Garfinkel, and J. S. Olcott,
“Strengthening by Ion Exchange,” J. Am. Ceram. Soc., 47 [5] 215–219
amount of overhang, as calculated in C1499, shall (1964).
be at least 12. However, these requirements may not be 3. M. B. Abrams, D. J. Green, and S. J. Glass, “Fracture Behavior of Engi-
neered Stress Profile Soda Lime Silicate Glass,” J. Non. Cryst. Solids, 321
appropriate should there be a significant increase in 10–19 (2003).
plate thickness or glass strength due to a strengthening 4. ASTM C158-02, “Standard Test Methods for Strength of Glass by Flexure
(Determination of Modulus of Rupture),” ASTM Annual Book of Stan-
treatment. It is also recommended that edge prepara- dards, Vol. 15.01, 2012.
tion procedures be identified as well as a minimum 5. P. F. James, M. Chem, and F. R. Jones, “Strengthening of Soda Lime Sil-
plate size/ring ratio combinations to ensure that valid ica Glass by Sol Gel and Melt-Derived Coatings,” J. Non. Cryst. Solids,
155 99–109 (1993).
strength data are generated, and the number of invalid 6. B. D. Fabes, W. F. Doyle, B. J. J. Zelinski, L. A. Silverman, and D. R.
tests is minimized. Uhlmann, “Strengthening of Silica Glass by Gel-Derived Coatings,”
J. Non. Cryst. Solids, 82 349–355 (1985).
7. L. Y. L. Wu, G. H. Tan, M. Qian, and T. H. Li, “Formulation of Trans-
parent Hydrophobic Sol-Gel Hard Coatings,” SIMTech Tech. Rep., 6 [2]
Summary 1–4 (2005).
8. ASTM C1499-09, “Standard Test Method for Monotonic Equibiaxial
Flexural Strength of Advanced Ceramics at Ambient Temperature,” ASTM
This study indicates that the equibiaxial flexure test Annual Book of Standards, Vol. 15.01, 2009.
9. J. A. Salem and L. M. Powers, “Guidelines for the Testing of Plates,”
can be a valid means of determining the flexure Proceedings of the 27th International Cocoa Beach Conference on
strength of large plates of float glass. Of the two equi- Advanced Ceramics and Composites: B, Ceramic Engineering and Science
www.ceramics.org/IJAGS Equibiaxial Flexure Strength of Glass 9

Proceedings, Vol. 24, No. 4, eds., W. M. Kriven and H. T. Lin, 357–364, 14. A. A. Wereszczak, T. P. Kirkland, K. T. Strong Jr, and T. J. Holmquist,
2003. “ORNL Quasi-Static Mechanical Characterization and Analysis: FY09
10. EN 1288 “Glass in Building – Determination of the Bending Strength of Annual Report to TARDEC,” ORNL/TM-2009/234, 2009.
Glass, Part 1, Fundamentals of Testing Glass and Part 5, Coaxial Double 15. A. A. Wereszczak, T. P. Kirkland, M. E. Ragan, K. T. Strong Jr, H. T.
Ring Test on Flat Specimens with Small Test Surface Areas,” European Lin, and P. Patel, “Size Scaling of Tensile Failure Stress in a Float Soad-
Committee for Standardization, 1991. Lime-Silicate Glass,” Int. J. Appl. Glass Sci., 1 [2] 143–150 (2010).
11. L. M. Powers, J. A. Salem, and A. S. Weaver, “Stresses in Ceramic Plates 16. S. T. Gulati, R. Akcakaya, and J. R. Varner, “Fracture Behavior of Tin vs.
Subjected to Loading between Concentric Rings,” Fracture Resistance Air Side of Float Glass,” Fractography of Glasses and Ceramics IV, eds.,
Testing of Monolithic and Composite Brittle Materials, ASTM STP1409, J. R. Varner and G. D. Quinn. Ceramic Transactions Vol. 122, The
30–45, 2002. American Ceramic Society, Westerville, OH, 2001.
12. M. H. Krohn, J. R. Hellmann, D. L. Shellman, D. J. Green, G. E. 17. R. R. Tummala and B. J. Foster, “Strength and Dynamic Fatigue of Float
Sakoske, and J. A. Salem, “Test Methodology for Strength Testing of Glass Surfaces,” J. Am. Ceram. Soc., 58 [3–4] 156–157 (1975).
Soda-Lime-Silica Float Glass Before and After Enameling,” J. Test. Eval., 18. K. M. Entwistle, “The Fracture Stress of Float Glass,” J. Mater. Sci., 28
30 [6] 1–8 (2002). 2007–2012 (1993).
13. M. H. Krohn, J. R. Hellmann, D. L. Shellman, C. G. Pantano, and G. E. 19. A. A. Wereszczak, M. K. Ferber, and W. Musselwhite, “Method for Identi-
Sakoske, “Biaxial Flexure Strength and Dynamic Fatigue of Soda-Lime-Sil- fying and Mapping Flaw Size Distribution on Glass Surface for Predicting
ica Float Glass,” J. Am. Ceram. Soc., 85 [7] 1777–1782 (2002). Mechanical Response,” Int. J. Appl. Glass Sci., 5 [1] 16–21 (2014).

You might also like